Dramatic rise in the proportion of ESBL-producing Escherichia coli and Klebsiella pneumoniae among clinical isolates identified in Canadian hospital laboratories from 2007 to 2016 Free

Abstract

Objectives

To assess the prevalence, antimicrobial susceptibilities and molecular characteristics of ESBL-producing Escherichia coli and Klebsiella pneumoniae infecting patients receiving care in Canadian hospitals from January 2007 to December 2016.

Methods

Clinical isolates of E. coli (n = 8387) and K. pneumoniae (n = 2623) submitted to CANWARD, an ongoing Canadian national surveillance study, were tested using the CLSI reference broth microdilution method to determine their susceptibility to 15 antimicrobial agents. ESBL-producing E. coli and K. pneumoniae confirmed by the CLSI phenotypic method and putative AmpC-producing E. coli underwent PCR testing and DNA sequencing to identify resistance genes. Annual proportions of isolates harbouring ESBL and AmpC genes were assessed by the Cochran–Armitage test of trend.

Results

The annual proportion of isolates of E. coli that were ESBL producing increased from 3.4% in 2007 to 11.1% in 2016 (P < 0.0001); >95% of ESBL-producing E. coli were susceptible to amikacin, colistin, ertapenem, meropenem and tigecycline. The proportion of isolates of K. pneumoniae that were ESBL producing increased from 1.3% in 2007 to 9.7% in 2016 (P < 0.0001); >95% of ESBL-producing K. pneumoniae were susceptible to amikacin and meropenem. CTX-M-15 was the predominant genotype in both ESBL-producing E. coli (64.2% of isolates) and ESBL-producing K. pneumoniae (51.0%). The annual proportion of isolates of E. coli that were AmpC producing [annual proportion mean 1.9% (range 0.3%–3.1%)] was unchanged from 2007 to 2016 (P > 0.5).

Conclusions

The prevalence of both ESBL-producing E. coli and K. pneumoniae increased significantly in Canada during the study period while the prevalence of AmpC-producing E. coli remained low and stable.

Introduction

Over the last 25 years, ESBL-producing Enterobacteriaceae have emerged and spread worldwide. ESBL-producing Enterobacteriaceae are characteristically resistant to oxyimino-cephalosporins (e.g. cefotaxime, ceftriaxone, ceftazidime), extended-spectrum penicillins (e.g. piperacillin) and monobactams (e.g. aztreonam).^1,2 They are also commonly fluoroquinolone resistant and possess an MDR phenotype.^1,2 Conjugative plasmids carrying CTX-M-type ESBL genes have been the principal driver of ESBL spread in Enterobacteriaceae. However, even though CTX-M-type ESBLs predominate in both Escherichia coli and Klebsiella pneumoniae, patients with bacteraemia due to E. coli and K. pneumoniae have been shown to possess substantially different risk profiles and transmission dynamics.³ ESBL-producing E. coli are frequently the pandemic clone, ST131.^1,2

In addition to ESBL carriage, resistance to oxyimino-cephalosporins in E. coli can also arise by the acquisition of a plasmid-encoded AmpC β-lactamase or by mutation(s) within the promoter/attenuator region of the chromosomal ampC gene resulting in constitutive overexpression.⁴ Expression of AmpC β-lactamases in E. coli is often associated with an MDR phenotype and with treatment failure in patients receiving oxyimino-cephalosporin therapy.^1,5 Resistance associated with AmpC β-lactamases may escape detection using current clinical laboratory antimicrobial susceptibility testing methods.^1,5 AmpC-producing E. coli and other Enterobacteriaceae typically differ from ESBL-producing isolates by demonstrating a negative phenotypic ESBL confirmatory test, resistance to cephamycins (e.g. cefoxitin) and susceptibility to cefepime.¹

