Journal of Antimicrobial Chemotherapy (2008) 62, Suppl. 2, ii41– ii54 doi:10.1093/jac/dkn351

Non-susceptibility trends among Enterobacteriaceae from bacteraemias in the UK and Ireland, 2001 – 06 David M. Livermore1*, Russell Hope1, Geraldine Brick1, Mark Lillie1 and Rosy Reynolds2 on behalf of the BSAC Working Parties on Resistance Surveillance 1

Health Protection Agency Centre for Infections, 61 Colindale Avenue, London NW9 5EQ, UK; 2 Department of Medical Microbiology, Southmead Hospital, Bristol BS10 5NB, UK

Background: Enterobacteriaceae are common agents of bacteraemia, with Escherichia coli accounting for 20% of the cases. Reflecting this importance, members of the family constitute 4 of the 12 collection groups in the British Society for Antimicrobial Chemotherapy (BSAC) Bacteraemia Surveillance Programme. Methods: MICs from the BSAC surveillance programme were reviewed, along with bacteraemia reports received by the Health Protection Agency (HPA) via its CoSurv/LabBase system. Isolates with unusual resistances were subjected to molecular analysis. Results: The BSAC and HPA systems both revealed dramatically increasing resistance to cephalosporins, ciprofloxacin and gentamicin among E. coli and Klebsiella spp., with cephalosporin resistance largely contingent on the spread of CTX-M extended-spectrum b-lactamases (ESBLs); fluoroquinolone resistance also increased among Proteus mirabilis and ESBL-negative E. coli. Carbapenem resistance remained extremely rare, but two Enterobacter spp., from the same patient in different years, had KPC carbapenemases, while a few isolates had carbapenem resistance contingent upon combinations of blactamase and impermeability, and ertapenem MICs for AmpC-derepressed Enterobacter spp. rose over time. Three new agents—ceftobiprole, doripenem and tigecycline—were tested. Ceftobiprole was broadly active, except against ESBL producers and Klebsiella oxytoca hyperproducing K1 enzyme, and was variable against AmpC-derepressed Enterobacter spp. and against Proteus vulgaris. Doripenem was more potent than imipenem. Tigecycline was almost universally active against E. coli, but low-level non-susceptibility (MIC 2 mg/L) was frequent among Klebsiella spp. Conclusions: E. coli and Klebsiella spp. showed dramatic shifts, with sharply rising non-susceptibility to cephalosporins, ciprofloxacin and gentamicin. The rise in cephalosporin resistance reflected dissemination of CTX-M ESBLs. Carbapenems remain broadly active and new agents offer potential. Keywords: BSAC, HPA, surveillance

Introduction Escherichia coli accounts for around one-fifth of all bacteraemias in the UK, a proportion only rivalled by Staphylococcus aureus and at least double that for any other pathogen. Other important Enterobacteriaceae in bloodstream infections include Klebsiella spp., causing 4.7% to 6.0% of bacteraemias; Proteeae, causing 2.6% to 4.1% and Enterobacter spp. causing 2.6% to 3.3%.1,2 These proportions have been stable over the past 15 years but there is evidence, from reports to the Health Protection Agency (HPA) and the European Antimicrobial Resistance Surveillance System (http://www.earss.rivm.nl), of

rising resistance, particularly in E. coli and to fluoroquinolones and cephalosporins.3 The rise in cephalosporin resistance may be attributable to the spread of extended-spectrum b-lactamases (ESBLs), particularly CTX-M types.4 The British Society for Antimicrobial Chemotherapy (BSAC) Bacteraemia Resistance Surveillance Programme is structured to reflect the importance of Enterobacteriaceae, which comprise 4 of the 12 collection groups (E. coli, Klebsiella spp., Enterobacter spp. and Proteeae), while Serratia and Citrobacter spp. are well represented in the ‘other Gram-negative’ group. This analysis reviews the BSAC results from 2001 to 2006 along with Hospital Trusts’ bacteraemia data, as reported to the HPA

.....................................................................................................................................................................................................................................................................................................................................................................................................................................

