Scholarly article on topic 'Comparison of commonly used antimicrobial susceptibility testing methods for evaluating susceptibilities of clinical isolates of Enterobacteriaceae and nonfermentative Gram-negative bacilli to cefoperazone–sulbactam'

Comparison of commonly used antimicrobial susceptibility testing methods for evaluating susceptibilities of clinical isolates of Enterobacteriaceae and nonfermentative Gram-negative bacilli to cefoperazone–sulbactam Academic research paper on "Biological sciences"

CC BY-NC-ND
0
0
Share paper
OECD Field of science
Keywords
{cefoperazone–sulbactam / "disk diffusion" / "phoenix system" / "susceptibility test" / "Vitek 2 system"}

Abstract of research paper on Biological sciences, author of scientific article — Shio-Shin Jean, Chun-Hsing Liao, Wang-Huei Sheng, Wen-Sen Lee, Po-Ren Hsueh

Abstract Background/Purpose The aim of this study was to investigate the cefoperazone–sulbactam (CFP–SUL) susceptibilities of important Gram-negative bacteria (GNB) by agar dilution (reference method), disk diffusion, and two automated methods. Methods A total of 799 GNB isolates, including Enterobacteriaceae (n = 500) and nonfermentative GNB (NFGNB, n = 299), were recovered from various clinical specimens collected at National Taiwan University Hospital, Taipei, Taiwan from November 2013 to December 2014. The agar dilution method, disk diffusion method, and two automated susceptibility systems (Phoenix and Vitek 2) were used for testing susceptibility of the isolates to CFP–SUL. Categories of susceptibility (susceptible, intermediate, or resistant) to CFP–SUL yielded from each method were interpreted according to CFP–SUL interpretive breakpoints proposed previously. The results of categorical agreement and errors obtained between the agar dilution method and the other three methods were analyzed. Results The Vitek 2 system had the highest error rates against Escherichia coli (n = 150) and Enterobacter cloacae (n = 77) isolates, i.e., 6.7% and 11.7% minor errors, 8.5% and 1.7% major errors, and 40% and 20% very major errors, respectively. Additionally, the Vitek 2 system was also found to have a significantly lower sensitivity (44.4%) and lower positive predictive value (18.2%) for detecting CFP–SUL nonsusceptible E. coli isolates than other methods. For carbapenem-nonsusceptible Enterobacteriaceae isolates, the Vitek 2 system failed to detect correct susceptibility to CFP–SUL. The three methods failed to correctly detect CFP–SUL susceptibility categories against all NFGNB isolates except Pseudomonas aeruginosa. Conclusion The Vitek 2 system is a suboptimal method in correctly detecting CFP–SUL susceptibility categories for E. coli, E. cloacae, and carbapenem-nonsusceptible Enterobacteriaceae isolates.

Academic research paper on topic "Comparison of commonly used antimicrobial susceptibility testing methods for evaluating susceptibilities of clinical isolates of Enterobacteriaceae and nonfermentative Gram-negative bacilli to cefoperazone–sulbactam"

ARTICLE IN PRESS

^ + MODEL ^^^B

Journal of Microbiology, Immunology and Infection (2015) xx, 1-10

ORIGINAL ARTICLE

Comparison of commonly used antimicrobial susceptibility testing methods for evaluating susceptibilities of clinical isolates of Enterobacteriaceae and nonfermentative Gram-negative bacilli to cefoperazone—sulbactam

Shio-Shin Jean a, Chun-Hsing Liao b, Wang-Huei Sheng c, Wen-Sen Lee d, Po-Ren Hsueh c e *

a Emergency Department, Department of Emergency and Critical Care Medicine, Wan Fang Hospital, Taipei Medical University, Taipei, Taiwan

b Department of Internal Medicine, Far Eastern Memorial Hospital, New Taipei City, Taiwan c Department of Internal Medicine, National Taiwan University Hospital, National Taiwan University College of Medicine, Taipei, Taiwan

d Division of Infectious Diseases, Department of Internal Medicine, Wan Fang Hospital, Taipei Medical University, Taipei, Taiwan

e Department of Laboratory Medicine, National Taiwan University Hospital, National Taiwan University College of Medicine, Taipei, Taiwan

Received 21 July 2015; received in revised form 25 August 2015; accepted 29 August 2015 Available online ■ ■ ■

Abstract Background/Purpose: The aim of this study was to investigate the cefoperazone —sulbactam (CFP— SUL) susceptibilities of important Gram-negative bacteria (GNB) by agar dilution (reference method), disk diffusion, and two automated methods. Methods: A total of 799 GNB isolates, including Enterobacteriaceae (n = 500) and nonfermentative GNB (NFGNB, n = 299), were recovered from various clinical specimens collected at National Taiwan University Hospital, Taipei, Taiwan from November 2013 to December 2014. The agar dilution method, disk diffusion method, and two automated susceptibility systems (Phoenix and Vitek 2) were used for testing susceptibility of the isolates to CFP—SUL.

* Corresponding author. Department of Laboratory Medicine, National Taiwan University Hospital, Number 7, Chung-Shan South Road, Taipei 100, Taiwan.

E-mail address: hsporen@ntu.edu.tw (P.-R. Hsueh).

http://dx.doi.org/10.1016/j.jmii.2015.08.024

1684-1182/Copyright © 2015, Taiwan Society of Microbiology. Published by Elsevier Taiwan LLC. All rights reserved.

KEYWORDS

cefoperazone —sulbactam; disk diffusion; phoenix system; susceptibility test; Vitek 2 system

+ MODEL

2 S.-S. Jean et al.

Categories of susceptibility (susceptible, intermediate, or resistant) to CFP—SUL yielded from each method were interpreted according to CFP—SUL interpretive breakpoints proposed previously. The results of categorical agreement and errors obtained between the agar dilution method and the other three methods were analyzed.

Results: The Vitek 2 system had the highest error rates against Escherichia coli (n = 150) and Enterobacter cloacae (n = 77) isolates, i.e., 6.7% and 11.7% minor errors, 8.5% and 1.7% major errors, and 40% and 20% very major errors, respectively. Additionally, the Vitek 2 system was also found to have a significantly lower sensitivity (44.4%) and lower positive predictive value (18.2%) for detecting CFP—SUL nonsusceptible E. coli isolates than other methods. For carbapenem-nonsusceptible Enterobacteriaceae isolates, the Vitek 2 system failed to detect correct susceptibility to CFP—SUL. The three methods failed to correctly detect CFP—SUL susceptibility categories against all NFGNB isolates except Pseudomonas aeruginosa. Conclusion: The Vitek 2 system is a suboptimal method in correctly detecting CFP—SUL susceptibility categories for E. coli, E. cloacae, and carbapenem-nonsusceptible Enterobacteriaceae isolates.

