| | Comparison of BDPhoenix and VITEK2 automated antimicrobial susceptibility test systems for extended-spectrum beta-lactamase detection in Escherichia coli and Klebsiella species clinical isolatesReceived 16 April 2002; accepted 21 August 2002. Abstract The present study compares the ability to detect extended-spectrum β-lactamases (ESBL) among a collection of 34 ESBL producing clinical isolates belonging to Escherichia coli and Klebsiella species with two new rapid susceptibility and identification instruments—VITEK2 (bioMérieux, Marcy l’Etoile, France) vs. BDPhoenix (BD Biosciences, Sparks, MD). ESBL content in these isolates was previously characterized on the basis of PCR amplification and sequencing results which were used as the reference method in our evaluation. BDPhoenix correctly determined the ESBL outcome for all strains tested (100% detection rate), whereas VITEK2 was not able to detect the ESBL status in 5 isolates (85% detection rate). Detailed analysis revealed that the discrepancies were mainly observed with ‘difficult-to-detect’ strains. Misidentification was either due to low oximino cephalosporin MIC in these strains or was associated with pronounced ‘cefotaximase’ or ‘ceftazidimase’ phenotypes. Klebsiella oxytoca chromosomal β-lactamase (K1) is phenotypically quite similar to ESBL enzymes. In order to evaluate whether the K1 and ESBL enzymes could be discriminated, we expanded our analysis by 8 clinical K. oxytoca strains with K1 phenotypes. VITEK2 gave excellent identification of these strains whereas 7 out of 8 were falsely labeled ESBL-positive by the BDPhoenix system.
1. Introduction  Extended-spectrum β-lactamases (ESBLs) constitute an ever-growing class of mainly plasmid-mediated β-lactamases found in Gram-negative bacilli. They confer resistance to broad spectrum cephalosporins (e.g., ceftazidime, cefotaxime, ceftriaxone and cefpodoxime) and/or the monobactam aztreonam, but do not affect cephamycins, carbapenems and β-lactamase inhibitors such as clavulanic acid Livermore 1995, Bradford 2001. Although most ESBLs confer resistance to one or more of the oximino cephalosporins, some enzymes do not always elevate the MICs to levels high enough to be called resistant according to the breakpoints advocated by the National Committee for Clinical Laboratory Standards NCCLS 2000, NCCLS 2002, despite clear demonstration of clinical failures of the drug if used in therapy Paterson et al 2001, Kim et al 2002. Thus, the accurate and reliable detection of the ESBL mechanism is a prerequisite for successful therapeutic management. For reasons of rapidity and convenience, integrated systems which automatically perform rapid identification and antimicrobial susceptibility testing are increasingly used in daily routine (Ferraro & Jorgensen, 1999). For most current instruments an ESBL prediction has been integrated into the automated procedure. Two detection strategies are in competition: (i) ESBL prediction from the MIC pattern (Sanders et al., 2000) or (ii) using ceftazidime or cefpodoxime as an indicator drug, followed by screening for synergy between oximino cephalosporins and clavulanic acid as recommended by the NCCLS (NCCLS, 2002). However, in order to promote more rapid results, the growth media and conditions as well as the testing procedures in cards and panels of the automated systems are often modified compared with conventional methods (Ferraro & Jorgensen, 1999). Up to now, valid data on the ability of such modifications to accurately detect the ESBL mechanism is still sparse. A second question that needs to be answered in this context concerns the discriminatory power of the automated procedures; currently, there is too little information available on how well these cards and panels can separate between ESBL and other resistance mechanisms that mimic the ESBL phenotype (e.g., K1 enzyme in Klebsiella oxytoca). The present study was conducted to compare the performance of two new automated instruments, VITEK2 (bioMérieux, Marcy l’Etoile, France) and BDPhoenix (BD Biosciences, Sparks, MD), in detecting the ESBL status in a challenge set of genetically characterized ESBL carriers (22 Klebsiella pneumoniae, 9 Escherichia coli and 3 K. oxytoca) and to assess discrimination of the automated microbiology systems between true ESBL content and an ESBL-simulating phenotype (8 K. oxytoca isolates with K1 β-lactamase production). 