*Correspondence address. Antibiotic Resistance Monitoring and Reference Laboratory, HPA Centre for Infections, 61 Colindale Avenue, London NW9 5EQ, UK. Tel: þ44-20-8327-7223; Fax: þ44-20-8327-6264; E-mail: [email protected] .....................................................................................................................................................................................................................................................................................................................................................................................................................................

ii41 # The Author 2008. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. from https://academic.oup.com/jac/article-abstract/62/suppl_2/ii41/690351 For Permissions, please e-mail: [email protected] College of Wisconsin Libraries-Serials user

Downloaded by Medical on 17 August 2018

Livermore et al. under its LabBase/CoSurv system which covers England, Wales and Northern Ireland but not Scotland nor the Irish Republic.5 Over 90% of Hospital Trusts report to the HPA, and the system captures data for around 70% of all bacteraemias in its geographic ambit. These reports are complementary to the BSAC surveillance, with a much larger sample, but with less standardization of methods or antimicrobial panels, and without the facility for molecular investigation of unusual isolates. The present paper also reviews the activity of the newer broad-spectrum antibiotics included in the BSAC surveillance, specifically ceftobiprole, doripenem and tigecycline.

Materials and methods The collection, identification and susceptibility testing protocols of the BSAC surveillance are described elsewhere in this supplement.6 Isolates inferred to have ESBLs were investigated for blaCTX-M by multiplex PCR;7 those non-susceptible to carbapenems were screened for blaKPC genes by specific PCR8 and for metallo-blactamase genes either by specific PCR9 or, from 2005, by multiplex PCR.10 The HPA LabBase/CoSurv system for the collection of laboratories’ voluntarily submitted data on bacteraemia isolates has been described previously.5

Results Pathogens, patients and sources From 2001 to 2006, the BSAC Bacteraemia Resistance Surveillance Programme collected 1480 E. coli, 1419 Klebsiella spp., 1220 Enterobacter spp., 1291 Proteeae and, among the ‘other Gram-negative bacteria’, 437 Serratia spp., 170 Citrobacter spp., 23 Raoultella terrigena, 19 Pantoea and 9 isolates of enterobacterial genera with fewer than five representatives each. We undertook analysis wherever there were more than 100 isolates of a species or genus, but excluded Citrobacter spp. on the ground that the collection was evenly split between Citrobacter freundii (n ¼ 88) and Citrobacter koseri (diversus) (n ¼ 61)—species that differ in their b-lactamase and resistance profiles.11 There were differences in sources between the E. coli and Proteeae isolates, on the one hand, and the Klebsiella, Enterobacter and Serratia groups on the other hand (Table 1). Specifically: (i) E. coli and Proteeae were more often from the oldest patients, with a mode age 80 years, compared with 70–79 years for Klebsiella, Enterobacter and Serratia; (ii) outpatients and those hospitalized for ,48 h accounted for just over half of the E. coli and Proteeae bacteraemias but ,40% of the Enterobacter, Klebsiella and Serratia groups; (iii) around 17% of the E. coli and Proteeae were from patients presenting at the accident and emergency department, compared with ,10% for the other species, where intensive care, haematology and oncology were frequent sources; and (iv) lastly, .40% of E. coli and Proteeae bacteraemias had a genitourinary origin, compared with 13% to 23% for Klebsiella, Enterobacter and Serratia spp., where ‘unknown’ was the most-cited source, .34% for each genus.

Resistance trends Many prevalence rates for non-susceptibility in the BSAC data set showed sizeable year-to-year fluctuation, probably reflecting the

small sample sizes (250 isolates per group per year) and the fact that several breakpoints cut the MIC distributions for widespread resistance types [European Committee on Antimicrobial Susceptibility Testing (EUCAST) breakpoints aim not to split MIC distributions for wild-type populations, but may do so for those with common modes of acquired resistance]. Consequently, we only accepted trends as convincing when: (i) they were apparent for both the proportion non-susceptible and the proportion resistant, and (ii) when they were supported by trends in the HPA CoSurv/LabBase data, with its much larger sample size.5 We believe that this strategy is robust, but note two limitations: (i) reporting to the HPA system is as ‘susceptible’ and ‘nonsusceptible’, with no distinction of intermediate and resistant; and (ii) reported categorizations are based on the definitions of nonsusceptibility used by the source laboratories at the time of reporting. In most cases, these were the then-current BSAC values, not the EUCAST-harmonized breakpoint values applied here to the BSAC surveillance data; however, a sizeable, if diminishing, minority of laboratories used Stokes’ method in the earlier years, while a small but growing minority use automated systems, some of them calibrated, by default or intention, to CLSI criteria.