Copyright © 2015, Taiwan Society of Microbiology. Published by Elsevier Taiwan LLC. All rights reserved.

Introduction

Antimicrobial drug resistance is a serious health care problem worldwide, and nosocomial infections due to multidrug-resistant (MDR) Gram-negative bacilli (GNB) are of particular concern.1,2 Initiation of appropriate antibiotics for treatment of severe infections, however, depends on the rapid identification of the pathogen and an understanding of the susceptibility profiles of pathogens to various antimicrobials.

Resistance to major antibiotics is often conferred by the expression of b-lacatmases.2,3 Sulbactam (SUL), an important b-lactamase inhibitor, was noted to possess good activity against some extended-spectrum b-lactamase (ESBL)-producing pathogens.4 Furthermore, this agent has been shown to augment the activity of cefoperazone (CFP) against a variety of Enterobacteriaceae species (especially Escherichia coli, Morganella morganii, and Klebsiella pneumoniae) which include some AmpC b-lactamase producers as well (Enterobacter cloacae, Enterobacter aero-genes, and Citrobacter freundii),5—7 and some important nonfermentative GNB species that are originally resistant to CFP.7—11

There are no recently validated minimum inhibitory concentration (MIC) interpretive breakpoints for CFP (with or without SUL) against GNB isolates by the Clinical and Laboratory Standards Institute (CLSI) and the European Committee on Antimicrobial Susceptibility Testing.12—14 CFP itself has fallen into disuse in the United States (US) and European Union and CFP—SUL had never been licensed in the US or anywhere in Western Europe. However, this combination agent is still extensively used in many hospitals of the Asian countries and has been reported to successfully treat several nosocomial infections, including mild-to-moderate nosocomial pneumonia, intra-abdominal infections, and sepsis in patients with febrile neutropenia.11,15,16 Nevertheless, there were only a few studies comparing the accuracy of various susceptibility testing methods for detection of

in vitro susceptibility of important GNB species to CFP or CFP—SUL.5,6,17,18

In this study, we conducted an in vitro study to investigate the performance of the agar dilution method, disk diffusion method, automated BD Phoenix (Becton Dickinson, Sparks, MD, USA), and Vitek 2 (bioMerieux, Marcy l'Etoile, France) systems for evaluating susceptibility of clinically important GNB isolates to CFP—SUL. Correlations between the susceptibility profiles obtained by the agar dilution method and the other three methods were also explored.

Methods Bacterial isolates

From November 2013 to December 2014, a total of 799 non-duplicate GNB isolates (1 isolate per patient) were randomly collected from various clinical specimens of patients who were hospitalized at National Taiwan University Hospital, Taipei, Taiwan, a 2500-bed university-affiliated tertiary care hospital. This study was approved by the Institutional Review Boards of the National Taiwan University Hospital (NTUH 201307067RINA). The GNB isolates comprised Enterobacteriaceae species, namely E. coli (n = 150), K. pneumoniae (n = 150), E. cloacae (n = 77), C. freundii (n = 13), Serratia marcescens (n = 30), Salmonella species (n = 50), Proteus mirabilis (n = 30), and nonfermentative GNB (NFGNB) species, namely Pseudomonas aeruginosa (n = 110), Acineto-bacter calcoaceticus—Acinetobacter baumannii (Acb) complex (n = 120), Stenotrophomonas maltophilia (n = 39), Chyrseobacterium indologenes (n = 10), Eliz-abethkingia meningoseptica (n = 12), and Burkholderia cepacia complex (n = 8). The isolates were identified using conventional identification methods as well as the Phoenix PMIC/ID-30 identification system (Becton Dickinson).

+ MODEL

Susceptibility of Enterobacteriaceae to CFP—SUL 3

Antimicrobial susceptibility testing

Antimicrobial susceptibilities of all isolates were determined concomitantly by the agar dilution, disk diffusion method, and two commercial automated susceptibility systems, namely the Phoenix system (Becton Dickinson) and the Vitek 2 system (bioMerieux). The disks (75 mg CFP and 30 mg SUL) used in the disk diffusion method were obtained from Becton Dickinson. For susceptibility testing by the agar dilution method, CFP—SUL at a fixed 2:1 ratio of CFP:SUL and serial two-fold CFP concentrations ranging from 0.015 mg/mL to 128 mg/mL in combination with SUL were used together with cation-adjusted Mueller-Hinton agar (Becton Dickinson). The inoculated agar plates were incubated in ambient air at 35°C for 16—20 hours. The MIC of each antimicrobial agent was defined as the lowest concentration that inhibited visible growth of the organism.

For evaluation of the two automated susceptibility systems, a Vitek AST-N310 card (bioMerieux) and a BD Phoenix Gram Negative Combo Panel (NMIC/ID-99; Becton Dickinson Diagnostics) were used. For the Vitek AST-N310 card, the ratio of concentration of CFP to SUL was 2:1 and the concentrations of CFP tested were < 8 mg/mL, 16 mg/mL, 32 mg/mL, and > 64 mg/mL. By contrast, with regard to the Phoenix NMIC/ID99 panel, as Barry and Jones5 demonstrated that the 8 mg/mL SUL concentration addition would convert more CFP-nonsusceptible (MIC > 16 mg/mL) isolates of Enterobacteriaceae and Pseudomonas spp. into CFP-susceptible ones than the 4 mg/mL SUL concentration use, the CFP concentrations were also tested in serial two-fold dilutions ranging from 0.5 mg/mL to 32 mg/mL in combination with a fixed SUL concentration of 8 mg/mL. Control strains of E. coli ATCC 25922, E. coli ATCC 35218, and P. aeruginosa ATCC 27853, were included in each set of tests.

Tests were repeated for isolates showing discrepant results between the agar dilution and the other three evaluated methods. If the results yielded by the initial and repeated tests were not identical, all of the susceptibility methods were repeated. The susceptibility results which were obtained two out of three times were recorded as the final results.