2. Materials and methods 2.1. Bacterial isolates, culture conditions, and inoculum preparation A total of 42 clinical isolates belonging to E. coli and Klebsiella species, 34 with ESBL and 8 with K1 enzymes, were included in the present study. Isolates were subcultured once on Columbia sheep blood agar before being tested. The inoculum preparation was performed according to the manufacturer’s suggested procedures, using a densitometer device for standardization (McFarland 0.5-0.6) and paying careful attention to use fresh colonies. The inoculated cards and panels were placed into their respective workstation for incubation and reading and were interpreted by the incorporated computerized expert system (VITEK2: Advanced Expert System, AES; BDPhoenix: BDXpert System). Species identification was performed by appropriate routine laboratory methods, such as the ID32E galleries (bioMérieux, Marcy l’Etoile, France), BBL Enterotube II and BBL Crystal E/NF ID System (Becton Dickinson Microbiology Systems, Cockeysville, MD). MICs were used as reference values when evaluating disagreements observed with the automated systems. MICs were determined by standard broth microdilution tests using commercial trays MICRONAUT-S (Merlin Diagnostika, Bornheim, Germany) and Mueller-Hinton broth (Difco Laboratories, Detroit, MI) according to the guidelines of the NCCLS NCCLS 2000, NCCLS 2002. Quality control was performed by running E. coli ATCC 25922, K. pneumoniae DSM 30104, K. pneumoniae ATCC 700603 (SHV-18), and K. oxytoca DSM 5175 for each new lot of cards, panels or MIC trays. 2.2. Characterization of ESBL carriers Clinical ESBL isolates belonging to E. coli and Klebsiella spp. were collected between 1998 and 2001 from individual patients at the Universitätsklinikum Hamburg-Eppendorf (Hamburg, Germany). Only one strain of a given species and susceptibility pattern per patient was accepted. All strains fulfilled the screening recommendations of the NCCLS document M100-S12 (NCCLS, 2002), i.e., they produced inhibition zone diameters of ≤27 mm for cefotaxime (30 μg), ≤27 mm for aztreonam (30 μg), ≤22 mm for ceftazidime (30 μg), or ≤25 mm for ceftriaxone (30 μg). The ESBL status was confirmed by a positive synergy test using the technique recommended by document M100-S12 NCCLS 2000, NCCLS 2002; i.e., in these strains the zone diameters for either ceftazidime plus clavulanic acid or cefotaxime plus clavulanic acid were at least 5 mm larger than the zone diameters for ceftazidime or cefotaxime alone. 2.3. PCR amplification and DNA sequencing ESBL producing isolates were grouped on the basis of PCR amplification and sequencing results. Template DNA was prepared by suspending four colonies from overnight cultures in 0.2 mL of H2O, and heating to 100°C for 10 min. PCRs were performed in 50 μL with the DyNAzyme DNA Polymerase Kit (Finnzymes, Espoo, Finland), using 1.5 mM MgCl2, 15 pmol of each primer and 1 unit DyNAzyme DNA polymerase together with 5 μL of template DNA. Primers for amplification of the TEM gene were as follows: OT-3, ATG AGT ATT CAA CAT TTC CG (corresponding to nucleotides 3950-3967 in the sequence published under accession number J01749 (Sutcliffe, 1978), and OT-4, TTA CCA ATG CTT AAT CAG TGA GG (nucleotides 4810-4788 in the same sequence) giving an amplification product of 860 bp; those for the SHV gene were: OS-5, TTA TCT CCC TGT TAG CCA CC (nucleotides 28-47 in the sequence listed under accession number AF148850 (Bradford, 1999) and OS-6, GAT TTG CTG ATT TCG CTC GG (nucleotides 824-805 in the same sequence) giving a product of 796 bp; those for the CTX-M gene were: CTX-M10for, GCA GCA CCA GTA AAG TGA TGG (nucleotides 277-297 in the sequence published under accession number X92506 (Bauernfeind et al., 1996), and CTX-M10rev, GCG ATA TCG TTG GTG GTA CC (nucleotides 811-192 in the same sequence) giving a product of 534 bp. Amplification parameters included a 4-min denaturation at 94°C, followed by 35 cycles of denaturation (94°C for 30 sec), annealing (60°C for 30 sec), and extension (72°C for 1 min), ending in a final extension period of 72°C for 20 min. PCR-amplified DNA fragments were directly sequenced on both strands by using an automated DNA-sequencing procedure (ABI Prism 310 DNA Sequencer, Applied Biosystems, Foster City, CA) involving fluorescent dye-labeled terminators (Applied Biosystems, Foster City, CA). Sequencing primers were the same as for PCR amplification. 