Effects of Phenotype and Genotype on Methods for Detection of Extended-Spectrum-β-Lactamase-Producing Clinical Isolates of Escherichia coli and Klebsiella pneumoniae in Norway

Study design.Consecutive nonduplicate isolates of E. coli and K. pneumoniae with reduced susceptibilities to oxyimino-cephalosporins (MIC > 1 mg/liter) were collected in 18 of 24 Norwegian diagnostic microbiology laboratories covering >90% of the Norwegian population from March through October 2003. Initial antimicrobial susceptibility testing was performed in each laboratory using agar disk diffusion systems from AB Biodisk (Solna, Sweden) or Rosco tablets (Taastrup, Denmark) on paper disk method agar (AB Biodisk) and/or the automated systems Vitek2 (bioMérieux, Marcy l'Etoile, France) and the MAST multipoint system (Mast Diagnostics, Merseyside, United Kingdom), in agreement with breakpoints from the Norwegian Working Group on Antibiotics (NWGA) (16). All laboratories used either cefotaxime (CTX), ceftazidime (CAZ), or cefpodoxime (CPD) alone or two of these substrates in various combinations to screen for reduced susceptibility to oxyimino-cephalosporins. Isolates expressing reduced susceptibilities to oxyimino-cephalosporins were submitted to the Reference Centre for Detection of Antimicrobial Resistance, Tromsø, Norway, with a registration form containing information on sex, age, inpatient and outpatient status, hospital department, and specimen type. Final bacterial identification was performed at the Reference Centre using the Vitek2 ID-GNB system (bioMérieux) or API ID32E (bioMérieux) and/or 16S rRNA gene sequence typing in cases of low discrimination. Strains confirmed as E. coli or K. pneumoniae were included in the study.

Antimicrobial susceptibility testing.E. coli (n = 89) and Klebsiella pneumoniae (n = 27) isolates with reduced susceptibilities to oxyimino-cephalosporins in the initial testing were examined at the Reference Centre using the following panel of Etest β-lactams (AB Biodisk) according to the manufacturer's instructions: ampicillin, amoxicillin-clavulanic acid (CLA), piperacillin, piperacillin-tazobactam, cefoxitin, CPD, CTX, CAZ, cefepime (FEP), aztreonam, imipenem, and meropenem. Vitek2 ASTN023 was used to determine susceptibility to non-β-lactam antibiotics. Interpretations were in accordance with NWGA guidelines. Breakpoints for cefpodoxime have not been established by the NWGA, and a MIC of >1 mg/liter was thus defined as indicative of reduced susceptibility.

Phenotypic detection of ESBL production.Phenotypic tests were performed on the same day from the same subculture with ampicillin selection (100 mg/liter). ESBL production on isolates expressing a reduced susceptibility (MIC > 1 mg/liter) to an oxyimino-cephalosporin (cefpodoxime and/or cefotaxime and/or ceftazidime) was examined using (i) CTX-CLA, CAZ-CLA, and FEP-CLA ESBL Etests and (ii) disks containing cefpodoxime, ceftazidime, or cefotaxime with and without CLA (called the combined disk method) (Oxoid, Basingstoke, United Kingdom). An ESBL phenotype was defined by reduced susceptibility (MIC > 1 mg/liter) to an oxyimino-cephalosporin (cefpodoxime and/or cefotaxime and/or ceftazidime) and a significant increase in susceptibility to oxyimino-cephalosporins tested in combination with CLA by the Etest and/or the combined disk method. For Etest analyses, ESBL production was defined as a ≥8-fold decrease in the MIC of cefotaxime, ceftazidime, or cefepime in the presence of CLA or the presence of so-called phantom or deformity zones. In the combined disk method, ESBL production was defined as an increase of ≥5 mm in the zone around CLA disks compared to the zones of corresponding disks without CLA. In comparison, a modified version of the Jarlier double-disk synergy (DDS) method (10) for detecting CLA synergy was used. Aztreonam (30 μg), cefpodoxime (10 μg), ceftazidime (30 μg), cefotaxime (5 μg), and cefpirome (30 μg) disks (Oxoid) were placed around an amoxicillin (20 μg)-clavulanic acid (10 μg) disk at a distance of 25 to 30 mm center to center. A clearly visible extension of the edge of the inhibition zone of any disk towards the amoxicillin-clavulanic acid disk was interpreted as positive for CLA synergy.