In this study we also investigated the susceptibilities of all isolates to imipenem and meropenem by the agar dilution method to delineate carbapenem susceptibility.13

Interpretation of susceptibility results

There are no CLSI-recommended MIC or disk diffusion interpretive criteria for susceptibility of Enterobacteriaceae spp., P. aeruginosa, species of Acb complex, B. cepacia complex, S. maltophilia and other non-Enter-obacteriaceae isolates to CFP—SUL.13 Furthermore, these MIC judgment criteria were also not documented in the expert rules of the Vitek AST-N310 card or Phoenix NMIC/ ID99 panel. Nevertheless, in this study, isolates were considered susceptible, intermediate, or resistant to CFP—SUL in accordance with the previous MIC breakpoints of CFP—SUL recommended by Jones et al6 and Barry and Jones,5 which are just consistent with the CLSI 2015 guidelines for Enterobacteriaceae spp.,13 and the CLSI 2010 guidelines for P. aeruginosa for CFP alone.12 For the disk

(CFP/SUL, 75/30 mg) diffusion method, the isolates with zone diameters > 21 mm were defined as susceptible; those with diameters ranging from 16 mm to 20 mm were defined as intermediate; and isolates with zone diameters < 15 mm were defined as resistant, as stated elsewhere.5'19 For the three dilution methods, after SUL was added, isolates with CFP MIC values < 16 mg/mL were defined as susceptible; those with a MIC value of 32 mg/mL were defined as intermediate; and isolates with CFP MIC values > 64 mg/mL were defined as resistant. Isolates showing no susceptibility to either imipenem or meropenem, as defined by the CLSI, were considered nonsusceptible to carbapenems.13

Data analysis

The rates of accuracy (categorical agreement) of the three evaluation modalities (disk diffusion, BD Phoenix system, and Vitek 2 system) were examined using GNB isolates. Using the results of the agar dilution method as the reference, we analyzed categorical agreement (CA) of the susceptibility (i.e., susceptible, intermediate, and resistant) categories between the evaluation system and agar dilution method. Errors were defined as follows: a minor error (mE) indicated that the result was intermediate in one system and susceptible or resistant in the other; a major error (ME) indicated a false-resistant result; and a very major error (VME) indicated a false-susceptible result. When calculating the rates of error, we applied the following denominators in respective species for estimation: the numberof reference resistant isolates for the VME rate, the number of reference susceptible isolates for the ME rate, and the number of all reference isolates for the mE rate, respectively.17 The acceptable intermethod error rates of VME, ME, and mE is < 1.5%, < 3%, and < 10%, respectively.20

Statistical analysis

Categorical variables are expressed as percentages of total isolates required for specific purposes in respective species, and were compared between different systems by the Chi-square test with Yate's correction or Fisher's exact test, as appropriate. Continuous variables were compared using the Student's t test or Mann—Whitney U test, depending on the validity of the normality assumption. In addition, sensitivity, specificity, positive predictive values (PPV), and negative predictive values (NPV) were calculated to compare the accuracy (i.e., degree of categorical agreement) of the disk diffusion, Vitek 2, and Phoenix systems with that of the reference method (agar dilution) in detecting CFP—SUL nonsusceptibilities. Finally, because only a few nonfermentative GNB isolates (C. indologenes, E. meningoseptica) have CLSI 2015-defined CFP MIC breakpoints,13 the discriminatory powers of the methods under evaluation were surveyed only for Enter-obacteriaceae species by estimating the area under the receiver-operating characteristic (ROC) curve. In this way, we were able to explore the differences in accuracy between the three comparator methods and the agar dilution method for correctly classifying the tested strains of enterobacterial GNB species with low CA rates (< 90%) into

S.-S. Jean et al.

Table 1 The range of minimum inhibitory concentrations (MICs) and disk diffusion diameter (in millimeters), MIC50, MIC90 levels, and susceptibility profiles (percentages of susceptible, intermediate, and resistant categories) versus cefoper-azone—sulbactam for Gram-negative bacteria, and their concordance and discordance between the susceptibility results evaluated by the agar dilution method (reference data) and the disk diffusion method, the BD Phoenix system, as well as the Vitek 2 system, respectively.

Species, results

Range MIC50 MIC90

(mg/mL) (mg/mL) (mg/mL)

Susceptibilities % of susceptibility results with:

S (%) I (%) R (%) CA

Escherichia coli (n = 150)

Agar dilution 0.03—64 1 16 94 2.7 3.3

Disk diffusion (diameter, mm) 13—35 92.7 5.3 2 97.3* 2.7

BD Phoenix <0.5/8—>32/8 <0.5/8 4/8 94 1.3 4.7 97.3* 2.7

Vitek 2 8— 64 8 8 85.3 5.3 9.3 84*-a<1) 6.7

0* 0* 8.5*

■b<1) 40

Klebsiella pneumoniae (n = 150)

Agar dilution 0.12—>128 0.25 16 90.4 1.9 7.7

Disk diffusion (diameter, mm) 13—28 86.5 5.8 7.7 96.7 3.3 0 0

BD Phoenix <0.5/8—>32/8 <0.5/8 32/8 88.5 1.9 9.6 96 3.3 0.7 0

Vitek 2 8— 64 8 8 94.2 0 5.8 95.3 4 0 16.7

Enterobacter cloacae (n = 77)

Agar dilution 0.03—128 0.5 64 76.6 10.4 13

Disk diffusion (diameter, mm) 11 — 40 76.6 20.8 2.6 84.4b(2) 15.6b(2) 0 0

BD Phoenix <0.5/8—>32/8 <0.5/8 >32/8 76.6 3.9 19.5 87b(3) 11.7b(3) 0 10

Vitek 2 8— 64 8 64 77.9 6.5 15.6 84.4 11.7b(4) 1.7 20

Citrobacter freundii (n = 13) Agar dilution

0.25-32

92.3 7.7 0

Disk diffusion (diameter, mm) 16—28 92.3 7.7 0 100 0 0 0

BD Phoenix <0.5/8—>32/8 <0.5/8 <0.5/8 92.3 0 7.7 92.3c(3) 7.7 0 0

Vitek 2 <8—>64 <8 <8 92.3 0 7.7 92.3 7.7 0 0

Serratia marcescens (n = 30)

Agar dilution 0.5—32 2 16 96.7 3.3 0

Disk diffusion (diameter, mm) 18—26 86.7 13.3 0 90 10 0 0

BD Phoenix <0.5/8—>32/8 2/8 16/8 96.7 0 3.3 93.3 3.3 3.4 0

Vitek 2 <8—32 <8 <8 96.7 3.3 0 100 0 0 0c(4)

