Received 6 July 2004; returned 14 August 2004; revised 6 September 2004; accepted 8 September 2004

Transcription

1 Journal of Antimicrobial Chemotherapy (2004) 54, DOI: /jac/dkh449 Advance Access publication 7 October 2004 Evaluation of the MicroScan ESBL plus confirmation panel for detection of extended-spectrum b-lactamases in clinical isolates of oxyimino-cephalosporin-resistant Gram-negative bacteria Enno Stürenburg 1 *, Melanie Lang 1, Matthias A. Horstkotte 1, Rainer Laufs 1 and Dietrich Mack 1,2 JAC 1 Institut für Infektionsmedizin, Zentrum für Klinisch-Theoretische Medizin, Universitätsklinikum Hamburg- Eppendorf, Martinistrasse 52, D Hamburg, Germany; 2 Chair of Medical Microbiology and Infectious Diseases, Swansea Clinical School, University of Wales Swansea SA2 8PP, UK Received 6 July 2004; returned 14 August 2004; revised 6 September 2004; accepted 8 September 2004 Objective: We aimed to assess the performance of the MicroScan ESBL plus confirmation panel using a series of 87 oxyimino-cephalosporin-resistant Gram-negative bacilli of various species. Methods: Organisms tested included 57 extended-spectrum b-lactamase (ESBL) strains comprising Enterobacter aerogenes (3), Enterobacter cloacae (10), Escherichia coli (11), Klebsiella pneumoniae (26), Klebsiella oxytoca (3) and Proteus mirabilis (4). Also included were 30 strains resistant to oxyimino cephalosporins but lacking ESBLs, which were characterized with other resistance mechanisms, such as inherent clavulanate susceptibility in Acinetobacter spp. (4), hyperproduction of AmpC enzyme in Citrobacter freundii (2), E. aerogenes (3), E. cloacae (3), E. coli (4), Hafnia alvei (1) and Morganella morganii (1), production of plasmid-mediated AmpC b-lactamase in K. pneumoniae (3) and E. coli (3) or hyperproduction of K1 enzyme in K. oxytoca (6). Results: The MicroScan MIC-based clavulanate synergy correctly classified 50 of 57 ESBL strains as ESBL-positive and 23 of 30 non-esbl strains as ESBL-negative (yielding a sensitivity of 88% and a specificity of 76.7%, respectively). False negatives among ESBL producers were highest with Enterobacter spp. due to masking interactions between ESBL and AmpC b-lactamases. False-positive classifications occurred in two Acinetobacter spp., one E. coli producing plasmid-mediated AmpC b-lactamase and two K. oxytoca hyperproducing their chromosomal K1 b-lactamase. Conclusion: The MicroScan clavulanate synergy test proved to be a valuable tool for ESBL confirmation. However, this test has limitations in detecting ESBLs in Enterobacter spp. and in discriminating ESBL-related resistance from the K1 enzyme and from inherent clavulanate susceptibility in Acinetobacter spp. Keywords: ESBLs, ESBL detection, ESBL discrimination, MIC-based clavulanate synergy test Introduction Production of extended-spectrum b-lactamases (ESBLs) has become a major challenge in treating Gram-negative infections. 1,2 Initially these enzymes were limited to Escherichia coli and Klebsiella species and were mainly restricted to certain geographical areas. Since the late 1990s, however, these enzymes have been spreading and ESBLs are reaching other genera, principally Enterobacter and Proteus. 3 6 Besides the growing species diversity, ESBL phenotypes have become more complex due to the production of multiple enzymes, including inhibitor-resistant ESBL variants, dissemination of CTX-M types, plasmid-borne AmpC, production of ESBLs in AmpC-producing strains, enzyme hyperproduction and porin loss The inability to detect such complex resistance phenotypes may have been a major factor in the uncontrolled spread of ESBL-producing organisms and related treatment failures, without any suspicion by the laboratory or physicians of their presence.... *Corresponding author. Tel: ; Fax: ; e.stuerenburg@uke.uni-hamburg.de JAC vol.54 no.5 q The British Society for Antimicrobial Chemotherapy 2004; all rights reserved.

