Nonduplicate K. pneumoniae clinical isolates were recovered from patient specimens during January 2005–July 2012 at the clinical microbiology laboratory of Westchester Medical Center. These included 208 bla_KPC-negative non-K. pneumoniae carbapenemase (non-KPC) ESBL-producing or third-generation cephalosporin-resistant K. pneumoniae isolates and 228 KPC (bla_KPC-positive)–producing K. pneumoniae isolates. In addition, 163 nonduplicate ESBL-producing E. coli clinical isolates from the same period were also analyzed for comparison. Isolates were randomly selected to span the entire study year with an approximately equal number of isolates per quarter; only 1 isolate from each patient was chosen and tested. The center is a 643-bed academic tertiary-care medical center in Westchester County, New York. The Institutional Review Board of New York Medical College approved this study.

The bacterial isolates were identified and evaluated for antimicrobial drug susceptibility with the MicroScan WalkAway 96 system (Siemens, Sacramento, CA, USA). ESBL production was phenotypically confirmed by a double-disk or broth microdilution method for suspected ESBL isolates according to the Clinical and Laboratory Standards Institute (CLSI) guidelines (19). The antimicrobial drug susceptibility of CTX-M–producing K. pneumoniae isolates against selected antimicrobial drugs was also assessed with standardized CLSI disk diffusion and Etest methods. Bacterial isolates were refrigerated on nutrient agar slants or were frozen (−80°C) in MicroBank cryovials containing 20% glycerol (Pro-Lab Diagnostics, Round Rock, TX, USA). For antimicrobial drug susceptibility testing of frozen isolates, fresh subcultures were used per CLSI guidelines.

For PCR, bacterial genomic DNA was extracted directly from colonies on nutrient slants or from fresh subcultures grown on Trypticase soy agar with 5% sheep blood (TSA II, BBL, Sparks, MD, USA) by boiling a dense suspension of an approximately no. 1 McFarland standard in sterile distilled water. As the DNA template in the PCR assays, 2–3 μL of the boiled cell suspension was used. PCR amplification of bla_CTX-M, bla_SHV, bla_TEM, and bla_KPC genes in K. pneumoniae and E. coli clinical isolates was performed by using a consensus primer pair specific to each type of β-lactamase as described (20–22). A multiplex PCR was developed and used for simultaneous detection of bla_CTX-M (551 bp) and bla_TEM (972 bp) genes. Two PCRs were performed for bla_SHV-ESBL and bla_KPC, respectively. PCRs were carried out by using the HotStart DNA polymerase master mix (QIAGEN, Germantown, MD, USA) with 30–35 cycles at an annealing temperature of 50°C for bla_CTX-M and bla_TEM, and 52°C for bla_SHV and bla_KPC. PCR products were analyzed by agarose gel electrophoresis or by using the QIAxcel system (QIAGEN). The specificity of PCR amplicons on representative isolates was confirmed by DNA sequencing.

For DNA sequencing, PCR products were purified by using the PCR Purification kit (QIAGEN) or the ExoSAP-IT PCR Clean-up kit (Affymetrix, Cleveland, OH, USA), according to the manufacturer’s instructions. The purified DNA amplicons were sequenced by using an ABI Prism BigDye Terminator (version 1.1) cycle sequencing ready reaction kit on the ABI Prism 3730xl or ABI 3500xl DNA Analyzers (Applied Biosystems, Foster City, CA, USA) in-house, or by a commercial facility (GeneWiz, South Plainfield, NJ, USA). The CTX-M, TEM, and SHV gene sequences were compared with sequences in GenBank by using the NCBI basic local alignment search tool (www.ncbi.nlm.nih.gov/BLAST).

Multilocus sequence typing (MLST) was performed by using primers and conditions as described by Diancourt et al. (23). PCR products from MLST were sequenced as described above. Allelic profiling and sequence types (STs) were determined by querying the K. pneumoniae MLST database maintained by the Pasteur Institute (www.pasteur.fr/recherche/genopole/PF8/mlst/Kpneumoniae.html).