2.4. Characterization of K. oxytoca K1 producers Eight routine K. oxytoca strains met the screening criteria for ESBL production according to NCCLS document M100-S12 (NCCLS, 2002) but were negative in both clavulanate synergy test and PCR amplification of ESBL genes. These isolates had reduced susceptibility, or were resistant, to cefuroxime, ceftriaxone, and aztreonam, whereas ceftazidime and cefotaxime consistently retained susceptibility against these strains. Since this is the typical phenotype for K1 enzyme (Livermore, 1995), these strains were attributed to this mechanism as the primary cause of the resistance observed. 2.5. Antimicrobial susceptibility testing cards and panels The card and panels used for this study were standard European configurations for members of the family Enterobacteriaceae and shared a comparable composition of β-lactam antimicrobials and concentration ranges (in μg/ml) as follows: (i) AST-N011 for VITEK2 (ampicillin, 4-32; ampicillin-sulbactam, 4/2-32/16; cefazolin, 4-64; cefepime, 1-16; cefixime, 0.25-2; cefotaxime, 1-32; ceftazidime, 1-32; cefuroxime, 2-32; imipenem, 2-16; piperacillin, 4-64; piperacillin-tazobactam, 4/4-128/4) and (ii) NMIC/ID-6 Combo Panel for BDPhoenix (ampicillin, 4-16; amoxycillin-clavulanate, 4/2-16/8; aztreonam, 2-16; cefazolin, 4-16; cefotaxime, 2-32; cefpirome, 2-32; cefsulodin, 4-32; ceftazidime, 2-16; cefuroxime, 4-16; imipenem, 1-8; meropenem, 1-8; piperacillin-tazobactam, 4/4-64/4). Whereas the BDPhoenix panel NMIC/ID-6 includes both species identification and antimicrobial testing, the VITEK2 card AST-N011 provides susceptibility testing only, so the species name was manually entered into the instrument in order to enable the instrument’s expert system to ascertain the β-lactamase phenotype. 3. Results 3.1. ESBL MIC profiles Thirty-four clinical E. coli and Klebsiella spp. isolates producing ESBL were grouped on the basis of PCR amplification and sequencing results (Table 1). MICs were determined to obtain the individual resistance profile and were used as reference values when evaluating disagreements observed with the automated systems. Overall, the 34 ESBL strains under study were intermediate or resistant to cefuroxime, the oximino cephalosporins and aztreonam, whereas susceptibility to imipenem (data not shown) and cefoxitin was retained (Table 1). However, the different ESBLs varied in the degree of in vitro resistance they caused. Some, including SHV-5, showed very broad activity and resulted in obvious resistance to all oximino cephalosporins, while others, including SHV-12, SHV-19, and TEM-26, had a ‘ceftazidimase’ phenotype, with much higher MIC values for ceftazidime than for other oximino cephalosporin drugs. A specific deviation toward cefotaxime (‘cefotaximase’) was seen in 4 strains containing CTX-M-1. Remarkably, one ESBL strain (EC-9) was also resistant to cefoxitin, a feature highly unusual for the ESBL enzymes described up to now. Further analysis revealed that EC-9 harbored TEM-52, which was described as a novel type of ESBL capable to hydrolyze the cephamycins (Poyart et al., 1998). It is of note that few strains, most obviously KP-19 and KP-22 harboring SHV-2, demonstrated only marginal ESBL activity, raising only the cefpodoxime MIC to 4 μg/ml while leaving those of cefotaxime, ceftriaxone, and ceftazidime scarcely above the values seen for nonproducers. |
a
CAZ, ceftazidime; CTX, cefotaxime; ATM, aztreonam; CRO, ceftriaxone; CXM, cefuroxime; FOX, cefoxitin; CPD, cefpodoxime.