2 ESBL detection by MicroScan ESBL plus panel Unfortunately, current diagnostic means failed to keep pace with changing conditions. The NCCLS has issued guidelines for phenotypic confirmation of suspected ESBL strains among E. coli, Klebsiella oxytoca and Klebsiella pneumoniae, but there are currently no interpretative criteria available for other genera, or the methods to be applied for ESBL detection against the background of coexisting resistance mechanisms. 16 The NCCLS recommendation still relies on the MIC difference test, in which a b-lactamase inhibitor is used to protect the activity of an indicator drug against an ESBL-producing strain. This test is considered positive when the suspected organism exhibits a lowering of >_3 two-fold dilution steps in its MICs of ceftazidime or cefotaxime in the presence of a fixed concentration (4 mg/l) of clavulanic acid, versus its MIC when tested alone. 16 For daily routine, this protocol has been translated into several customized microtitre formats, the MicroScan ESBL plus panel (Dade Behring) being a newer one among them. Its advantage is that it can be processed in a fully automated fashion, by using the Walkaway 96 SI instrument (Dade Behring, Schwalbach, Germany). In this study, we aimed to determine whether the MicroScan ESBL plus panel would be able to discriminate between ESBLs, AmpC b-lactamases, high levels of K1 b-lactamase and inherent clavulanate susceptibility in isolates of Acinetobacter spp., Enterobacter aerogenes, Enterobacter cloacae, E. coli, Hafnia alvei, K. oxytoca, K. pneumoniae and Morganella morganii. Some of these strains had been used in previous studies of commercially available Etest strips and microdilution panels. 17,18 Molecular procedures including PCR and nucleotide sequencing were used as the reference method. Materials and methods Bacterial strains The bacterial strains selected for this study included 87 oxyiminocephalosporin-resistant isolates of Gram-negative bacteria, the majority of which have been characterized previously. 17,18 The remaining strains were recent isolates gathered from patients attending the university hospital Hamburg-Eppendorf, Germany, affiliated with our institution. Species identification was performed by routine laboratory methods, such as the API32E (biomérieux, Marcy l Étoile, France). PCR of b-lactamase-encoding genes Reference confirmation for ESBL production was by molecular characterization, i.e. by PCR analysis for b-lactamase genes of the families CTX-M, SHV, TEM, OXA, VEB and PER and where applicable by nucleotide sequencing. The primers used to amplify the targeted genes were: TEM-F, 5 0 -TCCGCTCATGAGACAAT- AACC-3 0 and TEM-R, 5 0 -TTGGTCTGACAGTTACCAATGC-3 0 ; SHV-F, 5 0 -TTATCTCCCTGTTAGCCACC-3 0 and SHV-R, 5 0 -GAT- TTGCTGATTTCGCTCGG-3 0 ; 19 CTX-M-F, 5 0 -TCTTCCAGAATA- AGGAATCCC-3 0 and CTX-M-R, 5 0 -CCGTTTCCGCTATTAC- AAAC-3 0 ; OXA-1F, 5 0 -ACACAATACATATCAACTTCGC-3 0 and OXA-1R, 5 0 -AGTGTGTTTAGAATGGTGATC-3 0 ; 20 OXA-2F, 5 0 -TTCAAGCCAAAGGCACGATAG-3 0 and OXA-2R, 5 0 -TCCGA- GTTGACTGCCGGGTTG-3 0 ; 20 OXA-10F, 5 0 -CGTGCTTTGTAAA- AGTAGCAG-3 0 and OXA-10R, 5 0 -CATGATTTTGGTGGGA- ATGG-3 0 ; 20 VEB-1F, 5 0 -CGACTTCCATTTCCCGATGC-3 0 and VEB-1R, 5 0 -GGACTCTGCAACAAATACGC-3 0 ; 21 PER-F, 5 0 -ATG- AATGTCATTATAAAAGC-3 0 and PER-R, 5 0 -AATTTGGGCTTA- GGGCAGAA The nucleotide sequences were determined by bidirectional sequencing of PCR products, carried out by the Big- Dye dideoxy chain termination method on an ABI Prism 310 DNA sequencer (Perkin-Elmer Corp., Foster City, CA, USA). The nucleotide sequences were analysed using commercial software (Vector NTI suite; InforMax Inc.) against the GenBank database. ESBL-positive strains Sensitivity was calculated with 57 ESBL-positive strains, which included 40 bacteria (E. coli, K. oxytoca and K. pneumoniae) targeted by the NCCLS MIC difference test and 17 non-target bacteria, such as E. aerogenes, E. cloacae and Proteus mirabilis. In detail, the clinical isolates consisted of E. aerogenes (n = 3), E. cloacae (n = 10), E. coli (n = 11), K. oxytoca (n = 3), K. pneumoniae (n = 26) and P. mirabilis (n = 4). The E. aerogenes strains analysed harboured CTX-M-1 (n = 2) and SHV-5 (n = 1) ESBLs. The isolates of E. cloacae included SHV-12 (n = 3) and TEM-type (ABL: A184V) (n = 7) ESBLs. The E. coli strains included SHV-2 (n = 2), SHV-5 (n = 1), SHV-12 (n = 3), TEM-26 (n = 1), TEM-52 (n = 1), TEM-111 (n = 1), CTX-M-1 (n = 1) and CTX-M-23 (n = 1). 