Prevalence and characterization of extended-spectrum beta-lactamase-producing Enterobacteriaceae isolated in Canadian hospitals: Results from CANWARD 2007

1Department of Medical Microbiology, Faculty of Medicine, University of Manitoba; 2Department of Clinical Microbiology, Health Sciences Centre; 3National Microbiology Laboratory, Public Health Agency of Canada, Winnipeg, Manitoba Correspondence: Patricia J Baudry, Health Sciences Centre, Department of Clinical Microbiology, MS673-820 Sherbrook Street, Winnipeg, Manitoba, R3A 1R9. Telephone 204-787-4684, fax 204-787-4699, e-mail trishbaudry@hotmail.com Extended-spectrum beta-lactamases (ESBLs) are the most significant resistance determinants emerging and spreading worldwide among the Enterobacteriaceae (1). They were first reported in 1983 and since then, greater than 300 variants of ESBLs have been identified (http://www.lahey.org/Studies). The majority of ESBLs are Ambler Class A TEM, SHV and CTX-M variants. Until the late 1990s, TEM and SHV were the predominant ESBL variants and were mainly associated with nosocomial infections caused by Klebsiella pneumoniae (2,3). The epidemiology of ESBLs has undergone a rapid change over the past decade with the emergence and spread of CTX-M ESBLs. CTX-M ESBLs are now the most prevalent genotype and are mainly associated with Escherichia coli infections from both community and nosocomial infections (2,4). ESBLs are enzymes that compromise the efficacy of betalactam antibiotics, with the exception of the cephamycins and carbapenems, by hydrolysis of the beta-lactam ring (3,5). Antimicrobial therapy is frequently limited for the treatment of ESBL producers because they are often multidrug resistant (MDR). CANWARD 2007

The purpose of the present study was to determine the prevalence and molecular epidemiology of ESBL-producing Enterobacteriaceae isolated from Canadian hospitals in 2007.

MAtERIALS AND MEtHODS
Bacterial strains from surveillance study Bacterial isolates were collected as part of the Canadian Ward Surveillance Study (CANWARD 2007). CANWARD is a laboratory-based surveillance study coordinated by the Health Sciences Centre in Winnipeg, Manitoba. From January 1 through December 31, 2007, inclusive, 12 sentinel hospital centres across Canada submitted pathogens from patients attending hospital clinics, emergency rooms, medical and surgical wards, and intensive care units. Each centre was asked to submit pathogens (consecutive, one per patient/infection site) from blood (n=360), respiratory (n=200), urine (n=100), and wound/intravenous (n=50) infections. If the centres did not collect a minimum of 20 ESBL producers within these objectives, they were asked to continue to collect above the objectives until a minimum of 20 was fulfilled. All isolates were identified at participating sites by routine procedures performed at each laboratory. Isolates were shipped to the reference laboratory at the Winnipeg Health Sciences Centre on Amies charcoal swabs, subcultured onto blood agar, and stocked in skim milk at -80°C until minimum inhibitory concentration (MIC) testing was carried out. In the present study, only the analysis of unique ESBLproducing Enterobacteriaceae isolates that fell within the CANWARD 2007 objectives will be reported.

Antimicrobial susceptibility testing
Following two subcultures from frozen stock, the in vitro activities of various antimicrobials were determined in triplicate by microbroth dilution in accordance with Clinical and Laboratory Standards Institute (CLSI) guidelines (6). Colistin (polymyxin E) MICs were determined using the E-test method. Food and Drug Administration (USA) interpretation breakpoints were used for tigecycline (susceptible: 2 µg/mL or less, intermediate: 4 µg/mL and resistant: 8 µg/mL or greater) and colistin (susceptible: 2 µg/mL or less and resistant: 4 µg/mL or greater). Strains concomitantly resistant to three or more different antimicrobial classes were defined as MDR.
Any E coli, K pneumoniae, Klebsiella oxytoca or Proteus mirabilis with a ceftriaxone MIC of 1 µg/mL or greater was identified as a putative ESBL and underwent further analysis. The putative ESBL phenotype was confirmed by the disk diffusion method as described by CLSI. E coli ATCC 25922 and K pneumoniae ATCC 700603 were the control strains used in the study.