E. coli E. coli is the commonest agent of bacteraemia and, among all those considered in the BSAC surveillance, is the one showing the most striking changes in resistance. The BSAC data indicated significant up-trends in non-susceptibility and resistance to ceftazidime, cefotaxime, ciprofloxacin and gentamicin (P , 0.0001, except gentamicin, P ¼ 0.0035), with corresponding shifts also evident in the LabBase/CoSurv reports to the HPA (Table 2). Significant up-trends in non-susceptibility, in both the BSAC and HPA datasets, to cefuroxime and piperacillin/ tazobactam, which have no intermediate category, also were seen. Increasing non-susceptibility to amoxicillin/clavulanate too was apparent in both datasets, and although this latter trend was not confirmed by rising resistance in the BSAC data, we accept it as significant because (i) it parallels the trends for piperacillin/ tazobactam, and (ii) frank resistance to amoxicillin/clavulanate (MIC .16 þ 8 mg/L) is difficult to achieve when many E. coli isolates are inhibited by clavulanate alone at 8 mg/L. The BSAC data indicated rising non-susceptibility to cefoxitin (P , 0.001) but results for this drug are rarely reported under LabBase surveillance, precluding confirmation, and we remain sceptical, since the single cefoxitin breakpoint lies at the upper edge of the normal MIC distribution, exacerbating the risk of categorization errors. Both surveillance systems revealed a high proportion of E. coli non-susceptible to amoxicillin: the HPA LabBase data suggested that this fraction was still rising (P , 0.0001) but this was not supported by the BSAC data. Non-susceptibility to carbapenems and tigecycline was extremely rare in E. coli. An apparently falling non-susceptibility trend for imipenem in the LabBase data set is certainly spurious and perhaps reflects an improvement in the quality control of laboratories’ susceptibility testing. Nevertheless, two imipenem non-susceptible (MIC 8 and 16 mg/L) E. coli were collected by the BSAC surveillance in 2006 (Table 3), with none found previously. They were from separate centres. One, EO1441, had unstable imipenem resistance, lost on subculture, contingent on a combination of a CTX-M ESBL plus impermeability; the other, EO1452, had narrow-spectrum carbapenem resistance and

ii42 Downloaded from https://academic.oup.com/jac/article-abstract/62/suppl_2/ii41/690351 by Medical College of Wisconsin Libraries-Serials user on 17 August 2018

Resistance in Enterobacteriaceae Table 1. Sources of Enterobacteriaceae isolates Patient location at the time of bacteraemia (%)a

male

female

Mode of age distribution (years)b

40.1

44

56

80

general medicine (28%), genitourinary tract (44%), unknown accident and (31%), gastrointestinal emergency (16%), tract (14%) surgery (14%)

33.2

57.2

56

43

70–79

general medicine (21%), unknown (35%), genitourinary tract surgery (17%), (23%), gastrointestinal haematology/ tract (17%), oncology (15%) line-associated (14%)

4.7

23.5

66.9

61

38

70–79

surgery (17), haematology/ oncology (17%), general medicine (16%) intensive care (16%)

Proteeae (n ¼ 1291)

9.3

41.6

47.3

61

39

80

general medicine (27%), genitourinary (45%), unknown (31%) accident and emergency (19%), surgery (14%)

Serratia spp. (n ¼ 437)

4.4

23.1

69.8

66

34

70–79

surgery (21%), general medicine (18%), nephrology (15%), intensive care (12%)

Citrobacter spp. (n ¼ 170)

5.9

33.5

58.8

63

36

70–79

general medicine (22%), unknown (34%), genitourinary tract haematology/ (23%), line-associated oncology (15%) (20%), gastrointestinal surgery (13%), tract (15%) intensive care (12%), nephrology (11%)

Sex (%)a nonhospitald

hospital ,48 h

hospital .48 h

E. coli (n ¼ 1480)

8.4

49.7

Klebsiella spp. (n ¼ 1419)

6.4

Enterobacter spp. (n ¼ 1220)

Organism

Major specialtiesc

Major sourcesc

unknown (35%), line-associated (22%), gastrointestinal tract (14%), genitourinary tract (13%)

unknown (41%), line-associated (18%), genitourinary tract (14%), gastrointestinal tract (11%)

a

Percentages do not total 100 because data were missing for a few isolates. Age bands were ,5, 5 –19, 20– 29, 30– 39, 40–49, 50– 59, 60–69, 70–79 and 80 years. c Sources and specialties are listed when they accounted for .10% of cases. d Community or outpatients. b