22 The K. oxytoca isolates harboured CTX-M-1 (n = 2) and SHV-12 (n = 1). The K. pneumoniae strains harboured SHV-2 (n = 5), SHV-2a (n = 1), SHV-5 (n = 4), SHV-12 (n = 11), SHV-18 (n = 1), SHV-19 (n = 1), LEN-type (ABL: N53S, A201P, P218A) (n = 1), TEM-47 (n = 1) and TEM-110 (n = 1). The P. mirabilis strains included CTX-M-1 (n = 1), CTX-M-15 (n = 1), CTX-M-22 (n = 1) and TEM-92 (n = 1). ESBL-negative strains Specificity was calculated with 30 clinical strains resistant to oxyimino cephalosporins that failed to produce any ESBLs according to the PCR-based technique. Rather, in these strains, the likely type of b-lactamase present was inferred on the basis of the overall susceptibility profile. 23,24 The following patterns were used for phenotypic detection of (i) hyperproduction of AmpC chromosomal b-lactamase (AmpC high): resistant to ureidopenicillins and to cephalosporins in the oxyimino group (cefotaxime, cefotaxime/clavulanate, ceftazidime, ceftazidime/clavulanate and cefpodoxime) and the 7-amethoxy group (cefoxitin and cefotetan). 23 (ii) Within the AmpC high group, there were E. coli and K. pneumoniae strains that produced a plasmid-mediated AmpC b-lactamase, as detected by a multiplex PCR assay for detection of plasmid-mediated ampc b-lactamase genes (plasmid AmpC). 8,25 (iii) Hyperproduction of K. oxytoca chromosomal K1 b-lactamase (K1 high): resistant to ureidopenicillins, resistant to cefuroxime and aztreonam, and resistant or intermediate to cefotaxime and ceftriaxone, but susceptible to ceftazidime. 23 By these criteria, the non-esbl strains consisted of four strains of Acinetobacter spp. (interpretative reading not applicable); two strains of Citrobacter freundii (likely mechanism of resistance: AmpC high); three strains of E. aerogenes (AmpC high); three strains of E. cloacae (AmpC high); three strains of E. coli (plasmid AmpC, multiplex PCR: CIT-type); four strains of E. coli (AmpC high), one strain of H. alvei (AmpC high); six strains of K. oxytoca isolates (K1 high); three strains of K. pneumoniae (plasmid AmpC, multiplex PCR: CIT-type); one strain of M. morganii (AmpC high). Panel processing and interpretation Antibiotic susceptibilities were determined by overnight microdilution, with commercial dehydrated panels provided by Dade Behring. The ESBL plus panels were prepared and inoculated according to the manufacturer s recommended procedures. 871

3 E. Stürenburg et al. The inoculated panels were then placed into the Walkaway 96 SI instrument (Dade Behring) for incubation, final reading and interpretation of the results, which was conducted according to NCCLS criteria. 16 The following antimicrobial agents (concentration ranges in mg/l) were used in the MicroScan ESBL plus panel: aztreonam, ; cefepime, 1 32; cefotetan, 1 32; cefpodoxime, ; cefotaxime, ; cefotaxime/clavulanate, 0.125/4 16/4; cefoxitin, 2 32; ceftazidime, ; ceftazidime/clavulanate, 0.125/4 16/4; ceftriaxone, 1 64; imipenem, ; meropenem, ; piperacillin, Data analysis For isolates for which ceftazidime or cefotaxime MICs were >128 mg/l, the MIC was taken as 256 mg/l and the reduction in the ceftazidime (cefotaxime) MIC was calculated from that value. Strains for which the results between the methods were discordant were tested again, and only concordant and reproducible data were analysed. Results A total of 87 strains of Gram-negative bacilli resistant to oxyimino cephalosporins were chosen for the study. Fifty-seven (57) strains were shown by PCR and nucleotide sequencing to contain at least one of the genes bla TEM, bla SHV and bla CTX-M, confirming the ESBL status (Table 1). Thirty additional strains were characterized with other resistance mechanisms, such as inherent susceptibility to clavulanate (four strains of Acinetobacter spp.), hyperproduction of AmpC chromosomal b-lactamase (two strains of C. freundii, three strains of E. aerogenes, three strains of E. cloacae, four strains of E. coli, one strain of H. alvei, one strain of M. morganii), production of plasmid-mediated AmpC b-lactamase (three strains of K. pneumoniae, three strains of E. coli) or hyperproduction of K1 chromosomal b-lactamase (six strains of K. oxytoca) (Table 1). As an indicator for ESBL screening, the susceptibilities of the organisms to the five extended-spectrum b-lactams available on the MicroScan ESBL plus panel were as follows: cefotaxime (98%), cefpodoxime (98%), ceftriaxone (100%), aztreonam (94%) and ceftazidime (96%) when interpreted using the NCCLS criteria. 