b
ND, not detected. |
3.2. ESBL detection by automated susceptibility testing systems All 34 isolates were correctly identified as an ESBL producer by the BDPhoenix system and its BDXpert system. In comparison, by using VITEK2, five isolates out of the same collection of challenge strains were misidentified to have a more susceptible β-lactam resistance phenotype by the Advanced Expert System (Table 3). It was not surprising that the discrepancies were mainly observed with difficult-to-detect strains. In two isolates, namely KP-19 and KP-22, the oximino cephalosporin MICs were so uncharacteristically low (i.e., ≤2 μg/ml) that it was impossible for the AES to match that pattern to an ESBL phenotype (Table 1). Therefore, any MIC-based system would have underestimated these difficult-to-detect strains. Only cefpodoxime MICs (>2μg/mL) would have given correct ESBL detection in these two organisms. On the other hand, given the fact that porin-deficient strains are generally resistant or at least less susceptible to cefoxitin, which was not the case in these strains, it remains obscure why these isolates were categorized as ‘impermeability mutants’ by the Advanced Expert System. Discrepancies occurred also with the ‘ceftazidimase’ and ‘cefotaximase’ strains. These strains conferred clear resistance to either ceftazidime or cefotaxime but had only marginal effects on the MIC of other oximino cephalosporins (Table 1). Although the AES program mentioned that a specific deviation toward a single oximino cephalosporin was highly unusual, one K. pneumoniae strain containing ‘ceftazidimase’ SHV-19 (KP-11) was incorrectly identified as SHV-1 high-level producer, and two strains harboring ‘cefotaximase’ CTX-M-1 were falsely categorized as K1 hyperproducer (KX-7) or could not be identified (EC-10). The exact reason for the inability of the VITEK2 to detect the resistances expressed by these particular strains remains unknown but an explanation might be that these phenotypes had not been reported when the instrument was developed and thus they were absent in the database. 3.3. K1 detection by automated susceptibility testing systems K. oxytoca chromosomal β-lactamase (K1) is phenotypically quite similar to ESBL enzymes, and thus difficult to separate from this mechanism. In order to challenge the automated systems, our panel of testing organisms was expanded by 8 clinical K. oxytoca strains with a K1 phenotype (Table 2). The VITEK2 system gave accurate identification in all these strains. In contrast, by the BDPhoenix system and its BDXpert program, respectively, all but one of the K1 strains were falsely flagged as ESBL carriers (Table 3). An explanation might be that in the BDPhoenix system the ESBL testing cascade is triggered by growth in a 1 μg/ml-cefpodoxime screening well. Since MICs for producers of K1 β-lactamase in the studied collection were in part ≥2 μg/ml and thus overlapped with the ESBL results, cefpodoxime could not reliably discriminate between both phenotypes. According to our MIC data (Table 2), cefotaxime or ceftazidime would be much better choices to separate the ESBL phenotype from the K1 mechanism when using the current NCCLS breakpoint (i.e., 2 μg/ml). For example, if a ceftazidime MIC of ≥2 μg/ml had been used as ESBL indicator, the number of falsely ESBL positive isolates would have been reduced to zero. |
a
CAZ, ceftazidime; CTX, cefotaxime; CRO, ceftriaxone; ATM, aztreonam; CXM, cefuroxime; FOX, cefoxitin; CPD, cefpodoxime.
b
phenotypic ESBL confirmatory testing according to NCCLS document M100-S12 (NCCLS, 2002). |
|
a
Abbreviations: K1, K. oxytoca chromosomal K1 β-lactamase; d_OMP, deficient in outer membrane protein (i.e., impermeability mutant); High, high-level producer; ESBL, extended-spectrum β-lactamase.
b
Causes for the error in identification included absence of the correct phenotype in the database (no phenotype), peculiarities in the strains itself leading to uncharacteristically low MICs (low-MIC strain), or incorrect design of the underlying ESBL predicting algorithm (algorithm). |
4. Discussion The serious clinical consequences of an ESBL diagnosis create a great need for laboratory testing to accurately identify the presence of these enzymes in clinical isolates. Although detection of ESBL carriage is a problem for traditional microbiologic techniques, it may be even more so for automated susceptibility testing systems. Two different strategies for the identification of this particular resistance phenotype are in common use. In the VITEK2 system, ESBL prediction is based on the analysis of the MIC pattern to a selection of β-lactam antimicrobials (Sanders et al., 2000). This concept differs from that utilized in the new BDPhoenix system which uses cefpodoxime as a screening drug, followed by testing for synergy between oximino cephalosporins and clavulanic acid. In the present study, we have compared both investigational rapid susceptibility and identification instruments and their expert systems, respectively, in