To achieve a diversity of β-lactam resistance phenotypes, we selected 15 isolates of K. pneumoniae from 4 hospitals in New York City, all serviced by a central microbiology laboratory, on the basis of varying susceptibility patterns to imipenem, meropenem, and extended-spectrum cephalosporins. The isolates were from a variety of body sites (blood, sputum, and urine) and obtained in a 1-month period in 2005. The carbapenem-resistant quality control isolates K. pneumoniae 1534 (containing the KPC-1 carbapenemase) and Serratia marcescens 525 (which contains an SME-like β-lactamase) were from the Project ICARE (Intensive Care Antimicrobial Resistance Epidemiology) strain collection (1,11). The imipenem MICs for the isolates and the study identification of their hospital of origin are shown in Table 1.

The antimicrobial susceptibility profiles of the isolates were determined by the broth microdilution method with cation-adjusted Mueller-Hinton broth (BD Diagnostic Systems [BDDS], Sparks, MD, USA), as described in the National Committee for Clinical Laboratory Standards (NCCLS, Wayne, PA, USA) (now known as the Clinical and Laboratory Standards Institute [CLSI]) publication M7-A6 (16). Disk diffusion was performed as described in NCCLS document M2-A8 (17). Interpretations of MIC and disk diffusion results were made by using CLSI document M100-S15 (18).

The Etest method (AB Biodisk, Solna, Sweden) was performed as described by the manufacturer with Mueller-Hinton agar (BDDS); tests were interpreted at 18–20 h. The MicroScan WalkAway (Dade MicroScan, Inc.), BD Phoenix (BDDS), Sensititre AutoReader (Westlake, OH, USA), VITEK Legacy (bioMérieux), and VITEK 2 (bioMérieux) systems were tested according to manufacturers' protocols. The panels and cards used are listed in Table 2. All systems were tested with inocula from the same subculture. All strains for which the commercial MIC results were discrepant with MIC results from the broth microdilution reference method were retested by using all methods.

The carbapenem inactivation assay was performed on Mueller-Hinton agar with imipenem and meropenem disks as described by Yigit et al. (1). Isoelectric focusing was performed on crude lysates of isolates as previously described (1,2).

For detection of bla_KPC genes, a 489-bp internal gene fragment was amplified by using forward (5´-CTTGCTGCCGCTGTGCTG-3´) and reverse (5´-GCAGGTTCCGGTTTTGTCTC-3´) oligonucleotide primers where the 5´ base of each primer corresponds to position 223 or 711, respectively, with regard to the translational start site (bla_KPC-1 numbering, GenBank accession no. AF297554). PCR reagents included a final concentration of 0.5 μmol/L of each primer and 2 mmol/L MgCl₂. An annealing temperature of 60°C was used for amplification.

For DNA sequence determination, a 989-bp PCR product that included the entire bla_KPC structural gene was amplified by using oligonucleotide primers as previously described (4). Products were purified on QIAquick spin columns (Qiagen, Chatsworth, CA, USA). The nucleotide sequences of both strands of the bla_KPC gene from isolates 4 and 11 were determined from independent amplification products by using previously described primers (1). Cycle sequencing reactions were performed in a GeneAmp PCR system 9700 thermal cycler with the ABI BigDye Terminator v3.1 cycle sequencing kit (Perkin-Elmer, Applied Biosystems Division, Foster City, CA, USA). Products from sequencing reactions were purified on Cetri-Sep spin columns (Princeton Separations, Adelphia, NJ, USA) before analysis on an ABI 3130xl Genetic Analyzer. DNA sequencing data were analyzed by using DNASIS for Windows (Hitachi Software Genetic Systems, San Francisco, CA, USA).

Initial imipenem and meropenem susceptibility test results for 15 isolates of K. pneumoniae tested with the VITEK Legacy system with GNS 122 and 127 panels (flex system) in 1 hospital laboratory in New York City in a 1-month period in 2005 yielded a range of imipenem and meropenem MICs from susceptible (MIC <4 μg/mL) to resistant (MIC >16 μg/mL) (data not shown). One isolate was imipenem resistant (MIC >16 μg/mL) but meropenem susceptible (<4 μg/mL) on day 1, but on retesting the following day was meropenem resistant (16 μg/mL) and imipenem susceptible (<4 μg/mL). Imipenem Etest results, which were set up to arbitrate the conflicting results, were difficult to interpret because of variable numbers of colonies within the ellipses of inhibition (Figure). A carbapenem inactivation assay performed on 1 isolate indicated that it contained a carbapenem-inactivating enzyme (data not shown). The 15 isolates, which were collected from 4 different hospitals in New York City, were sent to the Centers for Disease Control and Prevention (Atlanta, GA, USA) for additional testing.

