First Canadian outbreak of Enterobacteriaceae-expressing Klebsiella pneumoniae carbapenemase type 3

1McGill University Health Centre, Montreal; 2Laboratoire de Santé Publique du Québec, Québec, Québec; 3Public Health Agency of Canada, Winnipeg, Manitoba Correspondence: Dr Victor Leung, Royal Victoria Hospital, 687 Pine Avenue West, Room H4-24, Montreal, Quebec H3A 1A1. E-mail victor.leung@mail.mcgill.ca Klebsiella pneumoniae carbapenemase (KPC) belongs to the Ambler class A β-lactamase group and is capable of hydrolyzing all β-lactam antibiotics. KPC is endemic in the eastern United States and is emerging internationally since its characterization in 2001 (1,2). KPC producers may be missed using routine phenotypic tests, which can lead to inappropriate antibiotic prescriptions and delays in infection control interventions. The the blaKPC gene, which is flanked by transposon Tn4401 and located on conjugative plasmids, is most frequently found in Klebsiella pneumoniae and is also transferable among genera in the Enterobacteriaceae family and even Pseudomonas species (3,4). KPCproducing K pneumoniae has been reported in Canada as isolated cases but remains rare (5,6). To date, other Enterobacteriaceae-expressing KPC have not been reported in Canada. In the present study, we describe the first Canadian outbreak of KPC-3-producing Enterobacteriaceae (KPC-Ent) isolates from patients in an intensive care unit (ICU).

The modified Hodge test (MHT) was used to phenotypically determine carbapenemase production (7).A customized algorithm was added to Vitek 2 to alert technologists to perform the MHT if Enterobacteriaceae isolates had imipenem minimum inhibitory concentrations (MICs) ≥2 µg/mL (the AST085 card that incorporates imipenem as the only carbapenem was used) or were resistant to all β-lactams except imipenem.Using the laboratory information system, no additional isolates were found in the six months before this first isolate.Clinical data were recorded using standardized collection forms.To establish an epidemiological link, each patient's length of stay in the ICU was plotted with an indicator marking when the first isolate was identified (Figure 1).The Centers for Disease Control and Prevention (CDC, Atlanta, Georgia, USA) surveillance definitions for classification of infections, were used (8).Species identification and antibiotic susceptibility were performed  Isolates were further characterized by sequencing the bla KPC gene, and pulsed-field gel electrophoresis (PFGE).The bla KPC gene was detected using polymerase chain reaction primers, which were previously described and encompassed the entire coding region resulting in an amplicon of 893 bp (10).The amplified product was purified using a MinElute purification kit (Qiagen Inc, Germany) and underwent bidirectional sequencing (Applied Biosystems, USA).Internal primers were included for sequencing bla KPC (11).BLAST software from the National Center for Biotechnology Information (USA) was used for sequence identity.PFGE was performed after XbaI restriction enzyme digestion (Figure 2).

resuLts
Between December 2009 and July 2010, 10 unique MHT-positive Enterobacteriaceae isolates from nine ICU-hospitalized patients were identified.A summary of the clinical characteristics of the case patients is shown in Table 1.Although not all patients were in the ICU simultaneously, each case patient's ICU hospitalization overlapped with several other case patients (Figure 1).A summary of the isolates, types of infection and antibiotic susceptibilities are shown in Table 2.The complete DNA sequences of the bla KPC gene and deduced amino acid sequences were 99.9% to 100% identical to that of K pneumonia CL-5761 containing the bla KPC-3 allele.Clonal diversity was interpreted using published criteria for PFGE and results showed four E coli, one Klebsiella oxytoca and one Serratia marcescens clonal group(s) (12).