Carbapenems are often reserved as last-line agents in the treatment of serious or highly resistant Gram-negative infections, including infections caused by ESBL-producing Enterobacteriaceae.^6,7 Exponential increases in infections caused by ESBL-producing Enterobacteriaceae have resulted in increased use of carbapenems and subsequently the emergence and increased isolation of carbapenem-resistant Enterobacteriaceae.⁸ ESBLs and AmpC β-lactamases generally lack the ability to hydrolyse carbapenems; however, reduced susceptibility or resistance to carbapenems may arise in Gram-negative bacilli when these enzymes are accompanied by decreased outer-membrane permeability. Infections attributable to ESBL-, AmpC- and carbapenemase-producing Enterobacteriaceae are burgeoning public health threats, challenge infection control programmes, and are often associated with delays in the administration of effective antimicrobial therapy, high mortality, prolonged hospital stay and increased hospital costs.^2,3,7,8

The treatment of infections caused by carbapenemase-producing Enterobacteriaceae is challenging. Second-line agents, such as colistin, tigecycline, aminoglycosides and fosfomycin, may be active against some isolates of carbapenemase-producing Enterobacteriaceae but these agents are frequently associated with toxicities and/or increasing resistance.¹ The recently approved combination agents ceftazidime/avibactam and meropenem/vaborbactam may provide therapy for patients infected with isolates of Gram-negative bacilli carrying Ambler class A β-lactamases, including ESBLs and KPC carbapenemases, AmpC β-lactamases and some class D β-lactamases (e.g. OXA-48 carbapenemase), including isolates carrying ESBL and AmpC enzymes in combination with impaired permeability.⁹ However, Enterobacteriaceae and Pseudomonas aeruginosa producing class B β-lactamases (metallo-β-lactamases) are not susceptible to ceftazidime/avibactam or meropenem/vaborbactam.^9,10 Other agents with activity against some carbapenemase-producing Enterobacteriaceae are in different stages of development, including imipenem/relebactam, plazomicin, cefiderocol, eravacycline and aztreonam/avibactam.

The primary objectives of the current study were to assess the prevalence, antimicrobial susceptibilities and molecular composition of ESBL-producing E. coli and K. pneumoniae infecting patients receiving care in Canadian hospitals from 2007 to 2016. Secondary objectives included determining the prevalence, antimicrobial susceptibilities and molecular composition of AmpC-producing E. coli.

Materials and methods

Bacterial isolates

The bacterial isolates tested in this study were collected as part of the CANWARD surveillance study¹¹ from January 2007 to December 2016. CANWARD is an ongoing Canadian national multicentre surveillance study that monitors the in vitro activities of antimicrobial agents tested against bacterial pathogens isolated by clinical laboratories in tertiary care hospitals. Each isolate accepted into CANWARD was deemed clinically significant by protocols in place at the submitting laboratory. Isolates are collected consecutively, one per patient, per site of infection, from both inpatients and outpatients attending emergency rooms, hospital clinics, medical/surgical wards and ICUs each year. Each participating laboratory is asked to submit annual quotas of respiratory, wound, urine and bloodstream isolates. All isolates are shipped to the CANWARD coordinating laboratory (Health Sciences Centre, Winnipeg, Canada), where their identities are confirmed by colony appearance, spot testing¹² and/or MALDI-TOF MS (Bruker Daltonics, Billerica, MA, USA). Isolates are then stocked in skim milk and stored at −80°C until antimicrobial susceptibility testing is performed. Tertiary care hospitals in 8 of the 10 Canadian provinces participated in the CANWARD study: 12 tertiary care hospitals in 2007, 10 in 2008, 15 in 2009, 14 in 2010, 15 in 2011, 12 in 2012, 15 in 2013, 13 in 2014, 13 in 2015 and 13 in 2016. Eight tertiary care hospitals participated in every year of CANWARD from 2007 to 2016. CANWARD receives annual approval by the University of Manitoba Research Ethics Board (H2009:059).

Antimicrobial susceptibility testing

Broth microdilution antimicrobial susceptibility testing was performed in accordance with CLSI standards.¹³ Custom-designed 96-well microtitre panels were prepared in-house, as previously described.¹¹ MICs were interpreted using CLSI interpretative breakpoints when available.¹⁴ US FDA MIC interpretative breakpoints were used for tigecycline (susceptible, ≤2 mg/L; intermediate, 4 mg/L; resistant, ≥8 mg/L)¹⁵ and EUCAST MIC breakpoints for colistin tested against Enterobacteriaceae (susceptible, ≤2 mg/L; resistant, ≥4 mg/L)¹⁶ were employed because neither CLSI nor FDA publishes breakpoints for these two agents.