DNA analyses.Bacterial DNA extraction was performed in a QIAGEN model M48 BioRobot (QIAGEN, Hilden, Germany) using a MagAttract DNA mini M48 kit (QIAGEN). 16S rRNA gene PCR-positive DNA extracts were screened for the presence of bla_TEM, bla_SHV, and bla_CTX-M by consensus PCRs in a GeneAmp PCR system 9700 (Applied Biosystems, Foster City, CA) using Applied Biosystems standard PCR mixtures with GeneAmp PCR buffer and Taq DNA polymerase. PCR information is given in Table 1. Bidirectional sequencing was performed using a BigDye v. 3.1 cycle sequencing kit and a model 3100 genetic analyzer (Applied Biosystems). Editing and alignment of DNA sequences were performed using the SeqMan II software package (DNAStar, Inc., Madison, WI).

Isoelectric focusing of β-lactamases.Analytic isoelectric focusing (IEF) of sonicated crude cell extracts was performed in precast Ampholine PAGplate polyacrylamide gels with a pH range of 3.5 to 9.5 (GE Healthcare, St. Giles, United Kingdom) using a Multiphor II apparatus (GE Healthcare). β-Lactamase activity was detected with nitrocefin (0.5 g/liter). The β-lactamases TEM-1 (isoelectric point [pI] 5.4) and SHV-1 (pI 7.6) and IEF protein standards of pI 4.45 to 9.6 (Bio-Rad Laboratories) were used for pI comparisons.

Quality control strains.E. coli J62 (bla_TEM-3), E. coli (bla_CTX-M-3), Kluyvera georgiana (bla_KLUG-1), K. pneumoniae ILT-2 (bla_CTX-M-14), and K. pneumoniae ILT-3 (bla_CTX-M-19), kindly provided by David Livermore and Laurent Poirel, as well as K. pneumoniae ATCC 700603 (bla_SHV-18) and E. coli ATCC 25922, were used.

ESBL detection and bla genotyping in E. coli.An ESBL phenotype was recognized in 52/87 (60%) isolates expressing reduced susceptibility to an expanded-spectrum cephalosporin by Etest. Fifty isolates were ESBL positive by the combined disk method. Fifty-two isolates were positive by the ESBL Etests and included those positive by the combined disk method.

Fifty of the 52 (96%) ESBL phenotype-positive E. coli isolates carried the bla_CTX-M, bla_SHV, or bla_TEM ESBL gene. The PCR results are summarized in Table 2. bla_CTX-M was detected in 45 isolates (90%). CTX-M sequence grouping and typing performed according to the method of Bonnet (3) revealed the CTX-M-1 group (n = 29), the CTX-M-9 group (n = 15), and the CTX-M-2 group (n = 1). Within the CTX-M-1 group, bla_CTX-M-15/28 (n = 23) was the most prevalent genotype. Sequence typing within the CTX-M-9 group (n = 15) revealed bla_CTX-M-9/9a (n = 4), bla_CTX-M-16 (n = 1), and various indistinguishable bla_CTX-M-9 genogroup (bla_{TOHO2/3/CTX-M-14/17/18/21/24}) types (n = 10). bla_SHV and bla_TEM were detected in 4 and 34 isolates, respectively. bla_SHV typing revealed bla_SHV-1 (n = 2) and bla_SHV-ESBL (n = 2). The 34 bla_TEM genes were detected in 31 bla_CTX-M-positive and three bla_SHV- and bla_CTX-M-negative strains. Sequence typing of the latter showed bla_TEM-52 (n = 1) and bla_TEM-128 (n = 2). bla_TEM-1 was detected in six randomly selected bla_CTX-M-positive isolates. Two E. coli isolates were negative for CLA synergy in the combined disk method but positive by CAZ-CLA and FEP-CLA but not CTX-CLA ESBL Etests. Both isolates contained bla_SHV-1 and had phenotypic profiles consistent with hyperproduction of SHV-1: moderate increases in ceftazidime MICs (2 to 4 mg/liter), wild-type cefotaxime MICs (0.125 to 0.25 mg/liter), and piperacillin-tazobactam MICs of >256 mg/liter. IEF analysis revealed single β-lactamase pI bands of approximately 7.5, consistent with an SHV-like enzyme in both strains (data not shown). In summary, the overall prevalence of ESBLs in E. coli isolates was 50 of 87 isolates (58%).