Salmonella spp. (n = 50)

Agar dilution 0.25—32 1 8 98 2 0

Disk diffusion (diameter, mm) 18—28 98 2 0 100 0 0 0

BD Phoenix <0.5/8—>32/8 <0.5/8 1/8 98 0 2 98 2 0 0

Vitek 2 <8—32 <8 <8 98 2 0 100b(5) 0 0 0c(4)

Proteus mirabilis (n = 30)

Agar dilution 0.5—8 1 2 100 0 0

Disk diffusion (diameter, mm) 25—34 100 0 0 100* 0 0 0

BD Phoenix <0.5/8—>32/8 <0.5/8 4/8 93.3 3.3 3.3 93.3* 3.3 3.3 0

Vitek 2 <8 <8 <8 100 0 0 100* 0 0 0b(4)

Pseudomonas aeruginosa (n = 110)

Agar dilution 2—128 8 64 80 13.6 6.4

Disk diffusion (diameter, mm) 12—29 82.7 6.4 10.9 92.7 7.3 0 0

BD Phoenix 4/8—>32/8 8/8 >32/8 80 7.3 12.7 94.5b(6) 5.5c(5) 0c(7) 0

Vitek 2 <8—>64 <8 >64 81.8 7.3 10.9 93.6b<7) 6.4c(6) 0 0c(8)

Species of Acinetobacter calcoaceticus—Acinetobacter baumannii complex (n = 120)

Agar dilution 1—128 4 64 78.3 6.7 15

Disk diffusion (diameter, mm) 6—34 76.7 10 13.3 91.7* 8.3 0* 0*

BD Phoenix <0.5/8—>32/8 <0.5/8 >32/8 67.5 1.7 30.8 79.2* 8.3 13.8* 11.1

Vitek 2 <8—>64 <8 32 89.2 5.8 5 82.5* 12.5 0* 33.3

Stenotrophomonas maltophilia (n = 39)

Agar dilution 8—>128 64 >128 5.1 23.1 71.8

Disk diffusion (diameter, mm) 8—26 12.8 51.3 35.9 56.4* 43.6 0 0*

BD Phoenix 1/8—>32/8 >32/8 >32/8 30.8 15.4 53.8 53.8* 35.9 0 14.3

Vitek 2 8— 64 8 16 94.9 2.6 2.6 10.3* 25.6 0 89.3

Other nonfermentative GNBa (n = 30) Agar dilution 4—>128

46.7 36.7 16.7

Susceptibility of Enterobacteriaceae to CFP—SUL

Table 1 (continued)

Species, results Range MIC50 MIC90 Susceptibilities % of susceptibility results with:

_<mg/mL) <mg/mL) <mg/mL) s(%) i(%) r(%) CA mE ME VME

Disk diffusion (diameter, mm) 6—31 60 16.7 23.3 73.3* 26.7 0 0

BD Phoenix <0.5/8—>32/8 32/8 >32/8 26.7 26.7 46.7 40*,c(9) 50b(8) 21.4 0

Vitek 2 <8—>64 32 >64 16.7 56.7 26.7 53.3c(10) 33.3b(9) 21.4b(10) 20

a Other nonfermentative GNB comprises Chyrseobacterium indologenes (n = 10), Elizabethkingia meningoseptica (n = 12), and Burkholderia cepacia complex (n = 8).

b Above average (for all Enterobacteriaceae isolates, with statistically significant differences): 1 5.0%, 2 4.8%, 3 4.4%, 4 5.2%, and 5 91.2%, (and for all nonfermentative GNB, also with statistically significant differences): 6 77.6%, 7 74.2%, 8 15.1%, 9 14.0%, and 10 1.5%.

c Below the average (for all Enterobacteriaceae isolates, with statistically significant differences): 1 91.2%, 2 95.2%, 3 94.8%, and 4 20.0%, (and for all nonfermentative GNB, also with statistically significant differences): 515.1%, 6 14.0%, 7 8.0%, 8 50.8%, 9 77.6%, and 10 74.2%.

CA = categorical agreement; GNB = Gram-negative bacteria; I = intermediate; mE = minor error; ME = major error; MIC = minimum inhibitory concentration; R = resistant; S = susceptible; VME = very major. error.

*The difference of rates of CA, mE, ME as well as VME in CFP—SUL susceptibilities yielded between three methods under evaluation is significant (p < 0.05) by statistical analysis.

categories of CFP—SUL susceptibility. The method was also applied to test the discriminatory power of the susceptibility systems against carbapenem-nonsusceptible (i.e., nonsusceptibility to either of the 2 carbapenem agents) enteric GNB isolates. All statistical calculations were two-tailed, and p < 0.05 was considered statistically significant. All statistical analyses were performed using SPSS software version 17.0 (SPSS Inc., Chicago, IL, USA).

Results

The comparisons of CFP—SUL susceptibility data, and categorical agreement rates between the agar dilution method and the other three in vitro methods for respective GNB species

The MIC range, MIC50 and MIC90 values, range of disk diffusion diameters, and susceptibility profiles of the GNB isolates are presented in Table 1. Nonparametric statistical analyses revealed that isolates of E. cloacae and Salmonella species had significantly smaller disk diffusion diameters than the other enteric GNB species isolates with the exception of K. pneumoniae (data not shown). In addition, CFP—SUL susceptibility evaluated by the Vitek 2 system had much higher ME rates for E. coli isolates than the disk diffusion method and the BD Phoenix system (p < 0.001). The CA rates ranged from 84.0% to 100% for Enterobacteriaceae. Comparison of the results of antimicrobial susceptibility testing (AST) to CFP—SUL obtained from the agar dilution method with those from the other three methods revealed high rates of categorical disagreement for three enterobacterial species, i.e., 10.0% by disk diffusion for S. marcescens, 13.0—15.6% by the three evaluation methods for E. cloacae, and 16.0% by the Vitek 2 system for E. coli isolates. Furthermore, high mE rates were obtained by the Vitek 2 system for E. coli, E. cloacae, and P. aeruginosa isolates (6.7%, 11.7%, and 6.4%, respectively) as well as by the disk diffusion test for S. marcescens isolates (10%). The Vitek 2 system had a very high rate of VME for isolates of E. coli, E. cloacae, and K. pneumoniae (40%,

20%, and 16.7%, respectively; Table 1). Of note, great deviations (with statistically significant differences) from the average values in CFP—SUL susceptibility results against overall Enterobacteriaceae organisms were also seen in the Vitek 2 system for E. coli and in the three comparator methods for E. cloacae isolates (Table 1). By contrast with the enterobacterial isolates, there were complex differences in the AST results in CA and in error rates between the three comparator testing methods, and significant deviations from the overall averages of the organisms for isolates of the nonfermentative GNB species with the exception of P. aeruginosa. Nevertheless, the disk diffusion test was reliable (i.e., CA rates >90%) for correctly categorizing CFP—SUL susceptibility for P. aeruginosa and species of Acb complex (Table 1).