By using the broth microdilution reference method, all the isolates were either intermediate or resistant to both imipenem and meropenem (Table 2). Two known imipenem-resistant isolates, K. pneumoniae 1534 and S. marcescens 525, were included as controls. Eight of 15 isolates from the 4 hospitals had the same pulsed-field gel electrophoresis (PFGE) type (data not shown), although their antibiograms varied for several antimicrobial agents (data not shown). Two isolates had a similar pattern but because of a 3-band difference were designated as type B. The remaining isolates showed patterns unrelated to types A or B (Table 1). All 15 isolates demonstrated 5 β-lactamase bands by isoelectric focusing (pIs = 5.4, 6.8, 7.0, 8.1, and 8.2), 1 of which was consistent with a KPC β-lactamase (pI = 6.8) (1,2). A 489-bp gene fragment was amplified from all 15 carbapenem-resistant K. pneumoniae isolates by using bla_KPC-specific oligonucleotide primers. DNA sequence analysis of purified PCR products that included the entire coding region of the bla_KPC genes of isolates 4 and 11 (which had unique PFGE profiles) identified the β-lactamase genes as bla_KPC-2 and bla_KPC-3, respectively.

The 15 isolates were tested for imipenem and meropenem resistance by disk diffusion, Etest, MicroScan WalkAway, BD Phoenix, Sensititre AutoReader, VITEK, and VITEK 2 panels and cards. The results of testing are summarized in Table 2. MicroScan WalkAway reported 1 isolate as susceptible to both imipenem and meropenem, whereas the Phoenix system called 2 isolates susceptible to both imipenem and meropenem. VITEK called 7 isolates (representing 3 different PFGE profiles) susceptible to both imipenem and meropenem, 3 isolates resistant to imipenem but susceptible to meropenem, 1 isolate susceptible to imipenem but resistant to meropenem, and 2 isolates susceptible to imipenem but intermediate to meropenem. Of the final 2 isolates, 1 was resistant to imipenem and intermediate to meropenem, and the other was resistant to both antimicrobial agents. Thus, 10 (67%) of 15 isolates were interpreted on initial testing as susceptible to imipenem, and 10 were susceptible to meropenem. When the VITEK 2 system was used, 5 (33%) of 15 isolates were reported as susceptible to imipenem. The VITEK 2 Advanced Expert System (AES) used the imipenem results to predict meropenem results; thus, these same 5 isolates were called meropenem susceptible. In addition, as a result of the AES's recognizing unusual susceptibility results in the antibiograms of 2 K. pneumoniae isolates, AES did not report an interpretation for meropenem for 2 isolates. Finally, the VITEK 2 reported 2 isolates as imipenem resistant and 1 isolate as imipenem intermediate, although the MICs reported by AES were listed as <2 μg/mL. Aside from these, all AES categorical interpretations were in agreement with the original VITEK 2 MIC results.

When Sensititre panels were used, 10 (67%) of 15 isolates were reported as susceptible to both imipenem and meropenem, 2 isolates were reported as imipenem intermediate and meropenem susceptible, and 3 isolates were reported as imipenem susceptible and meropenem intermediate. Thus, on initial testing, 13 (87%) of 15 isolates were reported as imipenem susceptible, and 12 (80%) of 15 were meropenem susceptible. Repeat testing of strain 4 yielded imipenem- and meropenem-resistant results for MicroScan, BD Phoenix, VITEK Legacy, and VITEK 2. However, the Sensititre AutoReader results showed the isolate as susceptible to imipenem and meropenem. Retesting of strain 14 on the BD Phoenix showed the isolate as resistant to imipenem and meropenem. However, with VITEK Legacy, the isolate remained susceptible to imipenem, but the response to meropenem switched from 8 μg/mL (intermediate) to <4 μg/mL (susceptible), which confirmed the observations of flip-flopping (i.e., reversing) results from the New York City laboratory. The Sensititre AutoReader results for strain 14 remained susceptible on repeat testing; MicroScan results remained resistant.