Putative ESBL-producing isolates of E. coli and K. pneumoniae were identified by testing with a ceftriaxone and/or ceftazidime MIC of ≥1 mg/L. All putative ESBL producers were confirmed by the CLSI phenotypic confirmatory disc test.¹⁴ A putative AmpC phenotype in E. coli was defined as an isolate where the ceftriaxone and/or ceftazidime MIC was ≥1 mg/L, the cefoxitin MIC was ≥32 mg/L, and the isolate tested ESBL-negative by the CLSI phenotypic confirmatory disc test.¹⁴ Cefoxitin was only tested against 7386 of the 8387 E. coli isolates collected as only 557 of 1558 E. coli in 2007 were tested against this agent.

Molecular characterization of ESBL-producing E. coli and K. pneumoniae, and AmpC-producing E. coli

All phenotypically confirmed ESBL-producing E. coli and K. pneumoniae were screened for bla_SHV, bla_TEM, bla_CTX-M and bla_OXA as previously described.¹⁷ Putative AmpC-producing E. coli were screened for bla_ENT, bla_DHA, bla_FOX, bla_CIT and bla_CMY as previously described.¹⁸ Putative AmpC-producing E. coli isolates that tested negative by PCR for the aforementioned acquired AmpC β-lactamase genes were subjected to DNA sequencing to detect promoter/attenuator mutations in the chromosomal ampC gene.¹⁹

Statistical analysis

Statistical significance was calculated by the Cochran–Armitage test of trend test using JMP version 14 (SAS, Cary NC). Statistical significance in this study was defined as a P value ≤0.05.

Results

ESBL-producing E. coli

Of the 8387 isolates of E. coli collected by CANWARD from 2007 to 2016, 12.3% (1031/8387) had ceftriaxone and/or ceftazidime MICs of ≥1 mg/L. Overall, 6.4% (539/8387) of all isolates of E. coli tested from 2007 to 2016 were phenotypically confirmed as ESBL producers (Table 1). The national prevalence of ESBL-producing E. coli was variable throughout the study period, with a significant increase observed nationally from 2007 (3.4%) to 2016 (11.1%) (P < 0.0001). There were also significant increases in the prevalence of ESBL-producing E. coli in three of the four national regions (Table 1). ESBL-producing E. coli were more frequently observed in isolates from males, in patients aged ≥18 years, in hospitalized patients (ICU, medical ward and surgical ward) and from wound specimens (Table 2). More than 95% of ESBL-producing E. coli were susceptible to amikacin, colistin, ertapenem, meropenem and tigecycline; 93.5% of ESBL-producing E. coli were susceptible to piperacillin/tazobactam (Table 3). Among ESBL-producing E. coli, CTX-M-type β-lactamases were the dominant family of enzyme identified (92.9%), with CTX-M-15 being the dominant variant observed (64.2%). The majority of ESBL-producing E. coli (68.3%; 368/539) harboured multiple β-lactamases, including the narrow-spectrum β-lactamases OXA-1 (45.8%; 247/539) and TEM-1 (31.4%; 169/539).

ESBL-producing K. pneumoniae

Of the 2623 isolates of K. pneumoniae collected by CANWARD from 2007 to 2016, 11.7% (306/2623) had ceftriaxone and/or ceftazidime MICs of ≥1 mg/L. Overall, 4.0% (104/2623) of all isolates of K. pneumoniae tested from 2007 to 2016 were phenotypically confirmed as ESBL-producers (Table 1). The national prevalence of ESBL-producing K. pneumoniae increased significantly from 2007 (1.3%) to 2016 (9.7%) (P < 0.0001); however, on a regional level only the increases observed in the British Columbia/Alberta/Saskatchewan/Manitoba region and Ontario were found to be significant (Table 1). ESBL-producing K. pneumoniae were more frequently observed in isolates from males, in patients aged ≤64 years and in hospitalized patients (ICU, medical ward and surgical ward) (Table 2). More than 95% of ESBL-producing K. pneumoniae were susceptible to amikacin and meropenem while only 63.6% of isolates were susceptible to piperacillin/tazobactam (Table 3). Genotypic diversity was greater among ESBL-producing K. pneumoniae (46 different genotypes in 104 K. pneumoniae isolates) than among ESBL-producing E. coli (36 different genotypes in 539 E. coli isolates) (Tables 4 and 5). Among ESBL-producing K. pneumoniae, CTX-M-type β-lactamases were the dominant family of enzyme identified (69.2%), with CTX-M-15 being the dominant variant observed (51.0%). The majority of ESBL-producing K. pneumoniae (89.4%; 93/104) harboured multiple β-lactamases, including the narrow-spectrum β-lactamases TEM-1 (52.9%; 55/104), OXA-1 (40.4%; 42/104) and SHV-1 (24.0%; 25/104).