their ability to detect ESBL carriage among a population of clinical E. coli and Klebsiella spp. with ESBL. Although the inclusion of a synergy test (cefotaxime-clavulanate and ceftazidime-clavulanate) in the previously developed VITEK system has proven useful to improve ESBL detection (Sanders et al., 1996), at this time in the VITEK2 system ESBL content is predicted by MIC distribution only (Sanders et al., 2000). Whereas other authors were able to detect even very rarely encountered ESBL variants with this technique Sanders et al 2000, Canton et al 2001, in our hands the MIC-based method failed to identify ESBL carriage for 5 of the 34 ESBL strains tested. In these instances, there was either a strong deviation toward a single oximino cephalosporin in the MIC pattern that made it difficult for the instrument’s expert system to match the susceptibility profile to an ESBL phenotype for the species. In this case, an updated and more detailed database could be helpful to improve the performance. Or the MICs of the oximino cephalosporins were so uncharacteristically low that it was impossible for the expert system to match that pattern to an ESBL phenotype. However, it is a well known phenomenon that many strains producing ESBL demonstrate an inoculum effect in that the oximino cephalosporin MICs rise as the inoculum increases (Jorgensen & Ferraro, 2000). Maybe a higher final inoculum density (i.e., 107 instead of 105 cfu/ml) could have detected this particular cases. Otherwise, the inclusion of a specific ESBL test on the VITEK2 card would be essential. The false-negative results with the VITEK2 system are serious errors since misidentification of an ESBL strain could lead to incorrect chemotherapy and in consequence be clinically fatal. Considering this correlation, poor discrimination of a K1 phenotype appears to be only a minor error. Nevertheless, misidentification of K1 phenotype is likely to result in implementation of expensive infection control measures in most hospitals, and thus is problematic, too. According to our observations and other sources Moland et al 1998, Thomson et al 1999, Gibb and Crichton 2000, a better discrimination of the K1 phenotype could be achieved quite easily if the ESBL predicting algorithm would be modified in a manner that places ceftazidime in the screening position of the testing cascade when testing K. oxytoca. In this context, it is interesting to note that a 1 μg/ml-ceftazidime well is already available on the BDPhoenix panel, but currently its result is processed only after evaluation of growth in the cefpodoxime well. In summary, our study demonstrated that both fully automated antimicrobial susceptibility testing systems, VITEK2 and BDPhoenix, are reliable tools for detection of ESBL in E. coli and Klebsiella spp. For VITEK2, the use of an independent ESBL screen might be useful to overcome the weakness to detect low-MIC ESBL strains which is inherent to any MIC-based system. In BDPhoenix, the ESBL predicting algorithm needs a modification to improve discrimination between the K1 and the ESBL phenotype when testing K. oxytoca strains. Acknowledgements The authors thank BD Diagnostic Systems for kindly supplying the BDPhoenix system and Gram-negative identification and susceptibility panels. We also thank bioMérieux for kindly providing the VITEK2 instrument, and Peter Schäfer and Heimke von Osten for critical reading of the English version of the manuscript. References Bauernfeind et al 1996.
1.
Bauernfeind A, Stemplinger I, Jungwirth R, Ernst S, Casellas JM.
Sequences of β-lactamase genes encoding CTX-M-1 (MEN-1) and CTX-M-2 and relationship of their amino acid sequences with those of other β- lactamases.
Antimicrob Agents Chemother. 1996;40:509–513. MEDLINE Bradford 1999.
2.
Bradford PA.
Automated thermal cycling is superior to traditional methods for nucleotide sequencing of blaSHV genes.
Antimicrob Agents Chemother. 1999;43:2960–2963. MEDLINE Bradford 2001.
3.
Bradford PA.
Extended-spectrum β-lactamases in the 21th century (characterization, epidemiology, and detection of this important resistance threat).
Clin Microbiol Rev. 2001;14:933–951. MEDLINE |
CrossRef
Canton et al 2001.
4.
Canton R, Perez-Vasquez M, Oliver A, Coque TM, Loza E, Ponz F, et al.
Validation of the VITEK 2 and the advanced expert system with a collection of enterobacteriaceae harboring extended spectrum or inhibitor resistant β-lactamases.
Diagn Microbiol Infect Dis. 2001;41:65–70. Abstract | Full Text |
Full-Text PDF (66 KB)
|
CrossRef
Ferraro and Jorgensen 1999.
5.
Ferraro MJ, Jorgensen JH.
Susceptibility testing instrumentation and computerized expert systems for data analysis and interpretation.
In:
Murray PR, Baron EJ, Pfaller MA, Tenover FC, Yolken RH editor. Manual of Clinical Microbiology. 7th Edition. Washington, DC: ASM Press; 1999;. Gibb and Crichton 2000.
6.
Gibb AP, Crichton M.
Cefpodoxime screening of Escherichia coli and Klebsiella spp. by Vitek for detection of organisms producing extended-spectrum β-lactamases.