16 Leaving aside ceftriaxone, these data add support to the NCCLS recommendation that 100% sensitivity of ESBL screening can only be achieved by testing more than one agent. 16 Importantly, no single drug at any one concentration accurately differentiated between strains producing ESBL and non-esbl phenotypes (data not shown). The majority of Table 1. MicroScan MIC difference test results using the ESBL plus confirmation panel CTX/CLA MIC difference test CAZ/CLA MIC difference test Status positive log 2 reduction in MIC log 2 reduction in MIC mean range negative/ ND+ d positive mean range negative/ ND+ d ESBL organisms (n = 57) ( ) ( ) 11 E. aerogenes (n = 3) (5.0) 1 e 3 E. cloacae (n = 10) ( ) ( ) 6 E. coli (n = 11) ( ) ( ) K. oxytoca (n = 3) ( ) ( ) K. pneumoniae (n = 26) a ( ) ( ) P. mirabilis (n = 4) ( ) ( ) 2 ESBL types (n = 58) a TEM (n = 12) ( ) ( ) 6 SHV (n = 36) ( ) ( ) 1 CTX-M (n = 10) ( ) ( ) 4 Non-ESBL strains (n = 30) Acinetobacter spp. (n = 4) ( ) ( ) 2 C. freundii AmpC high (n =2) 2 2 E. aerogenes AmpC high (n =3) 3 3 E. cloacae AmpC high (n =3) 3 3 E. coli AmpC high (n =4) 4 4 E. coli plasmid AmpC (n =3) b,c ( ) ( ) H. alvei AmpC high (n =1) 1 1 K. oxytoca K1 high (n = 6) ( ) 4 6 K. pneumoniae plasmid AmpC (n =3) b 3 3 M. morganii AmpC high (n =1) 1 1 CTX, cefotaxime; CLA, clavulanic acid; CAZ, ceftazidime. a One strain of K. pneumoniae produced two ESBL enzymes. b Plasmid derivates of AmpC b-lactamase included CIT-type enzymes (origin C. freundii). c Two strains of E. coli produced both a plasmid-mediated AmpC b-lactamase and an ESBL d When MIC values were above the test device range, interpretation was non-determinable, ND+. e The dash ( ) sign denotes that no strains were detected. 872

4 ESBL detection by MicroScan ESBL plus panel Figure 1. Plot showing the frequency of the MIC reductions (two-fold dilution steps) for ESBL producers as a whole (circles, n = 57) and for the subset of CTX-M producers (triangles, n = 10) and ESBL-producing Enterobacter strains (squares, n = 13), obtained with the clavulanate synergy test using cefotaxime (CTX) and ceftazidime (CAZ). Negative values in MIC reduction are due to the ability of clavulanic acid to induce the chromosomal AmpC b-lactamase, resulting in MICs of the combinations higher than that of the drug alone. The dashed line represents the current MIC breakpoint for ESBL confirmation (i.e. a lowering of >_ 3 two-fold dilutions in the MIC of either ceftazidime or cefotaxime by clavulanic acid), as recommended by the NCCLS. non-esbl phenotypes would be considered ESBL producers by these criteria, but very few would be confirmed as such by the confirmatory tests (as described below). If a lowering of >_3 two-fold dilutions in the MIC of either ceftazidime or cefotaxime was taken as the criterion for ESBL confirmation, the MIC difference test was able to detect 50 ESBL strains out of the total of 57 identified by molecular characterization (yielding 88% sensitivity). False negatives among ESBL producers were highest in tests with Enterobacter spp. (Table 1 and Figure 1). Addition of clavulanic acid to ceftazidime (or cefotaxime) failed to lower MICs at least eight-fold in tests with nine (or seven) out of 13 ESBL-producing Enterobacter strains. This poor performance was partly due to the ability of clavulanate to induce the chromosomal AmpC b-lactamase, which often resulted in MICs of the combinations higher than that of the drug alone (Table 1 and Figure 1). The remaining ESBL-producing isolates demonstrated a significant clavulanic acid effect with