AmpC-producing E. coli

Of the 8387 isolates of E. coli collected by CANWARD from 2007 to 2016, 1.9% (162/8387) demonstrated an AmpC phenotype (Table 1). The annual proportion of isolates of E. coli identified as AmpC-producing did not change significantly over the study period (P = 0.1311) and no significant regional differences were observed. The prevalence of AmpC-producing E. coli was not significantly different when isolates were analysed by patient gender, age, hospital location or specimen source (Table 2). More than 95% of AmpC-producing E. coli were susceptible to amikacin, colistin, ertapenem, meropenem and tigecycline; 94.1% and 90.1% of AmpC-producing E. coli were susceptible to cefepime and piperacillin/tazobactam, respectively (Table 3). Of 162 isolates of AmpC-producing E. coli, 87 (53.7%) contained an acquired AmpC β-lactamase gene, of which 86 (98.8%) carried CMY-2 and 1 (1.2%) carried FOX-5. The remaining 75 isolates (46.3%) were PCR-negative for the acquired AmpC β-lactamase genes tested in this study, and were found to contain promoter/attenuator mutations within the chromosomal ampC gene of E. coli (data not shown). No carbapenemase-producing isolates were identified in this study.

Discussion

In the current study, the proportions of ESBL-producing E. coli (mean, 6.4%; range 3.0%–12.4%) and K. pneumoniae (mean, 4.0%; range, 1.3%–9.7%) increased significantly (P < 0.0001) in Canada from 2007 to 2016 (Table 1). In the case of E. coli, significant increases were observed for three of the four regions of Canada; a significant increase was not observed for the New Brunswick/Nova Scotia region. For K. pneumoniae, significant regional increases in ESBL-producing isolates were only observed in Ontario and the western region of Canada (British Columbia/Alberta/Saskatchewan/Manitoba).

The proportion of isolates of E. coli identified as ESBL-producing in the current study is comparable to that previously published in a study of blood culture isolates of E. coli in a health region in western Canada from 2000 to 2010 [a mean of 4% (197/4698) of isolates were ESBL producers; range, 0.3% (1/335) in 2000 to 13.7% (63/459) in 2010].²⁰ A second study from the same health region reported a mean annual rate of 0.5% for ESBL-producing K. pneumoniae for the 10 year period from 2000 to 2009 [range, 0.1% (2/1440) in 2000 to 1.2% (24/2023) in 2009],²¹ considerably lower than the 4.0% observed in the current study. The SMART surveillance study, which included seven laboratory sites in Canada from 2010 to 2014, reported that the proportion of isolates of urinary E. coli that were ESBL producing increased from 10.4% in 2010 to 13.0% in 2014, which approached significance (P = 0.079).²²

E. coli ST131 was previously reported to be the major factor influencing the spread of CTX-M-15 in Canadian hospitals, with 102 of 153 CTX-M-15 producers belonging to ST131,¹ and this is also consistent with reports from the USA, where ST131 comprised 56% of CTX-M-15-producing isolates.²³ Given that CTX-M-15 was the dominant CTX-M variant observed in the current study (64.2%), we speculate that continued success of the ST131 clone of E. coli was responsible for the observed increase in ESBL-producing E. coli in the current study, as previously reported.^1,20