Diagn Microbiol Infect Dis. 2000;38:255–257. Abstract | Full Text |
Full-Text PDF (48 KB)
|
CrossRef
Jorgensen and Ferraro 2000.
7.
Jorgensen JH, Ferraro MJ.
Antimicrobial susceptibility testing (special needs for fastidious organisms and difficult-to-detect resistance mechanisms).
Clin Infect Dis. 2000;26:973–980. MEDLINE Kim et al 2002.
8.
Kim YK, Pai H, Lee H-J, Park S-E, Choi E-H, Kim J, et al.
Bloodstream infections by extended-spectrum β-lactamase-producing Escherichia coli and Klebsiella pneumoniae in children (epidemiology and clinical outcome).
Antimicrob Agents Chemother. 2002;46:1481–1491. MEDLINE |
CrossRef
Livermore 1995.
9.
Livermore DM.
β-Lactamases in laboratory and clinical resistance.
Clin Microbiol Rev. 1995;8:557–584. MEDLINE Moland et al 1998.
10.
Moland ES, Sanders CC, Thomson KS.
Can results obtained with commercially available MicroScan microdilution panels serve as an indicator of β-lactamase production among Escherichia coli and Klebsiella isolates with hidden resistance to expanded-spectrum cephalosporins and aztreonam.
J Clin Microbiol. 1998;36:2575–2579. MEDLINE NCCLS 2000.
11.
NCCLS.
M7-A5. Methods for Dilution Antimicrobial Susceptibility Tests for Bacteria That Grow Aerobically. Approved Standard. Fifth Edition. Wayne, Pennsylvania, USA: National Committee for Clinical Laboratory Standards; 2000;. NCCLS 2002.
12.
NCCLS.
M100-S12. Performance Standards for Antimicrobial Susceptibility Testing; Twelfth Informational Supplement. Wayne, Pennsylvania, USA: National Committee for Clinical Laboratory Standards; 2002;. Paterson et al 2001.
13.
Paterson DL, Ko W-C, von Gottberg A, Casellas JM, Mulazimoglu L, Klugman KP, et al.
Outcome of cephalosporin treatment for serious infections due to apparently susceptible organisms producing extended-spectrum β-lactamases (implications for the clinical microbiology laboratory).
J Clin Microbiol. 2001;39:2206–2212. MEDLINE |
CrossRef
Poyart et al 1998.
14.
Poyart C, Mugnier P, Quesne G, Berche P, Trieu-Cuot P.
A novel extended-spectrum TEM-type β-lactamase (TEM-52) associated with decreased susceptibility to moxalactam in Klebsiella pneumoniae.
Antimicrob Agents Chemother. 1998;42:108–113. MEDLINE Sanders et al 1996.
15.
Sanders CC, Barry AL, Washington JA, Shubert C, Moland ES, Traczewski MM, et al.
Detection of extended-spectrum-β-lactamase-producing members of the family Enterobacteriaceae with Vitek ESBL test.
J Clin Microbiol. 1996;34:2997–3001. MEDLINE Sanders et al 2000.
16.
Sanders CC, Peyret M, Moland ES, Shubert C, Thomson KS, Boeufgras JM, et al.
Ability of the VITEK 2 advanced expert system to identify β-lactam phenotypes in isolates of Enterobacteriaceae and Pseudomonas aeruginosa.
J Clin Microbiol. 2000;38:570–574. MEDLINE Sutcliffe 1978.
17.
Sutcliffe JG.
Nucleotide sequence of the ampicillin resistance gene of Escherichia coli plasmid pBR322.
Proc Natl Acad Sci USA. 1978;75:3737–3741. MEDLINE |
CrossRef
Thomson et al 1999.
18.
Thomson KS, Sanders CC, Moland ES.
Use of microdilution panels with and without β-lactamase inhibitors as a phenotypic test for β-lactamase production among Escherichia coli, Klebsiella spp., Enterobacter spp., Citrobacter freundii, and Serratia marcescens.
Antimicrob Agents Chemother. 1999;43:1393–1400. MEDLINE a Institut für Medizinische Mikrobiologie und Immunologie, Universitätsklinikum Hamburg-Eppendorf, Hamburg, Germany  Corresponding author. Tel.: +49-40-42803-3147; fax: +49-40-42803-4881.
PII: S0732-8893(02)00481-9 doi:10.1016/S0732-8893(02)00481-9 © 2003 Elsevier Science Inc. All rights reserved. | |
|
FULL TEXT