Performance of detection of CFP—SUL nonsusceptibility for three in vitro methods

Fisher's exact test revealed no significant differences in detecting CFP—SUL nonsusceptibility of C. freundii and Salmonella species, and S. marcescens and P. mirabilis isolates between the agar dilution method and the other three methods. By contrast, significant differences were found between the agar dilution method and disk diffusion test [7.3% vs. 6.0%; p < 0.001, odds ratio (OR) 0.014, and 95% confidence interval (CI) 0.004—0.056], as well as the BD Phoenix (7.3% vs. 6.0%; p < 0.001, OR 0.008, 95% CI 0.001—0.057) and Vitek 2 systems (7.3% vs. 14.7%; p = 0.027, OR 0.287, 95% CI 0.123—0.671) for E. coli isolates. As demonstrated in Table 2, there were also prominent differences in detecting CFP—SUL nonsusceptibility for K. pneumoniae, E. cloacae, P. aeruginosa, and species of Acb complex, as well as S. maltophilia and the other miscellaneous nonfermentative GNB isolates between the agar dilution method and the other three evaluation systems.

The performance of the three comparator methods with respect to susceptibility profiles (nonsusceptibility vs. susceptibility) of the isolates of respective GNB species,

+ MODEL

6 S.-S. Jean et al.

overall Enterobacteriaceae and nonfermentative GNB species are illustrated in Table 3. The Vitek 2 evaluation system had a significantly lower sensitivity (44.4%), PPV (18.2%) and CA rate (84.7%) for clinical E. coli isolates than the other two methods. The Vitek 2 system also had markedly low sensitivity (45.5%) for K. pneumoniae isolates. Moreover, the Vitek 2 system performed poorly (sensitivity 65.9%, PPV 57.4%) for all isolates of the seven Enterobacteriaceae species tested. Wide diversities were found in all of the performance parameters of the different testing methods for detecting CFP—SUL nonsusceptible strains of nonfermentative GNB species other than P. aer-uginosa (Table 3).

The ROC curve analysis of the capacity of Vitek 2 system to identify correct susceptibility categories among different Enterobacteriaceae spp

The ROC curve analysis demonstrated that the Vitek 2 system had a significantly poorer discriminatory power to detect the correct categories of susceptibility of E. coli isolates [area under the ROC curve (AUC) 0.654; p = 0.123, 95% CI 0.448—0.859] than those of isolates of E. cloacae

(AUC 0.903; p < 0.001, 95% CI 0.786—1.000) and overall isolates of the seven Enterobacteriaceae species (AUC 0.808; p < 0.001, 95% CI 0.719—0.897; Figures 1A—1C). In addition, we identified a total of 20 enteric GNB isolates (C. freundii, n = 1; E. coli, n = 4; K. pneumoniae, n = 5; and E. cloacae, n = 10) with a carbapenem-nonsusceptible phenotype. The performance of Vitek 2 system was much poorer than the disk diffusion and BD Phoenix methods in correctly classifying CFP—SUL susceptibility of these resistant GNB isolates (AUC 0.756; p = 0.076, 95% CI 0.513—0.999; figure not shown).

Discussion

This study highlights three important points. First, the Vitek 2 system performed poorly in recognizing the correct categories of CFP—SUL susceptibility among E. coli isolates. Second, the CFP—SUL susceptibility profiles obtained by the three comparator methods are in conflict with those from the agar dilution method for E. cloacae strains. Third, all three methods performed poorly in their ability to correctly categorize CFP—SUL susceptibility among the

Table 2 Comparisons of the capacity to detect nonsusceptibility to cefoperazone—sulbactam yielded from the three evaluation methods and the agar dilution method (reference) against Gram-negative bacteria (GNB).

Species, comparisons of the capacity of each method with the reference data in detecting the CFP—SUL nonsusceptible isolates

95% CI

Escherichia coli (n = 150)

Disk diffusion <0.001 0.014 0.004-0.056

BD Phoenix system <0.001 0.008 0.001-0.057

Vitek 2 system 0.027 0.287 0.123-0.671 Klebsiella pneumoniae (n = 150)

Disk diffusion <0.001 0.029 0.011-0.076

BD Phoenix system <0.001 0.022 0.007-0.066

Vitek 2 system <0.001 0.041 0.019-0.091 Enterobacter cloacae (n = 77)

Disk diffusion <0.001 0.018 0.003-0.126

BD Phoenix system <0.001 0.018 0.003-0.126

Vitek 2 system <0.001 0.041 0.010-0.161 Pseudomonas aeruginosa (n = 110)

Disk diffusion <0.001 0.033 0.011-0.100

BD Phoenix system <0.001 0.035 0.011-0.106

Vitek 2 system <0.001 0.023 0.006-0.092 Species of Acinetobacter calcoaceticus-Acinetobacter baumannii complex (n = 120)

Disk diffusion <0.001 0.033 0.011-0.101

BD Phoenix system <0.001 0.180 0.111-0.293

Vitek 2 system <0.001 0.121 0.073-0.202 Stenotrophomonas maltophilia (n = 39)

Disk diffusion 0.013 0.600 0.293-1.227

BD Phoenix system 0.089 0.833 0.647-1.073

Vitek 2 system >0.99 0.946 0.876-1.022 Other nonfermentative GNB (n = 30)a

Disk diffusion 0.001 0.303 0.141-0.651

BD Phoenix system 0.417 0.635 0.243-1.657

Vitek 2 system 0.642 0.714 0.231-2.208

a Other nonfermentative GNB comprised Chyrseobacterium indologenes (n Burkholderia cepacia complex (n = 8).

10), Elizabethkingia

CFP-SUL = cefoperazone-sulbactam; CI = confidence interval; GNB = Gram-negative bacteria; OD

meningoseptica (n = 12), and = odds ratio.