The factors driving increased numbers of ESBL-producing K. pneumoniae are less clear as published data do not indicate that the presence of any one dominant ST promotes spread.²¹ In one Canadian study of ESBL-producing K. pneumoniae, the major STs (ST17, ST20, ST573, ST575) formed only 32% of isolates.²¹ In another previous Canadian study, a large number of ESBL-producing K. pneumoniae (50%) were found to produce CTX-M-15, but 12 different genotypes were identified among 24 CTX-M-15-positive isolates, suggesting spread is not due to the success of a single clone.¹ In contrast, a study performed in Korea reported that 70% of ESBL-producing K. pneumoniae belonged to ST11; however, multiple genotypes were identified within ST11, indicating that several acquisition events had occurred in this clone.²⁴ Another study determined that there was a higher rate of community acquisition among ESBL-producing E. coli than among ESBL-producing K. pneumoniae, indicating that community transmission plays a more important role for ESBL-producing E. coli than for ESBL-producing K. pneumoniae.³ In contrast, ESBL-producing K. pneumoniae were shown to be associated with recent ICU admission and previous transplantation.³

Data generated in the current study also contribute evidence to the ongoing debate regarding whether it is necessary for clinical laboratories to identify and report ESBL-producing Enterobacteriaceae.²⁵ Current CLSI cephalosporin breakpoints for Enterobacteriaceae were partly created to eliminate the requirement for ESBL confirmation in suspect isolates.¹⁴ In this study, only 52.3% of E. coli and 34.0% of K. pneumoniae isolates that met ESBL screening criteria for ceftriaxone (MIC ≥1 mg/L) and/or ceftazidime (MIC ≥1 mg/L) were phenotypically confirmed as ESBL producers. Clearly current CLSI susceptible breakpoints for oxyimino-cephalosporins tested against Enterobacteriaceae overcall the presence of ESBLs. Previously, a study comparing North American and non-North American isolates reported that only 45.0% of isolates of E. coli and K. pneumoniae that were ESBL screen positive were confirmed by disc diffusion methods, contradicting the CLSI and EUCAST recommendation that lowered cephalosporin breakpoints are sufficient for the detection of resistance genes and clinical decision making with regard to antimicrobial agent selection.^14,26,27 In the current study, 2.2%, 35.4% and 31.8% of ESBL-producing E. coli and 10.3%, 25.0% and 32.0% of ESBL-producing K. pneumoniae demonstrated in vitro susceptibility to ceftriaxone, ceftazidime and cefepime, respectively. A lack of ESBL identification and reporting may result in lapses in infection control and increase the risk of transmission.

In the current study, the proportion of isolates of E. coli that were AmpC-producing fluctuated from year to year but did not increase significantly on a national level, nor were any regional increases observed. Nevertheless, with an overall proportion of 1.9% among all E. coli collected, AmpC-producing isolates do play a role in antimicrobial-resistant infections in Canadian hospitals. The molecular basis of AmpC-mediated resistance in this study was the result of approximately equal proportions of isolates producing a plasmid-mediated AmpC β-lactamase (i.e. CMY-2) and isolates harbouring mutations within their chromosomal ampC gene, and is consistent with previous reports.^1,28 The lack of any significant trend may be due to lack of clonal spread among isolates with acquired AmpC β-lactamases, or may be due to stable selection pressure for chromosomal AmpC hyperproducers.

There is at least one data limitation in the current study. First, the significant increases observed in the annual proportion of ESBL-producing E. coli and K. pneumoniae from 2007 to 2016 were not linear. We cannot account for the lack of linearity (variability) in ESBL proportions between years. It does not appear to be caused by sampling bias due to sites not participating every year. Rather, it is likely the result of, at least partially, sampling variation, which is a recognized limitation of surveillance studies that attempt to estimate the characteristics of a population by examining a limited subset of that population.

In summary, we have provided an update on the prevalence and molecular epidemiology of ESBL- and AmpC-producing E. coli and K. pneumoniae isolated from patients receiving care in Canadian hospitals from 2007 to 2016. Over this 10 year period, the annual proportion of ESBL-producing E. coli increased from 3.4% in 2007 to 11.1% in 2016 (P < 0.0001) and ESBL-producing K. pneumoniae increased from 1.3% in 2007 to 9.7% in 2016, with ESBL-producing E. coli (annual proportion mean 6.4%) being more common than ESBL-producing K. pneumoniae (annual proportion mean 4.0%). CTX-M-15 was the predominant genotype in both ESBL-producing E. coli (64.2% of isolates) and ESBL-producing K. pneumoniae (51.0%). During this same 10 year period, the proportion of isolates expressing acquired AmpC enzymes or overexpressing chromosomal AmpC enzymes has remained low and stable [annual proportion mean 1.9% (range 0.3%–3.1%)].