both cefotaxime and ceftazidime (Table 1 and Figure 1). Only in tests with the CTX-M producers was cefotaxime clearly the single best antibiotic in its ability to confirm ESBL status, since clavulanate synergy was much more pronounced with cefotaxime (mean log 2 reduction in MIC, 8.7) than with ceftazidime (mean log 2 reduction in MIC, 4.8). In fact, when used alone, MIC difference testing with ceftazidime would have failed to recognize four out of 10 CTX-Mproducing ESBL strains (Table 1 and Figure 1). Among the 30 isolates resistant to oxyimino cephalosporins but lacking ESBLs, we observed apparently positive results in MIC difference testing (three- to eight-fold increase in the MIC) with cefotaxime and ceftazidime in seven and five cases, respectively (Table 1 and Figure 2). Since the NCCLS requires only one of the ESBL tests to be positive for an organism to confirm ESBL production, this resulted in an overall specificity of only 76.7% (23/30 strains). Misidentification occurred in Acinetobacter spp. Figure 2. Plot showing the frequency of the MIC reductions (two-fold dilution steps) for non-esbl resistance phenotypes (circles: hyperproduction of AmpC chromosomal b-lactamase; triangles: hyperproduction of K1 chromosomal b-lactamase; squares: plasmid-mediated AmpC b-lactamase; diamonds: Acinetobacter spp.), obtained with the clavulanate synergy test using cefotaxime (CTX) and ceftazidime (CAZ). The dashed line represents the current MIC breakpoint for ESBL confirmation (i.e. a lowering of >_3 two-fold dilutions in the MIC of either ceftazidime or cefotaxime by clavulanic acid), as recommended by the NCCLS. (two strains), E. coli producing plasmid-mediated AmpC b-lactamase (three strains) and K. oxytoca hyperproducing their chromosomal K1 b-lactamase (two strains). An effect of clavulanic acid on the activities of both cefotaxime and ceftazidime was noted in Acinetobacter spp. and E. coli producing plasmidmediated AmpC enzymes, whereas in K. oxytoca hyperproducing K1 enzyme only cefotaxime was affected (Table 1 and Figure 2). With strains giving inconsistent results, both molecular characterization and susceptibility tests were repeated. On retesting, two E. coli were detected to have both a plasmid-mediated AmpC b-lactamase and an ESBL (CIT-type enzyme plus CTX-M-15; CIT-type enzyme plus CTX-M-33). Coexistence of AmpC and ESBL, however, does not provide a full explanation, since responsiveness to clavulanate remained unexplained in one repeatedly ESBL-negative E. coli strain. Even so, excluding the two E. coli strains producing both an ESBL and an AmpC enzyme from the calculation enhanced specificity to 82.1% (23/28 strains). Discussion The NCCLS has issued recommendations for the screening and confirmation of ESBLs in isolates of E. coli and Klebsiella spp. using broth microdilution methods. 16,26 The MicroScan product line offers a fully automated procedure adapting the NCCLS recommendation to confirm ESBL production in suspected strains. However, as b-lactamases are becoming more and more complex, several practical problems emerge in terms of following the NCCLS recommendations. In our study, we have evaluated the MicroScan procedure with regard to its discriminatory potential between ESBL production and other non-esblassociated mechanisms of resistance to extended-spectrum cephalosporins in various Gram-negative bacteria. One issue is that the presence of ESBLs can be masked by the expression of AmpC b-lactamases, which can be generated 873

5 E. Stürenburg et al. by chromosomal (e.g. in most Enterobacter spp., Serratia spp., C. freundii, M. morganii, Providentia spp. and Pseudomonas aeruginosa) or plasmid genes (mostly in E. coli and Klebsiella spp.). Plasmid-mediated AmpC b-lactamases are thought to have originated from the chromosomes of the former Enterobacteriaceae species. 27 Regardless of background, AmpC enzyme can interfere with clavulanate synergy tests. 