Pulsed-field gel electrophoresis (PFGE) on CTX-M ESBL–producing K. pneumoniae isolates representing each CTX-M genotype was performed as described (24). The GelCompare II software, (version 2.0; Applied Maths, Austin, TX, USA) was used to calculate the Dice similarity coefficients and generate dendrograms by cluster analysis with the unweighted-pair group method using average linkages. Pulsotype designations were assigned at the ≥80% profile similarity level.

Of the 121 ESBL-producing K. pneumoniae isolates originally recovered during 2005–2009, bla_SHV and bla_TEM genes were detected in 102 (84.3%) and 61 (50.4%) of 121 isolates_, respectively (Table 1). Overall, 25 CTX-M-type ESBL K. pneumoniae were identified. However, none of the 81 K. pneumoniae isolates from 2005 through 2008 was positive for bla_CTX-M genes. CTX-M–type ESBL was first detected in 2 (5.0%) of 40 K. pneumoniae isolates from 2009. The prevalence of K. pneumoniae isolates carrying the CTX-M–encoding genes increased to 6 (17.6%) of 34 in 2010 and 12 (34.3%) of 35 in 2011. The level remained high (27.8%, 5/18) in the first 7 months of 2012. Overall, only 2 (1.7%) of 121 ESBL-producing K. pneumoniae isolates from 2005 through 2009 carried the bla_CTX-M genes, compared with 23 (26.4%) of 87 isolates from 2010 through 2012 (p<0.0001, Fisher exact test), indicating the rapid emergence and spread of CTX-M enzymes among ESBL-producing K. pneumoniae clinical isolates since 2009.

One hundred sixty-three ESBL-producing E. coli clinical isolates from 2005 through 2012 were analyzed by PCR for detection of bla_ESBL genes of the SHV, TEM, and CTX-M types (Table 2). Unlike the situation with K. pneumoniae, bla_CTX-M genes were detected in ESBL-producing E. coli isolated as early as 2005. Overall, 89 (54.6%) of 163 ESBL E. coli isolates from the 8-year period carried bla_CTX-M genes. CTX-M was the leading ESBL type in all years examined except 2008. The bla_CTX-M genes from 47 (52.8%) of 89 CTX-M–producing E. coli isolates were sequenced. CTX-M-15 was determined in 45 (95.7%) of 47 CTX-M–producing E. coli isolates analyzed. CTX-M-1 and CTX-M-3 genotypes were each found in 1 ESBL E. coli isolate.

Two hundred twenty-eight KPC-producing K. pneumoniae isolates from 2005 to 2012 were examined by PCR for detection of bla_CTX-M genes. All K. pneumoniae isolates were positive for the bla_KPC gene by PCR as described (22). None was positive for the bla_CTX-M gene.

Selected clinical/epidemiologic features of the 25 patients with CTX-M ESBL–producing K. pneumonia and certain microbiological characteristics of the isolates are shown in Table 3, Appendix. Mean patient age was 56 years, and 13 (52%) of the patients were male. Sixteen patients (64%) had bacteriuria. CTX-M–producing K. pneumoniae isolates were recovered from 13 (52%) patients within 72 hours of hospital admission; however, 18 (72%) of these patients had been hospitalized in the 8 months before the current admission.

The bla_CTX-M genes from all 25 CTX-M ESBL–producing K. pneumoniae isolates from 2009 through 2012 were sequenced. CTX-M-15 was identified in 19 (76.0%) and was the dominant CTX-M genotype. The remaining 6 isolates were determined to be CTX-M-3 (n = 4), CTX-M-1 (n = 1), and CTX-M-2 (n = 1), respectively. Twenty-four (96.0%) had coexisting bla_SHV β-lactamases, which were predominantly non-ESBL bla_SHV-11 (n = 15) and bla_SHV-1 (n = 5)_. Four additional K. pneumoniae carried ESBL-type bla_SHV β-lactamases, including bla_SHV-12 (n = 1), bla_SHV-27 (n = 1), and bla_SHV-28 (n = 2). Seventeen (68.0%) were positive for TEM-type β-lactamases, and all were confirmed to be bla_TEM-1.

