CTX-M: changing the face of ESBLs in Europe

Abstract

Since around 2000—earlier in Poland and Spain and later in France and the UK—dramatic shifts have occurred in the prevalence and types of extended-spectrum β-lactamases (ESBLs) in Europe. Before this watershed, most producers were nosocomial isolates, often Klebsiella spp. or Enterobacter spp. from specialist care units, and had mutant TEM or SHV ESBLs. Subsequently, CTX-M ESBLs have become dominant, with much greater penetration into Escherichia coli, and with many infections in ‘complicated community’ patients, usually with underlying disease, recent antibiotic usage, or healthcare contact. The degree of clonality among producers varies with the country, as does the enzyme type produced, with group 9 (CTX-M-9 and -14) enzymes dominant in Spain and group 1 enzymes (particularly CTX-M-3 and -15) dominant elsewhere. Irrespective of the particular enzyme, most producers are multiresistant. These changing patterns present major therapeutic and infection control challenges, with the public health intervention points unclear.

extended-spectrum β-lactamases, antibiotic resistance, β-lactamases, CTX-M β-lactamases

Introduction

Until the late 1990s, European surveys of extended-spectrum β-lactamases (ESBLs) almost exclusively found TEM and SHV enzyme variants, often SHV-2 and SHV-5, and largely found these in Klebsiella spp. Many ESBL-producing outbreak strains of Klebsiella pneumoniae were described, with a few ‘epidemic’ clones affecting multiple hospitals, notably serotype K25 lineages with SHV-4 enzyme that spread in Belgium and France,1,2 and also an Enterobacter aerogenes clone with TEM-24 enzyme that became (and remains) widespread in Belgium and France.3–6 Affected patients were mostly in specialist units, and European surveys in 1994 and 1997–98 found ESBLs in 23–25% of all Klebsiella spp. from ICUs.7,8 CTX-M ESBLs were recorded rarely, although there were large outbreaks of Salmonella Typhimurium with CTX-M-4 and -5 enzymes in Latvia,9 Russia and Belarus6 in the mid-1990s.

These patterns have now changed dramatically, with CTX-M enzymes replacing TEM and SHV mutants as the predominant ESBLs in many European countries, with Escherichia coli joining K. pneumoniae as a major host, and with producers increasingly isolated from community patients. CTX-M enzymes are supplanting TEM and SHV variants in East Asia too,10,11 just as one of them (CTX-M-2) did in the early 1990s in Argentina.12 Only in North America are TEM and SHV mutants still dominant, though substantial outbreaks of E. coli with CTX-M-15 β-lactamase have occurred in Canada13,14 and a scatter of ICAAC abstracts describe CTX-M producers in the USA.

This paper, written on behalf of the β-Lactamase Work Packages of the EU-funded COBRA (Combating Resistance to Antibiotics by Broadening the Knowledge on Molecular Mechanisms behind Resistance to Inhibitors of Cell Wall Synthesis) Consortium summarizes the current European situation by country. It does not aim to be a comprehensive account of the biochemistry, genetics or evolution of CTX-M β-lactamases; others have reviewed these topics recently and well.15,16 It should however be said that over 50 CTX-M β-lactamases are recognized and that they divide into five clusters on sequence homology, namely the CTX-M-1, -M-2, -M-8, -M-9 and -M-25 groups (Author Webpage). Group 1 and 2 types evolved by the escape of chromosomal genes from Kluyvera ascorbata,17 whereas group 8 and 9 enzymes evolved via similar escapes from Kluyvera georgiana.18 Once mobilized, bla_CTX-M genes can be hosted by many elements, but most often by large multiresistance plasmids. ISEcp1 insertion sequence elements were often involved in the initial mobilization events13,19 but this is not universally true, particularly for group 2 and 9 enzymes,16,20 and analysis of a 12 kb region surrounding bla_CTX-M-10 from different isolates revealed links to a phage-related element, which may have facilitated the initial escape of bla_CTX-M-10 from Kluyvera spp. Expression is often from the promoter in ISEcp1, but may be from those in unusual class 1 integrons.21