Characterization of ESBL genes
Genotypic characterization of ESBLs was performed by polymerase chain reaction and sequencing of bla SHV , bla TEM , bla CTX-M and bla OXA-like genes as previously described (7)(8)(9). A Basic Local Alignment Search Tool search of the DNA sequence was conducted to determine specific genotypes.

Genetic relationships
Genetic relationships of the ESBL-producing Enterobacteriaceae were assessed by pulsed-field gel electrophoresis following digestion with XbaI as previously described (7).
Statistical analysis χ 2 analysis was used to evaluate statistical significance, as appropriate, using Graphpad Quickcalcs (Graphpad Software Inc).

Epidemiology of ESBL-producing Enterobacteriaceae
A total of 7881 clinical isolates were collected as part of the CANWARD 2007 surveillance study. E coli, K pneumoniae, K oxytoca and P mirabilis ranked first (n=1702, 21.6%), fifth (n=457, 5.8%), 11th (n=119, 1.5%) and 14th (n=100, 1.3%), respectively, among all pathogens collected. Of those that fell within the CANWARD objectives, 93 of 1560 (6.0%) E coli, 12 of 442 (2.7%) K pneumoniae, seven of 119 (5.9%) K oxytoca and three of 111 (2.7%) P mirabilis had a ceftriaxone MIC of 1 µg/mL or greater and were identified as putative ESBL producers. Of those with a ceftriaxone MIC of 1 µg/mL or greater, 53 E coli and seven K pneumoniae were identified phenotypically as unique ESBL producers and were further analyzed. The prevalence of ESBL-producing E coli was 3.4% (53 of 1560) and ranged from 1.1% in emergency rooms, 1.9% in intensive care units, 3.3% in hospital clinics, 6.2% in medical wards to

*Comparison of resistant (R) with nonresistant (susceptible [S] and intermediate [I]) isolates among ESBL-producing and non-ESBL-producing isolates. † Tested against fewer isolates of non-ESBL-producing isolates (E coli n=504; K pneumoniae n=185). Amox/Clav Amoxicillin/clavulanate; Colistin (polymyxin E); MIC 50/90 Minimum inhibitory concentration (in µg/mL) required to inhibit the growth of 50%/90% of organisms; N/A Not available; NS Not significant, P>0.05; TMP/SMX Trimethoprim/sulfamethoxazole; TZP Piperacillin-tazobactam
53 (90.6%) and five of seven (71.4%) ESBL-producing E coli and K pneumoniae were identified as MDR. All ESBL producers remained susceptible to the carbapenems. A comparison of antimicrobial susceptibilities between ESBL-producing and non-ESBL-producing isolates to various antimicrobials are summarized in Table 2. Not surprisingly, the ESBL producers displayed higher resistance rates to all antimicrobials tested with the exception of the carbapenems and tigecycline in comparison to non-ESBL producers.

Distribution of beta-lactamase genes
All isolates phenotypically identified as ESBL producers by the CLSI disk diffusion assay were subjected to PCR and sequencing to detect ESBL genes. All bla TEM , bla SHV , bla CTX-M and bla OXA-1, -2, -10 groups were detected in 49%, 1.9%, 96.2% and 47.2% of ESBL-producing E coli, respectively, and 14.3%, 100%, 57.1% and 42.9% of ESBL-producing K pneumoniae, respectively (Table 3) The distribution of beta-lactamase genes among ESBLproducing E coli and K pneumoniae are summarized in Table 3.

Genetic relationships among ESBL-producing E coli and K pneumoniae
Molecular typing using pulsed-field gel electrophoresis was conducted to study clonal relationships among ESBL-producing E coli and K pneumoniae. A dendrogram depicting the genetic relationships among ESBL-producing E coli and K pneumoniae isolates from Canadian hospitals is shown in Figure 1. The majority of ESBL-producing E coli were determined to be genetically unrelated (less than 80% homology). Several small clusters (two to five isolates) of genetically related ESBLproducing E coli carrying the same type of CTX-M were observed. However, there were two clusters of two genetically related E coli from different regions of Canada with CTX-M genes from different CTX-M groups (CTX-M-1 and CTX-M-9 groups) suggesting two different acquisition events among the same strain. Clonal spread was suggested to have occurred at one of the participating sites on two separate occasions involving two ESBL-producing E coli. The ESBL-producing K pneumoniae were determined to be genetically unrelated. A fingerprint was not obtained for one ESBL-producing K pneumoniae due to autodigestion of the DNA.