28,29 With the chromosomal type of enzyme, clavulanic acid may act as an inducer of high-level production, and may then attack the indicator cephalosporin, thus masking any synergy arising from a coexisting ESBL. 28,29 Accordingly, in the present study, clavulanatebased synergy testing failed to detect reliably ESBL production in tests with 54% 69% ESBL-producing Enterobacter strains. The overall ability of clavulanic acid to lower the MICs of either cefotaxime or ceftazidime, due to these constraints, was diminished to 50/57 strains (88% sensitivity) and 46/57 strains (81% sensitivity), respectively. Even though they are not inducible, plasmid-encoded AmpC b-lactamases typically are expressed at medium to high levels. Like their counterpart on the chromosome, plasmid-encoded AmpC b-lactamases provide a broader spectrum of resistance than ESBL and are not blocked by commercially available inhibitors. 8,27 Thus, high-level expression of a plasmid-mediated AmpC enzyme may also prevent recognition of an ESBL. In our study, two E. coli strains produced both a CTX-M b-lactamase and a CIT-type plasmid-mediated AmpC b-lactamase. In preliminary studies with these strains, dominant AmpC production covered and masked underlying ESBL production. Thus, these strains were initially considered only in the AmpC group until retesting was performed and the ESBL content was detected. Approaches to overcome these difficulties include the use of tazobactam or sulbactam, which are much less likely to induce AmpC b-lactamases and are therefore preferable inhibitors for ESBL detection tests with these organisms, or testing cefepime as an ESBL detection agent. 28 Cefepime is a more reliable detection agent for ESBLs in the presence of an AmpC b-lactamase, as this drug is stable to AmpC b-lactamases, but labile to ESBLs. 28 In a previous study employing the Etest ESBL, cefepime with and without clavulanic acid had promising utility in identifying ESBLs in isolates that also contained an AmpC b-lactamase. 18 Another issue of vital importance is which indicator cephalosporin should be tested. Synergy testing may fail to identify CTX-M positive isolates as ESBL producers if ceftazidime is used as the sole detection agent (as seen in the subset analysis of the CTX-M producers tested in this study). Relying on cefotaxime alone, on the other hand, is unreliable in distinguishing ESBL producers from hyperproducers of the K1 enzyme in K. oxytoca. As shown previously with the ESBL Etest and (as here) with a microdilution technique, many of the cefotaxime synergy-positive K. oxytoca isolates are hyperproducers of their chromosomal K1 b-lactamase and are lacking an ESBL. 17,18,30 Which bacteria should be excluded from routine ESBL testing has not received much attention. In the present study, we have tested six Acinetobacter strains. We observed ostensibly positive results with cefotaxime and ceftazidime in two out of six strains, although PCR analysis showed that these strains were lacking an ESBL. These results confirmed the suspicion that Acinetobacters often give apparently positive results, due to their susceptibility of the inhibitors themselves, and thus should not be tested for ESBL production by inhibitor-based methods. A similar difficulty exists with Stenotrophomonas, where clavulanate synergy tests can produce false-positive results via inhibition of the L2 chromosomal b-lactamase. 34 Given that ESBLs are spreading and are reaching non-fermenters as well, principally Acinetobacter, these particularities could become clinically important in the future. Overall, this study demonstrated that the MicroScan MIC difference test, using both ceftazidime and cefotaxime, is a sensitive tool in its ability to confirm the ESBL status of a given Gram-negative isolate. However, as with other techniques, its constraints must be recognized. First, and most importantly, clavulante synergy can give false-negative results in an AmpC environment, whether chromosomal or plasmid-encoded. Secondly, false-positive ESBL results can occur, due to the presence of other clavulanate-responsive b-lactamases, mainly the K1 b-lactamase of K. oxytoca, or due to clavulanate susceptibility, which is inherent to the species, as for Acinetobacter spp. As ESBL phenotypes become more and more complex, emphasis on an upgrade of the NCCLS recommendations seems to be warranted. Acknowledgements We thank Djahesh Noor