The accuracy of the DDS method was evaluated in comparison to the results obtained by ESBL Etests, the combined disk method, and bla typing. The DDS method revealed CLA synergy with at least one substrate with 55/87 isolates, including all strains that scored positive in the ESBL Etest analysis. The two SHV-1-hyperproducing strains also expressed CLA synergy by the DDS method. Interestingly, three isolates with negative results by both the ESBL Etest and the combined disk method displayed reproducible CLA synergy to aztreonam and cefpirome in the DDS method. The isolates shared common features, including CLA synergy to aztreonam and cefpirome and negative bla_{CTX-M/SHV/TEM} PCRs. IEF analyses revealed a single β-lactamase band for which the pI was 9.0, consistent with their AmpC profile of moderately elevated cefoxitin MICs (32 to 48 mg/liter) and increased MICs of oxyimino-cephalosporins and aztreonam (cefpodoxime, 24 to 48 mg/liter; cefotaxime, 2 to 4 mg/liter; ceftazidime, 2 to 6 mg/liter; and aztreonam, 2 to 4 mg/liter). In summary, we have no explanation for the CLA synergy observed in 3 out of 55 DDS-positive isolates. These strains were not defined as ESBL positive.

ESBL detection and bla genotyping in K. pneumoniae.An ESBL phenotype was recognized in 21/25 (84%) K. pneumoniae isolates expressing reduced susceptibility to an expanded-spectrum cephalosporin by the Etest. Eighteen isolates were ESBL positive by the combined disk method. Twenty-one isolates were positive by the ESBL Etests and included those positive by the combined disk method.

A total of 19 ESBL genes, namely, bla_CTX-M (n = 3), bla_SHV (n = 15), and bla_TEM (n = 1), were detected in 19 isolates (Table 2). bla_SHV-5, bla_SHV-5a, and bla_SHV-12 were the most prevalent ESBL genotypes. bla_CTX-M amplicons were typed as being in the bla_CTX-M-15/28 (n = 2) and bla_CTX-M-9 (n = 1) genogroups. The three bla_CTX-M strains were also shown to contain bla_SHV-11 or bla_SHV-14. Both bla_SHV-11 and bla_SHV-14 have previously been reported to express a non-ESBL phenotype in K. pneumoniae (15, 30). The bla_TEM-52 (n = 1) and bla_TEM-1 (n = 8) genes were detected in nine strains. Discordant ESBL phenotype results were observed with three K. pneumoniae strains. Interestingly, one bla_SHV-28 isolate expressed significant CLA synergy in FEP-CLA and CAZ-CLA ESBL Etests but was negative in the combined disk method. The difference in inhibition zones of ceftazidime disks with and without CLA was 3 mm (24 versus 21 mm).

A similar phenotypic profile was observed for one bla_SHV-1 isolate and one bla_SHV-11 isolate expressing phenotypes analogous to those of the two SHV-1-hyperproducing E. coli strains described above. They were consequently regarded as SHV-1 and SHV-11 hyperproducers. In summary, the overall prevalence of ESBL-positive strains in the K. pneumoniae collection was 19/25 (76%). The lack of chromosomally encoded AmpC β-lactamases in the genus Klebsiella may explain the relatively higher occurrence of ESBL production in K. pneumoniae than in E. coli isolates (50/87; 58%) with reduced susceptibilities to oxyimino-cephalosporins (18, 26).

All 25 K. pneumoniae isolates with reduced susceptibilities to oxyimino-cephalosporins were examined by the DDS test. CLA synergy was observed in 21 strains that were identical to the ESBL Etest-positive strains. The putative SHV-1- and SHV-11-hyperproducing strains also expressed CLA synergy in the DDS test.

Performance of ESBL tests.The performance of ESBL Etests and the combined disk method in the detection of ESBL production in E. coli and K. pneumoniae is summarized in Tables 3 and 4. Significant CLA synergy was easily detected by all substrates in both methods for the most prevalent ESBL genogroups: the bla_CTX-M-15/28 genotypes and bla_SHV-5/12 genotypes in E. coli and K. pneumoniae, respectively. However, discordant ESBL test results were observed in one major and several minor ESBL genotypes.

bla_CTX-M-9 genogroup E. coli strains, except for a single CTX-M-16 isolate, scored negative in the ceftazidime combined disk method (14/15; 93%), in contrast to the results of the CAZ-CLA ESBL Etest, by which most of these strains (14/15; 93%) scored positive. However, a CAZ/CAZ-CLA MIC ratio of ≥8 was observed only for the CTX-M-16 strain, whereas the other 13 CTX-M-9-positive strains scored positive by the CAZ-CLA ESBL Etest by the appearance of deformity in the ellipse only (the “eagle effect”). The single bla_CTX-M-9 genogroup K. pneumoniae strain scored positive for ESBL production by both methods using all three oxyimino-cephalosporin substrates.