+ MODEL

Table 3 Comparisons of the cefoperazone—sulbactam susceptibility profiles (nonsusceptible vs. susceptible) obtained from the three evaluation methods and the reference (agar dilution) test for various kinds of Gram-negative bacteria.

Species (n)

No. of isolates:

Results (%) for:

+ by both tests

Method + and REF -

Method -and REF

- by both tests

Sensitivity Specificity PPV

Agreement

Escherichia coli (150) Disk diffusion 9 2

BD Phoenix 8 1

Vitek 2 4 18

Klebsiella pneumoniae (150)

140 123

100* 88.9* 44.4*

98.6 99.3 87.2

81.8* 88.9* 18.2*

100 99.3 96.1

98.7* 98.7* 84.7*

Disk diffusion 11 3 0 136 100* 97.8 78.6 100 98.0

BD Phoenix 11 3 0 136 100* 97.8 78.6 100 98.0

Vitek 2 5 0 6 139 45.5* 100 100 95.7 96.0

Enterobacter cloacae (77)

Disk diffusion 17 1 1 58 94.4 98.3 94.4 98.3 97.4

BD Phoenix 17 1 1 58 94.4 98.3 94.4 98.3 97.4

Vitek 2 15 2 3 57 83.3 96.6 88.2 95.0 93.5

Citrobacter freundii (13)

Disk diffusion 1 0 0 12 100 100 100 100 100

BD Phoenix 1 0 0 12 100 100 100 100 100

Vitek 2 1 0 0 12 100 100 100 100 100

Serratia marcescens (30)

Disk diffusion 1 3 0 26 100 89.7 25 100 90

BD Phoenix 0 1 1 28 0 96.6 0 96.6 93.3

Vitek 2 1 0 0 29 100 100 100 100 100

Salmonella spp. (50)

Disk diffusion 1 0 0 49 100 100 100 100 100

BD Phoenix 1 0 0 49 100 100 100 100 100

Vitek 2 1 0 0 49 100 100 100 100 100

Proteus mirabilis (30)

Disk diffusion 0 0 0 30 ND 100 ND 100 100

BD Phoenix 0 2 0 28 ND 93.3 0 100 93.3

Vitek 2 0 0 0 30 ND 100 ND 100 100

Overall isolates of the 7 Enterobacteriaceae species (500)

Disk diffusion 40 9 1 450 97.6* 98* 81.6* 99.8 98*

BD Phoenix 38 8 3 451 92.7* 98.3* 82.6* 99.2 99.3*

Vitek 2 27 20 14 439 65.9* 95.6* 57.4* 96.9 93.2*

Pseudomonas aeruginosa (110)

Disk diffusion 19 3 0 88 100 96.7 86.4 100 97.3

BD Phoenix 18 3 1 88 94.7 96.7 85.7 98.9 96.4

Vitek 2 18 2 1 89 94.7 97.8 90 98.9 97.3

Species of Acinetobacter calcoaceticus- Acinetobacter baumannii complex (120)

Disk diffusion 26 3 1 90 96.3* 96.8* 89.7* 98.9* 96.7*

BD Phoenix 23 15 3 79 88.5* 84* 60.5* 96.3* 85*

Vitek 2 13 0 13 94 50* 100* 100* 87.9* 89.2*

Stenotrophomonas maltophilia (39)

Disk diffusion 34 3 0 2 100* 40 91.9 100* 92.3*

BD Phoenix 27 0 10 2 73* 100 100 16.7* 74.4*

Vitek 2 2 0 35 2 5.4* 100 100 5.7* 10.3*

Other nonfermentative GNB species (30) ia

Disk diffusion 11 1 5 13 68.8 92.9* 91.7* 72.2 80

BD Phoenix 13 9 3 5 81.3 35.7* 59.1* 62.5 60

Vitek 2 14 11 2 3 87.5 21.4* 56* 60 56.7

Overall isolates of the nonfermentative GNB species (299)

Disk diffusion 90 10 6 193 93.8* 95.1* 90* 97* 94.6*

BD Phoenix 81 27 17 174 82.7* 86.6* 75* 91.1* 85.3*

Vitek 2 47 13 51 188 48* 93.5* 78.3* 78.7* 78.6*

a Other nonfermentative GNB (30) isolates comprised Chyrseobacterium indologenes (10), Elizabethkingia meningoseptica (12), and Burkholderia cepacia complex (8).

GNB = Gram-negative bacteria; ND = not done; NPV = negative predictive value; PPV = positive predictive value; REF = reference (indicates the results of agardilution); + = positive result for predicting nonsusceptibility; - = negative resultfor excluding nonsusceptibility.

*The difference of performance in parameters between the method(s) under evaluation and the agar dilution method (reference test) is significant (p < 0.05) by statistical analysis.

+ MODEL

S.-S. Jean et al.

nonfermentative GNB isolates with the exception of P. aeruginosa.

For susceptibility testing by the broth microdilution method (the results equivalent to those of the agar dilution method recommended in CLSI), the CFP—SUL combination regimen consisting of a fixed 2:1 concentration ratio had been shown to maximize the spectrum of activity against many GNB species, and best simulate the parenteral formulation as well as the pharmacokinetics of this combination regimen.5 Furthermore, Barry and Jones5 clearly proved that the in vitro performance of these disks correlated well with that of the broth microdilution CFP—SUL (also at the 2:1 concentration ratio) susceptibility test that

was recommended and outlined elsewhere.6 Consequently, despite the use of disks containing 75 mg CFP and 30 mg SUL (close to 2:1 ratio) in susceptibility testing not being justified by the CLSI,12'13 we adopted these disks to determine the CFP—SUL susceptibility test against the clinically important GNB isolates in this study.

Prior to this investigation, only one study had investigated the abilities of various AST evaluation methods against specific antibiotic agents for clinical heteroresistant E. cloacae and A. baumannii isolates.21 To the best of our knowledge, no other study has compared the performance of the disk diffusion method with that of two commonly used, commercially available automated susceptibility

Figure 1. The discriminatory power [area under the receiver-operating-characteristic (ROC) curve] of the Vitek 2 system to correctly identify categories of susceptibility (susceptible, intermediate, or resistant) to cefoperazone—sulbactam among the isolates of (A) Escherichia coli, (B) Enterobacter cloacae, and (C) overall Enterobacteriaceae species. ROC = receiver-operating-characteristic.