for excellent technical assistance on this project and Brigitta Pölkner, Jana Lauterberg, Bärbel Erichsen and Kerstin Wagner (all Dade Behring) for making this study possible. We are grateful to the following for sharing their ESBL strains with us for this research project: Mrs Aline Wenger, Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland (10 E. cloacae isolates); Dr Elke Halle, Institut für Mikrobiologie und Hygiene, Campus Charité Mitte, Berlin, Germany (one P. mirabilis strain); Dr Karin Schwegmann, Niedersächsisches Landesgesundheitsamt, Hannover, Germany (one P. mirabilis strain). This study was supported by Dade Behring. References 1. Stürenburg, E. & Mack, D. (2003). Extended-spectrum b-lactamases: implications for the clinical microbiology laboratory. Journal of Infection 47, Bradford, P. A. (2001). Extended-spectrum b-lactamases in the 21st century: characterization, epidemiology, and detection of this important resistance threat. Clinical Microbiology Reviews 14, Albertini, M. T., Benoit, C., Berardi, L. et al. (2002). Surveillance of methicillin-resistant Staphylococcus aureus (MRSA) and Enterobacteriaceae producing extended-spectrum b-lactamase (ESBLE) in Northern France: a five-year multicentre incidence study. Journal of Hospital Infection 52, Spanu, T., Luzzaro, F., Perilli, M. et al. (2002). Occurrence of extended-spectrum b-lactamases in members of the family Enterobacteriaceae in Italy: implications for resistance to b-lactams and other antimicrobial drugs. Antimicrobial Agents and Chemotherapy 46, Coudron, P. E., Moland, E. S. & Sanders, C. C. (1997). Occurrence and detection of extended-spectrum b-lactamases in members of the family Enterobacteriaceae at a veterans medical center: seek and you may find. Journal of Clinical Microbiology 35, Bonnet, R., De Champs, C., Sirot, D. et al. (1999). Diversity of TEM mutants in Proteus mirabilis. Antimicrobial Agents and Chemotherapy 43,

6 ESBL detection by MicroScan ESBL plus panel 7. Bonomo, R. A. & Rice, L. B. (1999). Inhibitor resistant class A b-lactamases. Frontiers in Bioscience 4, e34 e Philippon, A., Arlet, G. & Jacoby, G. A. (2002). Plasmiddetermined AmpC-type b-lactamases. Antimicrobial Agents and Chemotherapy 46, Xiang, X., Shannon, K. & French, G. (1997). Mechanism and stability of hyperproduction of the extended-spectrum b-lactamase SHV-5 in Klebsiella pneumoniae. Journal of Antimicrobial Chemotherapy 40, Martinez-Martinez, L., Pascual, A., Hernandez-Alles, S. et al. (1999). Roles of b-lactamases and porins in activities of carbapenems and cephalosporins against Klebsiella pneumoniae. Antimicrobial Agents and Chemotherapy 43, Gazouli, M., Tzouvelekis, L. S., Prinarakis, E. E. et al. (1996). Transferable cefoxitin resistance in enterobacteria from Greek hospitals and characterization of a plasmid-mediated group 1 b-lactamase (LAT-2). Antimicrobial Agents and Chemotherapy 40, Gazouli, M., Tzouvelekis, L. S., Vatopoulos, A. C. et al. (1998). Transferable class C b-lactamases in Escherichia coli strains isolated in Greek hospitals and characterization of two enzyme variants (LAT-3 and LAT-4) closely related to Citrobacter freundii AmpC b-lactamase. Journal of Antimicrobial Chemotherapy 42, M Zali, F. H., Heritage, J., Gascoyne-Binzi, D. M. et al. (1997). Transcontinental importation into the UK of Escherichia coli expressing a plasmid-mediated AmpC-type b-lactamase exposed during an outbreak of SHV-5 extended-spectrum b-lactamase in a Leeds hospital. Journal of Antimicrobial Chemotherapy 40, Stürenburg, E., Sobottka, I., Mack, D. et al. (2002). Cloning and sequencing of Enterobacter aerogenes OmpC-type osmoporin linked to carbapenem resistance. International Journal of Medical Microbiology 291, Bonnet, R. (2004). Growing group of extended-spectrum b-lactamases: the CTX-M enzymes. Antimicrobial Agents and Chemotherapy 48, National Committee for Clinical Laboratory Standards. (2003). Performance Standards for Antimicrobial Susceptibility Testing Thirteenth Informational Supplement M100-S13. NCCLS, Wayne, PA, USA. 17. Stürenburg, E., Sobottka, I., Feucht, H. H. et al. (2003). Comparison of BDPhoenix and VITEK2 automated antimicrobial