The minor ESBL types with aberrant ESBL test results included six E. coli strains within the bla_CTX-M-1 (n = 4; bla_CTX-M-1 and bla_CTX-M-3/22) and bla_TEM-128 (n = 2) genogroups that were negative by the ceftazidime combined disk test (Table 4). In contrast, all these strains were positive by the CAZ-CLA ESBL Etest (Table 4). The bla_CTX-M-1 and bla_TEM-128 strains were positive by deformity in the ellipse only, whereas the two bla_CTX-M-3/22 strains had the marginally positive CAZ/CAZ-CLA MIC ratios 12 (K5-58) and 11 (K8-8), respectively. Moreover, one bla_SHV-2 strain and two bla_SHV-28 K. pneumoniae strains scored negative in one or two of the ESBL Etests (Table 3). Corresponding results were obtained by the combined disk method, except with one ESBL bla_SHV-28 K. pneumoniae strain (K2-79) that scored negative for all substrates (Table 4).

MIC profiles for ESBL-producing strains.The MIC means and ranges for oxyimino-cephalosporins and aztreonam within the different ESBL genogroups are presented in Table 5. CTX-M and SHV ESBL-producing strains generally expressed cefotaximase and ceftazidimase profiles, respectively. The MICs of all substrates for the strains with the most-prevalent ESBL genotypes, bla_CTX-M-15/28 in E. coli and bla_SHV-5/12 in K. pneumoniae, were high to moderate.

bla_CTX-M-9 genogroup E. coli strains expressed low MICs of ceftazidime (mean, 0.55 mg/liter) and aztreonam (mean, 1.8 mg/liter), as described previously (4, 24), except for the CTX-M-16 isolate. Interestingly, some of the minor bla genotypes showed clinically significant differences in MIC levels for different substrates. E. coli bla_TEM-128 strains (n = 2) had higher mean MICs for cefpodoxime (48 mg/liter) and cefotaxime (9 mg/liter) than for ceftazidime (0.63 mg/liter) and aztreonam (1 mg/liter). Moreover, seven bla_CTX-M-1 and bla_CTX-M-2 genogroup E. coli strains were intermediately susceptible to ceftazidime (MICs, 1 to 8 mg/liter). Interestingly, the two bla_SHV-28-positive K. pneumoniae strains expressed comparatively low MICs for cefotaxime (0.19 and 1 mg/liter) and cefpodoxime (0.75 and 12 mg/liter). The K2-79 strain, with a cefpodoxime MIC of 0.75 mg/liter, scored negative in the combined disk method.

Coresistance.The majority of ESBL-positive E. coli strains (36/50; 70%) expressed resistance to two or more non-β-lactam antibiotics (aminoglycosides [AG], fluoroquinolones [FQ], nitrofurantoin [NIT], and/or trimethoprim-sulfamethoxazole [SXT]) and were defined as multidrug resistant (MDR). Sixteen out of 23 (70%) bla_CTX-M-15/28 E. coli strains were resistant to three or more non-β-lactam antibiotics, in contrast to 3/15 (20%) of the bla_CTX-M-9 strains. Twenty-eight (62%), 30 (67%), 18 (40%), and 33 (73%) out of 45 CTX-M-positive strains were resistant to AG, FQ, NIT, and SXT, respectively. Only one of the bla_TEM and bla_SHV ESBL-positive E. coli strains (n = 5) was resistant to more than two non-β-lactam antibiotics, and those strains were all susceptible to FQ. Twelve (63%) of the ESBL-producing K. pneumoniae strains expressed MDR. Nine (45%), 3 (15%), 18 (90%), and 4 (20%) K. pneumoniae strains were resistant to AG, FQ, NIT, and SXT, respectively.

Epidemiological data.ESBL-producing strains were detected in 16 laboratories. There were no indications of nosocomial outbreaks during the study period. Urinary tract isolates (n = 42), specifically, 34 (68%) E. coli isolates and 8 (42%) K. pneumoniae isolates, were dominant. ESBL-positive blood culture isolates were not detected. Twenty-two (44%) of the E. coli isolates were from outpatients; however, hospital contact cannot be ruled out as the means of ESBL-positive strain acquisition. We did not detect any specific bla_CTX-M genogroups in hospitalized or nonhospitalized patients, and MDR phenotypes were detected in both groups. Only two (10%) K. pneumoniae isolates were from outpatients.