+ MODEL

Susceptibility of Enterobacteriaceae to CFP—SUL 9

systems in evaluating the CFP—SUL susceptibility profiles of such a large collection of the clinically important GNB isolates. Although the automated tests are relatively less labor-intensive for yielding AST data than the disk diffusion method, disk diffusion is still a convenient method for obtaining AST results against antibiotic(s). As shown by Hardy et al,19 resistance among Enterobacteriaceae species to CFP—SUL, as also defined by the broth microdilution method, was highly concordant (85%, 17/20) with that obtained by disk diffusion testing.19 This concordant rate in fact shows an insignificant contrast with the proportion of isolates determined by the disk diffusion method to be nonsusceptible to CFP—SUL in our study (97.6%, p = 0.063).

We found that all three comparator methods performed poorly in detecting nonsusceptible strains of non-fermentative GNB species with the exception of P. aeruginosa. In addition to the high ME rate (6.7%) yielded from the BD Phoenix system in the study conducted by Juretschko et al,22 Donay et al17 also found a strikingly lower CA rate (70.6%) by the BD Phoenix system for P. aeruginosa isolates against piperacillin—tazobactam. By contrast, the CA rate detected by the BD Phoenix system for P. aeruginosa isolates against CFP—SUL (94.5%) in our survey was similar to that also obtained from the BD Phoenix system against CFP (96.6%) in the survey of Menozzi et al.18 Therefore, the BD Phoenix system appears to be an accurate method for testing susceptibility of P. aeruginosa isolates to CFP—SUL.

By comparison with the minimal performance requirements recommended by the National Committee for Clinical Laboratory Standards in 1994 and International Organization for Standardization,20'23 we found that use of the formula suggested by Menozzi et al18 for the Vitek 2 system resulted in significantly low CA rates (< 90%) as well as disproportionately high error rates (mostly much higher than 3.0%) for E. coli and E. cloacae isolates. Among the tested organisms of those two species of the Enter-obacteriaceae family, we found that the Vitek 2 system showed a high number of minor errors among E. coli (n = 10), K. pneumoniae (n = 6), and E. cloacae (n = 9) isolates and a high number of major errors among E. coli (n = 12) isolates. In addition, despite the E. coli isolates accounting for only 30% of all Enterobacteriaceae isolates in our survey, the high number (n = 18) of false non-susceptible E. coli strains detected by the Vitek 2 system has a great impact on calculating the PPV for E. coli isolates {< 20% [4/(4 + 18)]} as well as for Enterobacteriaceae isolates {< 60% [27/(27 + 18+2)]}, as seen in Table 3. By contrast, although higher mE rates were also observed by the disk diffusion method for E. cloacae and S. marcescens (15.6% and 10%, respectively) than for other Enterobacteriaceae species, no VME or ME strain was found for these two species.

The main drawback of this in vitro study is that the resistance mechanisms of the GNB isolates analyzed against ß-lactam agents (especially ß-lactamases) were not well characterized. Of 220 E. coli and K. pneumoniae isolates proven to be ESBL producers, Jang et al24 found that the Vitek 2 system produced high rates of VME (27.4%) when testing susceptibility to cefepime and an error rate of 4.5% for ceftazidime compared to the reference data of broth microdilution test. Lat et al25 also found that the Vitek 2 system resulted in high VME rates on the susceptibility of K.

pneumoniae carbapenemase-producing K. pneumoniae strains to meropenem (27%) and cefepime (67%) when CLSI 2010-determined MIC breakpoints were applied.12,25 In addition, a survey conducted by Pailhories et al26, who recently investigated 14 ertapenem-nonsusceptible E. cloacae strains with potential ESBL production, observed that nine (64.3%) strains with mE in categorical discordance of susceptibility were detected from the Vitek 2 system when compared to MIC data from the agar dilution test. In our survey, the Vitek 2 method correspondingly showed the poorest discriminatory ability in identifying the correct CFP—SUL susceptibilities for carbapenem-nonsusceptible enterobacterial strains among three comparator testing methods. It remains to be investigated if the results of susceptibility testing for MDR strains of some members of the Enterobacteriaceae family against many b-lactam drugs obtained from the Vitek 2 system agree well with those obtained by the reference method. Additionally, the high VME rates of the Vitek 2 system in our survey referred to the very limited VME numbers (2 VMEs for E. coli and E. cloacae and 1 for K. pneumoniae) that were divided by small isolate numbers exhibiting a true CFP—SUL resistant phenotype (5 for E. coli, 6 for K. pneumoniae, and 10 for E. cloacae, respectively) for these three species. If the number of total isolates of respective species was used as a new denominator, significantly lower VME rates will be yielded (1.3% for E. coli, 0.7% for K. pneumoniae, and 2.6% for E. cloacae, respectively) than those in Table 1 of our survey. To avoid considerable deviations in reestimating the VME rates, more E. coli and E. cloacae strains proven with CFP—SUL resistant susceptibility by the reference testing might be needed. Finally, because the half-life of CFP differs considerably from that of SUL,5 this impact regarding in vitro CFP—SUL susceptibilities obtained by disk or any automated method on its application for clinical management is worth being investigated.

In conclusion, although the Vitek 2 system still has acceptable accuracy in detecting the correct CFP—SUL susceptibility categories for isolates of some Enter-obacteriaceae species, we found it to be an unreliable method for E. coli and E. cloacae. Furthermore, using the interpretive criteria applied in our study, none of the three tested methods are appropriate alternative means for yielding data on susceptibility of nonfermentative GNB to CFP—SUL with the exception of the disk diffusion method for P. aeruginosa and species of Acb complex. As the database of automated systems is updated, future studies should be undertaken to explore the changes in AST results obtained by these systems against CFP—SUL for important GNB species.

Conflicts of interest

None declared.

Acknowledgments

This study was partly supported by TTY Biopharm Co. Ltd, (Taipei, Taiwan).

+ MODEL

References

1. Boucher HW, Talbot GH, Bradley JS, Edwards JE, Gilbert D, Rice LB, et al. Bad bugs, no drugs: no ESKAPE! An update from the Infectious Diseases Society of America. Clin Infect Dis 2009;48:1-12.

2. Jean SS, Lee WS, Lam C, Hsu CW, Chen RJ, Hsueh PR. Carba-penemase-producing Gram-negative bacteria: current epidemics, antimicrobial susceptibility and treatment options. Future Microbiol 2015;10:407-25.