susceptibility test systems for extended-spectrum b-lactamase detection in Escherichia coli and Klebsiella species clinical isolates. Diagnostic Microbiology and Infectious Disease 45, Stürenburg, E., Sobottka, I., Noor, D. et al. (2004). Evaluation of a new cefepime-clavulanate ESBL Etest to detect extended-spectrum b-lactamases in an Enterobacteriaceae strain collection. Journal of Antimicrobial Chemotherapy 54, Arlet, G., Rouveau, M. & Philippon, A. (1997). Substitution of alanine for aspartate at position 179 in the SHV-6 extended-spectrum b-lactamase. FEMS Microbiology Letters 152, Steward, C. D., Rasheed, J. K., Hubert, S. K. et al. (2001). Characterization of clinical isolates of Klebsiella pneumoniae from 19 laboratories using the National Committee for Clinical Laboratory Standards extended-spectrum b-lactamase detection methods. Journal of Clinical Microbiology 39, Poirel, L., Le Thomas, I., Naas, T. et al. (2000). Biochemical sequence analyses of GES-1, a novel class A extended-spectrum b-lactamase, and the class 1 integron In52 from Klebsiella pneumoniae. Antimicrobial Agents and Chemotherapy 44, Stürenburg, E., Kühn, A., Mack, D. et al. (2004). A novel extended-spectrum b-lactamase CTX-M-23 with a P167T substitution in the active-site omega loop associated with ceftazidime resistance. Journal of Antimicrobial Chemotherapy 54, Livermore, D. M. (1995). b -Lactamases in laboratory and clinical resistance. Clinical Microbiology Reviews 8, Livermore, D. M., Winstanley, T. G. & Shannon, K. P. (2001). Interpretative reading: recognizing the unusual and inferring resistance mechanisms from resistance phenotypes. Journal of Antimicrobial Chemotherapy 48, Perez-Perez, F. J. & Hanson, N. D. (2002). Detection of plasmid-mediated AmpC b-lactamase genes in clinical isolates by using multiplex PCR. Journal of Clinical Microbiology 40, National Committee for Clinical Laboratory Standards. (2003). Methods for Dilution Antimicrobial Susceptibility Tests for Bacteria that Grow Aerobically Sixth Edition: Approved Standard M7-A6. Wayne, PA, USA. 27. Hanson, N. D. (2003). AmpC b-lactamases: what do we need to know for the future? Journal of Antimicrobial Chemotherapy 52, Thomson, K. S. (2001). Controversies about extended-spectrum and AmpC b-lactamases. Emerging Infectious Diseases 7, Livermore, D. M. & Brown, D. F. (2001). Detection of b-lactamase-mediated resistance. Journal of Antimicrobial Chemotherapy 48,Suppl 1, Potz, N. A., Colman, M., Warner, M. et al. (2004). False-positive extended-spectrum b-lactamase tests for Klebsiella oxytoca strains hyperproducing K1 b-lactamase. Journal of Antimicrobial Chemotherapy 53, Urban, C., Go, E., Mariano, N. et al. (1995). Interaction of sulbactam, clavulanic acid and tazobactam with penicillinbinding proteins of imipenem-resistant and -susceptible Acinetobacter baumannii. FEMS Microbiology Letters 125, Higgins, P. G., Wisplinghoff, H., Stefanik, D. et al. (2004). In vitro activities of the b-lactamase inhibitors clavulanic acid, sulbactam, and tazobactam alone or in combination with b-lactams against epidemiologically characterized multidrug-resistant Acinetobacter baumannii strains. Antimicrobial Agents and Chemotherapy 48, Pandey, A., Kapil, A., Sood, S. et al. (1998). In vitro activities of ampicillin-sulbactam and amoxicillin-clavulanic acid against Acinetobacter baumannii. Journal of Clinical Microbiology 36, Walsh, T. R., MacGowan, A. P. & Bennett, P. M. (1997). Sequence analysis and enzyme kinetics of the L2 serine b-lactamase from Stenotrophomonas maltophilia. Antimicrobial Agents and Chemotherapy 41, Poirel, L., Karim, A., Mercat, A. et al. (1999). Extended-spectrum b-lactamase-producing strain of Acinetobacter baumannii isolated from a patient in France. Journal of Antimicrobial Chemotherapy 43, Vahaboglu, H., Ozturk, R., Aygun, H. et al. (1997). Widespread detection of PER-1-type extended-spectrum b-lactamases among nosocomial Acinetobacter and Pseudomonas aeruginosa isolates in Turkey: a nationwide multicenter study. Antimicrobial Agents and Chemotherapy 41, Nordmann, P. (1998). Trends in b-lactam resistance among Enterobacteriaceae. Clinical Infectious Diseases 27, S