remains under study. Four tigecycline non-susceptible E. coli were collected, scattered through the surveillance, all with ‘intermediate’ MICs of 2 mg/L. From 2002, the nature of cephalosporin resistance was investigated for the BSAC collection. The number of isolates with non-CTX-M ESBLs fluctuated from two to six per annum, without trend, while the number with CTX-M ESBLs rose from 2 in 2002 to 27 in 2006, exceeding 10% of all E. coli collected in the latter year (Table 4). Resistances to amoxicillin/clavulanate, ceftazidime, cefotaxime, ciprofloxacin, gentamicin, cefoxitin, cefuroxime and piperacillin/tazobactam were strongly associated with ESBL production (P , 0.0001) and so the rises in non-susceptibility to these drugs were strongly associated with the proliferation of ESBL producers. In

addition, there was also an independent trend of increasing ciprofloxacin non-susceptibility and resistance (P , 0.001) among isolates lacking ESBLs. Logistic regression analysis of the BSAC data set, adjusted for inter-centre variation,12 implicated hospitalization for .48 h before acquisition of infection and male sex as independent predictors of increased resistance or non-susceptibility in E. coli; hospital-acquired isolates were more often ESBL producers, but an association of male sex and resistance was only partly ESBL-related. Of the two new antibiotics tested, doripenem had a unimodal MIC distribution, centred at 0.015 mg/L, and all MIC values lay between 0.008 and 0.06 mg/L, except for the two imipenemresistant isolates detailed in Table 3. Ceftobiprole had MICs of

ii43 Downloaded from https://academic.oup.com/jac/article-abstract/62/suppl_2/ii41/690351 by Medical College of Wisconsin Libraries-Serials user on 17 August 2018

Livermore et al. Table 2. Non-susceptibility and resistance trends (%) among Enterobacteriaceae isolates

Organism E. coli Amoxicillin

Amoxicillin/clavulanate

Cefotaxime

Cefoxitin Ceftazidime

Cefuroxime Ciprofloxacin

Ertapenem

Gentamicin

Imipenem

Piperacillin/tazobactam Tigecycline

ESBL Klebsiella spp. Amoxicillin/clavulanate

Cefotaxime

Cefoxitin Ceftazidime

Cefuroxime Ciprofloxacin

P value for trend

2001

2002

2003

2004

2005

2006

BSAC non-S BSAC R HPA non-S BSAC non-S BSAC R HPA non-S BSAC non-S BSAC R HPA non-S BSAC non-S5R HPA non-S BSAC non-S BSAC R HPA non-S BSAC non-S5R HPA non-S BSAC non-S BSAC R HPA non-S BSAC non-S BSAC R HPA non-S BSAC non-S BSAC R HPA non-S BSAC non-S BSAC R HPA non-S BSAC non-S5R HPA non-S BSAC non-S BSAC R HPA non-S BSAC

57.6 55.5 54.7 23.7 5.7 17.3 NT NT 1.7 10.2 FEW 2.9 1.2 2.2 11.4 8 8.2 8.2 6.7 NT NT FEW 4.9 4.1 3.6 0 0 1 2 4.2 NT NT FEW 0

60.8 56.8 55.7 24.0 6.4 21.3 2.4 0 2.9 8.0 FEW 3.6 2 2.5 8.4 8.1 6.8 6.4 8.7 0 0 FEW 5.6 4 4.9 0 0 0.3 4.8 6.5 0.4 0 FEW 3.2

61.7 58.9 56.5 27.4 6.5 21.7 2.0 2.0 5.2 9.7 FEW 4.4 3.2 4.3 8.1 10.8 10.9 10.5 12.8 0 0 FEW 2.8 2.4 6.4 0 0 1.2 6.9 6.3 0.8 0 FEW 2.4

56.0 53.6 57.0 28.6 4.4 20.6 5.6 5.6 6.1 16.1 FEW 4.8 2.4 6.4 11.7 11.7 18.1 16.1 15.8 0 0 FEW 6.5 6.5 7.4 0 0 0.1 10.5 6.1 0 0 FEW 6.0

59.9 58.3 59.0 27.5 7.7 22.0 8.5 7.7 8.6 13.0 FEW 8.9 2.8 9.2 13 15.2 17.4 16.6 19.4 0 0 FEW 7.3 6.5 7.8 0 0 0.2 3.2 7.3 0 0 FEW 7.3

67.4 64.5 60.0 43.8 8.3 43.8 12.0 11.6 10.8 20.2 FEW 11.2 7.9 11.6 19.4 16.4 25.6 25.2 22.8 0.8 0.8 FEW 11.2 9.1 9 0.8 0.4 0.2 10.3 8.9 0.4 0 FEW 12.0

0.10 0.072 ,0.0001