The antimicrobial drug susceptibilities of CTX-M–producing K. pneumoniae isolates are summarized in Table 4. Of the 25 CTX-M–producing K. pneumoniae isolates examined in this study, only 12% (n = 3) and 36% (n = 8) of isolates were susceptible to ciprofloxacin and gentamicin, respectively. Low susceptibility rates were also observed for pipercillin/tazobactam (36%), tetracycline (20%) and trimethoprim/sulfamethoxazole (4%). Twenty-three of the 25 (92%) isolates tested were susceptible to carbapenems. Notably, the 2 carbapenem-resistant K. pneumoniae isolates (PK30 and PK107) carried bla_CTX-M-3, bla_SHV-11, and bla_TEM-1. One of these K. pneumoniae isolates also showed resistance to colistin with an MIC of 64µg/mL. Both patients died of complications associated with bloodstream and respiratory tract infections. Three of 22 CTX-M–producing K. pneumoniae isolates examined by Etest were nonsusceptible to tigecycline (MICs 3 µg/mL, 3 µg/mL, and 8 µg/mL).

All 25 CTX-M–producing K. pneumoniae isolates examined were resistant to cefotaxime, and all but 1 isolate showed higher MICs of cefotaxime than of ceftazidime. The 50% minimum inhibitory concentration (MIC₅₀) for cefotaxime among these isolates was >256 µg/mL. By contrast, the MIC₅₀ and 90% inhibitory concentration for ceftazidime were 16 µg/mL and 128 µg/mL, respectively. Two CTX-M–producing K. pneumoniae isolates (8.0%) were susceptible (MIC ≤4 µg/mL) and 5 isolates (20%) were intermediate in susceptibility (8 µg/mL) to ceftazidime according to the 2010 revised CLSI breakpoints (Figure 1). In addition, we determined the susceptibilities of 22 CTX-M–producing K. pneumoniae isolates against cefotaxime and ceftazidime by using the standard disk diffusion method. All CTX-M–producing K. pneumoniae isolates examined were resistant to cefotaxime by disk diffusion (mean inhibitory zone size 8.3 mm; range 6–13 mm). Two of these isolates were susceptible (≥21 mm) and 5 had intermediate (18–20 mm) susceptibility to ceftazidime by disk diffusion (Table 3, appendix). The disk diffusion results showed a category agreement with the Etest MIC of 100% for cefotaxime and 90.9% for ceftazidime with 2 minor errors.

Of 17 representative CTX-M-producing K. pneumoniae isolates analyzed by PFGE, 8 different pulsotypes (PF1–8) were identified with Dice coefficients of ≥80% similarity (Figure 2). Ten of 17 K. pneumoniae isolates examined belonged to 3 major groups (PF3, PF4, PF5) with 3–4 isolates in each group. The remaining pulsotypes contained only 1 or 2 K. pneumoniae isolates. No clear temporal relationship was shown among the highly related isolates.

MLST was performed on 18 CTX-M–producing K. pneumoniae isolates. These isolates were selected to represent different CTX-M genotypes, pulsotypes, antimicrobial susceptibility profiles, and years of isolation. Twelve STs were recognized for the K. pneumoniae isolates examined (Table 3, appendix). Notably, all 3 CTX-M group 1, non–CTX-M-15 K. pneumoniae isolates analyzed (KP38, PK107, and PK135) had ST11, whereas 10 different STs (ST15, ST16, ST17, ST48, ST147, ST252, ST258, ST280, ST392, and ST437) were identified for the 14 CTX-M-15 K. pneumoniae isolates. Isolate F351 was the only non–CTX-M-1 group K. pneumoniae isolate identified in this study and was determined to be a separate group (ST792) by MLST. Of the 14 CTX-M–producing K. pneumoniae isolates evaluated simultaneously by DNA sequencing, PFGE and MLST, a high genetic divergence was demonstrated by the detection of 4 CTX-M genotypes (CTX-M-1, CTX-M-2, CTX-M-3, and CTX-M-15), 8 pulsotypes (PF1–8) and 11 STs (ST11, ST15, ST16, ST17, ST48, ST147, ST252, ST280, ST392, ST437, and ST792) (Figure 2).