dIsCussIon
To the best of our knowledge, this is the first outbreak of KPC-Ent in Canada.To have multiple genera involved in this outbreak is concerning.Horizontal transmission is suggested by the epidemiological link (Figure 1) and clonal clusters in four pairs of patients (Figure 2).Although the outbreak involved multiple genera, PFGE was the only molecular typing method used.Interestingly, the E coli isolates in lanes 1 and 6 belong to a different pulsovar (Figure 2) and were isolated from the same patient 72 days apart.This strongly suggests in vivo intraspecies transfer of the bla KPC gene.In vivo transfer of the bla KPC gene between K pneumoniae isolates from the same patient has been recently documented (13).There is also evidence for interspecies conjugative transfer of the bla KPC gene (14).We hypothesize that this outbreak has resulted from ongoing horizontal transfer of Enterobacteriaceae with subsequent in vivo interspecies transfer of the bla KPC-3 gene.Plasmid and transposon characterization and multilocus sequence typing were not performed to further characterize genetic relatedness.Previous studies have demonstrated remarkable diversity of the molecular features of KPC genes with plasmids differing in size and incompatibility groups.Future epidemiological investigations of KPC producers should use different molecular epidemiological approaches to characterize transmission with more certainty.The antibiotic susceptibility patterns of the 10 isolates are shown in Figure 2. Vitek 2 susceptibilities were different than the Etest results for imipenem.We would have misclassified eight of the isolates as susceptible using Etest, whereas five isolates would have been misclassified as susceptible using Vitek 2 using the Clinical and Laboratory Standards Institute (CLSI) 2009 breakpoints for Enterobacteriaceae (15).In June 2010, the CLSI revised the carbapenem breakpoints for Enterobacteriaceae (16).The CLSI revisions are based on evaluation of pharmokinetic and pharmacodynamic data, clinical data and MIC distributions of carbapenemase-producing strains.The breakpoints were lowered by two doubling dilutions for meropenem, imipenem and doripenem, and by three doubling dilutions for ertapenem.Although the use of lower breakpoints decreases the chance of falsely misclassifying Enterobacteriaceae isolates, it still has the potential to result in major errors.Using these new breakpoints, the imipenem and meropenem Etest would have misclassified seven of the isolates as susceptible, while Vitek 2 would have misclassified only one isolate as susceptible.However, this observation is based on the Vitek 2 AST085 card, which only has imipenem.
The CLSI 2010 supplemental update for the Performance Standards for Antimicrobial Susceptibility Testing (16) states that if a laboratory uses new interpretive criteria, the MHT does not need to be performed other than for epidemiological or infection control purposes.If we had only relied on the revised breakpoints recommended by the most recent CLSI, we would have missed one or seven cases depending on the method and antibiotics used for susceptibility testing.This would have prevented the early implementation of infection control measures.Based on our limited experience and the general availability of tests in hospital-based microbiology laboratories, we recommend that laboratories in Canada continue to use the MHT on Enterobacteriaceae isolates with elevated breakpoints to carbapenems.If positive by MHT, further testing at a reference laboratory for the underlying resistance mechanism is warranted because the prevalence of KPCs in Canada is believed to be low.
The possibility of major errors in detecting KPCs using automated susceptibility systems including Vitek 2 has previously been reported and is believed to be a result of an inoculum effect and porin changes affecting carbapenem permeability (17).There are little data on the ability of Etest to consistently detect KPC-Ent.Our isolates also contained other β-lactamases (data not shown), which may have contributed to the variable susceptibilty profile.Data suggest that ertapenem MICs may be the most sensitive, but meropenem and imipenem MICs are more specific for KPC detection (18,19).Only one patient in our outbreak received tigecycline for a VRE infection before the KPC-Ent isolation.Interestingly, S marcescens colonizing the urinary tract had an MIC of 3 µg/mL.The lower levels of tigecycyline in the urinary tract may have had a selective pressure effect.None of our patients were treated with colistin before or after identification of KPC-Ent.Patients who were treated for infections continued to receive a combination of a carbapenem and aminoglycoside.The efficacy of this treatment was difficult to discern because there were many other factors contributing to patient status.Overall, there were four deaths and none could be attributed entirely to KPC-Ent infection.However, we did not follow the status of patients who were discharged.Two of the case patients remained hospitalized in the ICU at the end of the study.

ConCLusIons
We described the first Canadian outbreak of KPC-3-producing Enterobacteriaceae isolates from patients in an ICU.After conducting two point prevalence studies in the ICU, using rectal swabs and the recommended CDC method followed by polymerase chain reaction confirmation, two new isolates (E coli and S marcescens) expressing KPC-3 from different patients have been found.The extent of this current outbreak is not fully realized, but further point prevalence studies will be performed.Infection control measures, including isolation of patients, contact precautions and increased hand hygiene education, have been implemented but the impacts of these measures and further active surveillance using rectal swabs will need to be evaluated.The ideal screening methods and optimal treatment for patients with

Figure 2 ) 3 1Patients
Figure 2) Macrorestriction analysis (XbaI) by pulsed-field gel electrophoresis of Enterobacteriaceae isolates.*Salmonella braenderup molecular mass marker; † Isolate was obtained from same patient (lane 3) four days later; § Isolate was obtained from same patient (lane 1) 72 days later; ¶ Isolate was obtained from same patient (lane 10) seven days later.E Escherichia; K Klebsiella; S Serratia

Figure 1 )
Figure 1) Timeline depicting patient length of stay in the intensive care unit.Each horizontal gray bar represents length of stay, with bars on the same y-axis representing the same patient.Some patients were admitted to the intensive care unit more than once during the hospitalization.The black ovals indicate the date the Enterobacteriaceae isolate was obtained from the patient.Patient 3 has two isolates Citrobacter freundii and Klebsiella oxytoca.Patients 1 and 2 had Escherichia coli pulsovar A. Patients 3 and 4 had K oxytoca pulsovar D. Patient 5 had E coli pulsovar B. Patients 6 and 9 had E coli pulsovars D and D1 respectively.Patients 7 and 8 had Serratia marcescens pulsovars A