3. Rice LB. Mechanisms of resistance and clinical relevance of resistance to ß-lactams, glycopeptides, and fluoroquinolones. Mayo Clin Proc 2012;87:198-208.

4. Akova M. Sulbactam-containing ß-lactamase inhibitor combinations. Clin Microbiol Infect 2008;14:185-8.

5. Barry AL, Jones RN. Criteria for disk susceptibility tests and quality control guidelines for the cefoperazone-sulbactam combination. J Clin Microbiol 1988;26:13-7.

6. Jones RN, Barry AL, Packer RR, Gregory WW, Thornsberry C. In vitro antimicrobial spectrum, occurrence of synergy, and recommendations for dilution susceptibility testing concentrations of the cefoperazone-sulbactam combination. J Clin Microbiol 1987;25:1725-9.

7. Kuo HY, Wang FD, Yen YF, Lin ML, Liu CY. In vitro activities of piperacillin or cefoperazone alone and in combination with ß-lactamase inhibitors against Gram-negative bacilli. New Microbiol 2009;32:49-55.

8. Fass RJ, Gregory WW, D'Amato RF, Matsen JM, Wright DN, Young LS. In vitro activities of cefoperazone and sulbactam singly and in combination against cefoperazone-resistant members of the family Enterobacteriaceae and non-fermenters. Antimicrob Agents Chemother 1990;34:2256-9.

9. Greenberg RN, Baker M, Meade DW, Chang Y, Danko LS. In-vitro antimicrobial studies with the combination of cefoperazone and sulbactam. J Antimicrob Chemother 1993;32:912-3.

10. Jones RN, Guzman-Blanco M, Gales AC, Gallegos B, Castro AL, Martino MD, et al. Susceptibility rates in Latin American nations: report from a regional resistance surveillance program. Braz J Infect Dis 2011;2013(17):672-81.

11. Levin AS. Multiresistant Acinetobacter infections: a role for sulbactam combinations in overcoming an emerging worldwide problem. Clin Microbiol Infect 2002;8:144-53.

12. Clinical and Laboratory Standards Institute. Performance Standards for Antimicrobial Susceptibility Testing: Twentieth Informational Supplement M100-S20. Wayne, PA: CLSI; 2010.

13. Clinical and Laboratory Standards Institute. Performance Standards for Antimicrobial Susceptibility Testing: Twenty-fifth Informational Supplement M100-S25. Wayne, PA: CLSI; 2015.

14. European Committee on Antimicrobial Susceptibility Testing. Breakpoint Tables for Interpretation of MICs and Zone Diameters, Version 5.0. Available at: http://www.eucast.org/ clinical_breakpoints/ [Accessed 25 June 2015].

15. Bin C, Hui W, Renyuan Z, Yongzhong N, Xiuli X, Yingchun X, et al. Outcome of cephalosporin treatment of bacteremia due to CTX-M-type extended-spectrum ß-lactamase-producing Escherichia coli. Diagn Microbiol Infect Dis 2006;56:351-7.

S.-S. Jean et al.

16. Lee N, Yuen KY, Kumana CR. Clinical role of ß-lactam/ß-lac-tamase inhibitor combinations. Drugs 2003;63:1511—24.

17. Donay JL, Mathieu D, Fernandes P, Prégermain C, Bruel P, Wargnier A, et al. Evaluation of the automated phoenix system for potential routine use in the clinical microbiology laboratory. J Clin Microbiol 2004;42:1542—6.

18. Menozzi MG, Eigner U, Covan S, Rossi S, Somenzi P, Dettori G, et al. Two-center collaborative evaluation of performance of the BD phoenix automated microbiology system for identification and antimicrobial susceptibility testing of Gramnegative bacteria. J Clin Microbiol 2006;44:4085—94.

19. Hardy DJ, Barry AL, Fuchs PC, Gerlach EH, McLaughlin JC, Pfaller MA. Error rates in cefoperazone and cefoper-azone—sulbactam disk tests with Enterobacteriaceae and Pseudomonas aeruginosa. Eur J Clin Microbiol Infect Dis 1992; 11:926—30.

20. International Organization for Standardization (ISO). Clinical Laboratory Testing and In Vitro Diagnostic Test Systems. Susceptibility Testing of Infectious Agents and Evaluation of Performance of Antimicrobial Susceptibility Test Devices. Part 2: Evaluation of performance of antimicrobial susceptibility test devices. International Standard 20776—2. Geneva, Switzerland. 2007.

21. Lo-Ten-Foe JR, de Smet AM, Diederen BM, Kluytmans JA, van Keulen PH. Comparative evaluation of the VITEK 2, disk diffusion, Etest, broth microdilution, and agar dilution susceptibility testing methods for colistin in clinical isolates, including heteroresistant Enterobacter cloacae and Acineto-bacter baumannii strains. Antimicrob Agents Chemother 2007; 51:3726—30.

22. Juretschko S, Labombardi VJ, Lerner SA, Schreckenberger PC, Pseudomonas AST Study Group. Accuracies of ß-lactam susceptibility test results for Pseudomonas aeruginosa with four automated systems (BD Phoenix, Microscan WalkAway, Vitek, and Vitek 2). J Clin Microbiol 2007;45:1339—42.

23. National Committee for Clinical Laboratory Standards. Development of in vitro susceptibility testing criteria and quality control parameters: approved guideline M23-A. National Committee for Clinical Laboratory Standards, Villanova, PA: NCCLS; 1994.

24. Jang W, Park YJ, Park KG, Yu J. Evaluation of MicroScan WalkAway and Vitek 2 for determination of the susceptibility of extended-spectrum ß-lactamase-producing Escherichia coli and Klebsiella pneumoniae isolates to cefepime, cefo-taxime and ceftazidime. J Antimicrob Chemother 2013;68: 2282—5.

25. Lat A, Clock SA, Wu F, Whittier S, Della-Latta P, Fauntleroy K, et al. Comparison of polymyxin B, tigecycline, cefepime, and meropenem MICs for KPC-producing Klebsiella pneumoniae by broth microdilution, Vitek 2, and Etest. J Clin Microbiol 2011; 49:1795—8.

26. Pailhoriès H, Cassisa V, Lamoureux C, Chesnay A, Lebreton C, Lemariée C, et al. Discordance in the minimal inhibitory concentrations of ertapenem for Enterobacter cloacae: Vitek 2 system versus Etest and agar dilution methods. Int J Infect Dis 2014; 18:94—6.