Presence of the KPC Carbapenemase Gene in Enterobacteriaceae Causing Bacteremia and Its Correlation with In Vitro Carbapenem Susceptibility

The serine carbapenemase KPC (Klebsiella pneumoniae carbapenemase) has emerged as a beta-lactamase capable of inactivating carbapenem antibiotics. First identified in Klebsiella pneumoniae (21), KPC has since been detected in other Enterobacteriaceae (7). The gene encoding KPC, bla_KPC, is plasmid transmissible among Enterobacteriaceae, which has implications for infection control (3, 20). The presence of bla_KPC may not always result in carbapenem resistance in vitro (19), thereby impeding detection during routine workup. KPC-producing bacteria have primarily been reported from the New York City area; however, bla_KPC is present among Enterobacteriaceae isolates as far west as Arkansas (7). The aim of this study was to systematically screen Enterobacteriaceae bacteremia isolates for reduced susceptibility to carbapenems and to correlate results with the presence of bla_KPC.

(This work was presented in abstract form at the 47th Interscience Conference on Antimicrobial Agents and Chemotherapy, Chicago, IL, September 2007 [12a].)

Microbiological and molecular analyses were performed on bacterial isolates from inpatients with Enterobacteriaceae bacteremia at Barnes-Jewish Hospital in St. Louis, MO. We included patients with bacteremia occurring between 1 August 2006 and 31 January 2007. Isolates were tested for susceptibility to the three carbapenem antibiotics (ertapenem, imipenem, and meropenem) and noncarbapenem antibiotics, using the disk diffusion method (6) (Sensi-Disc antibiotic disks; Becton Dickinson and Co., Sparks, MD).

Total DNA was extracted using the QIAamp DNA minikit (Qiagen, Valencia, CA). A real-time PCR assay of all available isolates (n = 243) was developed for initial screening for the presence of bla_KPC using primers and cycle parameters as described previously (17). All isolates that were positive for the bla_KPC gene by real-time PCR were confirmed with a conventional PCR assay as described previously (5). The three positive isolates were further characterized by DNA sequencing of the bla_KPC PCR product using primers (forward, 5′-ATGTCACTGTATCGCCGTC-3′; reverse, 5′-CTCAGTGCTCTACAGAAAACC-3′) and thermocycling parameters described by Yigit et al. (21), with a BigDye Terminator cycle sequencing kit, v3.1 (Applied Biosystems Inc., Foster City, CA) in an MJ Research PTC-200 DNA Engine thermal cycler (Bio-Rad Laboratories, Waltham, MA). Sequencing reaction mixtures were purified by ethanol precipitation, separated, and analyzed using an ABI Prism 3100 genetic analyzer (ABI, Foster City, CA) following the manufacturer's protocols. Forward and reverse strands of two independent PCR products from each isolate were sequenced. Sequences were aligned and compared to published sequences for the bla_KPC-2 gene using Vector NTI v10.3.0 software (Invitrogen, Carlsbad, CA) and found to be identical to the bla_KPC-2 published sequence.

Patient data on demographics, comorbidities, treatment, and in-hospital mortality were abstracted from medical records. The Washington University Human Research Protection Office approved this study.

During the study period, 247 Enterobacteriaceae isolates were recovered from blood cultures at Barnes-Jewish Hospital. Four isolates were unavailable for testing, leaving 243 Enterobacteriaceae isolates from 223 patients. Ninety isolates (37.0%) were Escherichia coli, 79 (32.5%) were Klebsiella pneumoniae, 25 (10.3%) were Enterobacter spp., 13 (5.3%) were Proteus mirabilis, 11 (4.5%) were Klebsiella oxytoca, 7 (2.9%) were Citrobacter spp., 6 (2.5%) were Serratia marcescens, and 12 (4.9%) were other species. Seven (2.9%) isolates had reduced susceptibility to one or more carbapenems (Table 1). Two isolates were resistant to all carbapenems tested; both were bla_KPC positive. Three isolates were nonsusceptible only to ertapenem; none of these were bla_KPC positive.

Three (1.2%) isolates carried the bla_KPC gene. These isolates infected three patients (Table 2) and included one K. pneumoniae, one Enterobacter cloacae, and one P. mirabilis isolate. The in vitro ertapenem nonsusceptibility assay detected bla_KPC with high sensitivity (100% [three/three]) and high specificity (98.3% [236/240]), similar to results for imipenem (100% [three/three] and 100% [240/240], respectively) and meropenem (66.6% [two/three] and 99.6% [239/240], respectively) (Table 1). The positive predictive value (PPV) of ertapenem nonsusceptibility for detecting bla_KPC was 43% (three/seven) versus 100% (three/three) for imipenem and 66.6% (two/three) for meropenem. The PPV of ertapenem as sole carbapenem showing resistance was 0% (zero/three); the PPV of resistance to all three carbapenems for detecting bla_KPC was 100% (two/two). One (33%) of the patients infected with a bla_KPC⁺ isolate and 41 (18.6%) of the patients infected with a bla_KPC⁻ isolate died.

KPC-positive bacteria were present in 1.3% (3/223) of bacteremia episodes in our study, which is relatively low. However, plasmid transfer and subsequent dissemination can occur (3, 21). In a study by Landman et al., the susceptibility of K. pneumoniae isolates to carbapenems decreased from 97% to 76% within 5 years, probably due to bla_KPC (11). In a United States-wide surveillance study, the prevalence of bla_KPC among various Enterobacteriaceae was 0.5% (7), whereas a study of Brooklyn hospitals reported 38% prevalence in K. pneumoniae (11). Our data confirm that bla_KPC is not restricted to the northeastern United States and warrant surveillance of carbapenem susceptibilities among Enterobacteriaceae.

Ertapenem has been proposed as the carbapenem that most accurately detects the presence of bla_KPC by disk diffusion (1, 4, 12). This may be because diameter cutoffs for inhibition zones were set more stringently for ertapenem than for other carbapenems (6). Ertapenem was the most frequently nonsusceptible carbapenem in our study; however, the PPV of ertapenem nonsusceptibility for identifying bla_KPC was low (43%). This is possibly due to carbapenem resistance mediated by mechanisms other than bla_KPC (16). Other studies (15, 19) have found that carbapenem susceptibility testing by the disk diffusion method is unreliable at predicting the presence of bla_KPC. Possible explanations for undetected bla_KPC carriage are an unexpressed bla_KPC gene, the inoculum effect (4), and misinterpretation of the resistance pattern to signify an extended-spectrum beta-lactamase producer (15). An MIC that is in the upper range of susceptibility may be the only indication of bla_KPC. Lowering the imipenem MIC breakpoints (13) or PCR-based screening (2, 9) might increase the chance of detecting resistance.

A limitation of our study is that we did not assess isolates for additional beta-lactamases other than bla_KPC, which is a constellation increasingly encountered (12, 14). We also had a relatively small sample size and a single-center design, and we restricted analysis to bacteremia isolates. We did not test gram-negative bacteria outside the Enterobacteriaceae family for bla_KPC (18). In conclusion, our study is among the first prospective investigations into the endemic epidemiology of bla_KPC-positive bacteria, demonstrating that bla_KPC is currently present at a low level in a major Midwestern city. Disk diffusion tests currently remain the simplest screening tests to detect bla_KPC-positive bacteria in clinical microbiology laboratories.