Acknowledgements

The data in this paper were previously presented in part at the ASM MICROBE meeting (2017) in New Orleans, LA (Poster SUNDAY-109). CANWARD data can also be found at www.can-r.ca, the official website of CARA.

We thank the participating centres, investigators and laboratory site staff from the CANWARD sites for their continued support and cooperation. We also thank the Public Health Agency of Canada - National Microbiology Laboratory (PHAC-NML) for their support of this project.

Members of CARA

The CARA principal members include George G. Zhanel, Daryl J. Hoban, Heather J. Adam, Melanie R. Baxter, Kimberly A. Nichol, Philippe R. S. Lagacé-Wiens, Andrew Walkty and James A. Karlowsky.

Members of CANWARD

The participating CANWARD sites (investigator) are: Royal University Hospital, Saskatoon, SK (Dr J. Blondeau); Children’s Hospital of Eastern Ontario, Ottawa, ON (Dr R. Slinger); Queen Elizabeth II Health Sciences Centre, Halifax, NS (Dr R. Davidson); Health Sciences Centre, Winnipeg, MB (Dr G. Zhanel/Dr D. Hoban); London Health Sciences Centre, London, ON (Dr J. Delport); South East Health Care Corp., Moncton, NB (Dr C. Ellis); Hôpital Maisonneuve-Rosemont, Montreal, QC (Dr M. Laverdière); Montreal General Hospital, Montreal, QC (Dr V. Loo); Royal Victoria Hospital, Montreal, QC (Dr V. Loo); Mount Sinai Hospital/University Health Network, Toronto, ON (Dr S. Poutanen); University of Alberta Hospital, Edmonton, AB (Dr J. Fuller); Vancouver Hospital, Vancouver, BC (Dr D. Roscoe); The Ottawa Hospital, Ottawa, ON (Dr M. Desjardins); St. Michael’s Hospital, Toronto, ON (Dr L. Matukas); CHRTR Pavillon Ste. Marie, Trois-Rivières, QC (Dr M. Goyette); St. Joseph’s Hospital, Hamilton, ON (Dr C. Lee); Centre Hospitalier Universitaire de Sherbrooke, Sherbrooke, QC (Dr A. Carignan); Cité de la Santé, Laval, QC (Dr M. Bergevin); and L’Hôtel-Dieu de Quebec, Quebec City, QC (Dr R. Pelletier).

Funding

The CANWARD study was supported in part by the University of Manitoba, Diagnostic Services Manitoba, the National Microbiology Laboratory, Astellas, Merck, Pfizer, Sunovion, The Medicines Company, Abbott, Achaogen, Cubist, Paladin Labs, Bayer, Janssen Ortho/Ortho McNeil, Affinium, Basilea, AstraZeneca, Paratek, Tetraphase, Theravance, Sanofi-Aventis and Zoetis.

Transparency declarations

G. G. Z. and D. J. H. have received research funding from Astellas, Merck, Pfizer, Sunovion, The Medicines Company, Abbott, Achaogen, Cubist, Paladin Labs, Bayer, Janssen Ortho/Ortho McNeil, Affinium, Basilea, AstraZeneca, Paratek, Tetraphase, Theravance, Sanofi-Aventis and Zoetis. All other authors: none to declare.

This article forms part of a Supplement sponsored by the University of Manitoba and Shared Health Manitoba, Winnipeg, Canada.

Disclaimer

The opinions expressed in this paper are those of the authors and do not necessarily represent those of Astellas, Merck, Pfizer, Sunovion, The Medicines Company, Abbott, Achaogen, Cubist, Paladin Labs, Bayer, Janssen Ortho/Ortho McNeil, Affinium, Basilea, AstraZeneca, Paratek, Tetraphase, Theravance, Sanofi-Aventis and Zoetis.

References

Author notes