DISCuSSION
The prevalence and molecular epidemiology of ESBL-producing E coli and Klebsiella species in Canada has been monitored since the late 1990s (7)(8)(9)(10)(11)(12). Throughout these studies, we have observed increases in the prevalence and changing trends in the molecular epidemiology of ESBL producers. The present study updates the prevalence and molecular epidemiology of ESBL producers among Canadian hospitals.
Our study found the prevalence of ESBL-producing E coli, K pneumoniae, K oxytoca and P mirabilis isolated from Canadian hospitals in 2007 to be 3.4%, 1.6%, 0% and 0%, respectively. These results are similar to the prevalence rates we observed from our Canadian Intensive Care Unit (CAN-ICU) Surveillance study from 2005 to 2006, where the prevalences of ESBL-producing E coli and Klebsiella species were determined to be 3.7% and 1.8%, respectively (9,12). However, in contrast to the CAN-ICU study, the majority of ESBL producers from the CANWARD 2007 study were obtained from medical wards (30 of 60 ESBL-producers; 50.0%) with very few coming from intensive care units (four of 60 ESBL-producers; 6.7%) (P=0.0001).
CTX-M ESBLs were the most common variants observed among both ESBL-producing E coli and K pneumoniae, with CTX-M-15 being the most common genotype among both as well as CTX-M-14 among E coli. This is not novel among ESBL-producing E coli (9,10,12,13). However, this is the first report of CTX-M becoming the predominant enzyme among ESBL-producing K pneumoniae in Canada, replacing SHV (7,11). In addition, we describe the first CTX-M-65-producing E coli in Canada. CTX-M-65, a variant of CTX-M-14, was first reported and identified in the United States in 2008 (14).
The majority of ESBL-producing E coli also carried the narrowspectrum bla TEM-1 gene, whereas ESBL-producing K pneumoniae more commonly carried a narrow-spectrum bla SHV gene.
E coli has become the predominant organism producing ESBLs in Canada over the past decade with the spread of CTX-M (9,12). However, with the emergence and increase of CTX-M-producing K pneumoniae, we may observe increases in their prevalence if CTX-M spreads as efficiently among K pneumoniae as it did among E coli.
Surprisingly, ESBL-producing E coli and K pneumoniae showed a trend to be more commonly isolated from blood, respiratory tract and wound specimens in comparison with urine specimens. This suggests that ESBL producers may be becoming more invasive; however, other factors that were not assessed by the present study could also be contributing to this observation, such as previous antimicrobial therapy, length of hospital stay and underlying disease.
Therapeutic options for ESBL producers are severely limited because they are often MDR. Both ESBL-producing E coli and K pneumoniae displayed high coresistance rates to the fluoroquinolones, trimethoprim/sulfamethoxazole, doxycycline and gentamicin, with a high percentage exhibiting a MDR phenotype. Although resistance rates were high, ESBLproducing E coli and K pneumoniae still remained susceptible to the carbapenems and ESBL-producing E coli remained susceptible to tigecycline. Alternative therapies such as nitrofurantoin (urinary tract infection only) and colistin (polymyxin E) still remain potential options for ESBLproducing E coli because they were 98.0% and 96.0% susceptible, respectively.
Molecular typing of the isolates revealed that the majority of ESBL producers were genetically unrelated. This suggests that horizontal transfer of plasmids bearing the ESBL gene plays a larger role in the spread of ESBLs across Canada compared with clonal spread of an epidemic strain.
In summary, we update the prevalence and molecular epidemiology of ESBL-producing Enterobacteriaceae from hospitals across Canada in 2007. The prevalence of ESBL-producing E coli, K pneumoniae, K oxytoca and P mirabilis in Canadian hospitals is 3.4%, 1.6%, 0% and 0%, respectively. CTX-M has become the most prevalent enzyme among both ESBL-producing E coli and K pneumoniae. Coresistance rates were high among ESBL producers to many non-betalactam agents, with a high percentage exhibiting a MDR phenotype. The spread of ESBL producers across Canada is polyclonal and is not due to the clonal spread of a single strain.