CTX-M enzymes are more active against cefotaxime and ceftriaxone than ceftazidime, but point mutations can increase activity against ceftazidime; thus CTX-M-15 and -32 differ from CTX-M-3 and -1, respectively, solely by Asp-240→Gly substitutions, but are 100-fold more active against ceftazidime.22,23 Cephamycins, temocillin and carbapenems are not hydrolysed by CTX-M enzymes; clavulanate, tazobactam and sulbactam inhibit activity, but producers are often resistant to β-lactamase inhibitor combinations because of concurrent production of inhibitor-resistant penicillinase, (e.g. OXA-1), sometimes encoded by the same plasmids.13,24,25

CTX-M enzymes, by country

The United Kingdom

No isolates with CTX-M enzymes were reported in the UK before 2000, when a single Klebsiella oxytoca with CTX-M-9 was found.26 This was followed, in 2001, by a clonal outbreak in a Birmingham hospital, involving 30 patients and caused by a K. pneumoniae strain with CTX-M-26 β-lactamase.27 Also in 2001, a survey examined over 900 E. coli from 28 hospitals in the UK and Ireland and recorded four isolates with CTX-M-15 enzyme.28 These were unique by PFGE, were from three hospitals, and were a harbinger of what was to follow. Unnoticed strains with CTX-M enzymes were present earlier, as revealed by retrospective detection of CTX-M group 9 enzymes in Salmonella enterica serotype Virchow isolates from the 1990s, several of them from patients with a history of foreign travel.29

In mid-2003 the Health Protection Agency began to receive requests to investigate ‘outbreaks’ (in the loosest sense) of ESBL-producing E. coli from community as well as hospital patients. On receipt, over 90% of these isolates proved to have CTX-M-15 enzyme or, much more rarely, CTX-M-3 or -9.30 In the subsequent 24 months the Agency received over 1200 producers, from over half of all the clinical microbiology laboratories in the UK. These represent only a small fraction of all producers isolated. A prospective survey was undertaken late in 2004, covering 16 laboratories in south-east England, and seeking up to 100 consecutive cephalosporin-resistant Enterobacteriaceae per site. This yielded 1127 isolates with confirmed cephalosporin resistance.31 Among those with substantive mechanisms (ESBLs, AmpC or hyperproduced K1 enzyme), 51% were E. coli, and the largest group, comprising 292 isolates, were E. coli with CTX-M enzymes. CTX-M types also were the dominant ESBLs in Klebsiella spp. (Table 1). Just 4 years previously the dominant cephalosporin resistance types had been AmpC-derepressed Enterobacter spp. and Klebsiella spp. with TEM and SHV ESBLs, and only four E. coli with CTX-M enzymes were found.28,32

Table 1

Mechanisms found among consecutive cefpodoxime-, cefotaxime- or ceftazidime- resistant E. coli, Klebsiella spp. or Enterobacter spp. collected at 16 laboratories in London and south-east England from August to October 200431

	E. coli (n = 574)	Klebsiella spp. (n = 224)	Enterobacter spp. (n = 157)
CTX-M ESBL	292	190	6
Non-CTX-M ESBL	88	25	20
AmpC—chromosomal or plasmid-mediated	41	1	72
Hyperproduction of K1 chromosomal β-lactamase	0	8	0
No substantive mechanism defined; low-level resistance, mostly to cefpodoxime only	153	0	59

UK E. coli isolates with CTX-M-15 β-lactamase include five major serotype O25 clones, designated A–E, along with a great diversity of non-clonal producers.30 The major clones had 78% similarity by PFGE and may share a common ancestor. The most prevalent of them, A, is less resistant to cephalosporins than the others, because of an IS26 element between bla_CTX-M-15 and its normal promoter. It is dominant in several parts of the UK (Figure 1), whilst other areas have the more localized B–E clones, or predominantly have non-clonal producers. Around Preston (north-west England; light grey circle in Figure 1), some strain A E. coli additionally have a plasmidic AmpC β-lactamase, CMY-23.33 A minority (7%) of E. coli with CTX-M enzymes have group 9 (CTX-M-9 or -14) enzymes; these are non-clonal or form small local clusters.34 Less work has been done on Klebsiella spp. with CTX-M enzymes but, again, CTX-M-15 is dominant.31

Figure 1

Occurrence of the epidemic E. coli strain A with CTX-M-15 β-lactamase in the UK, 2003–5.

Regardless of species or enzyme type, most producers are multiresistant to aminoglycosides (except gentamicin for E. coli strain A), classical tetracyclines, quinolones, trimethoprim and sulphonamides. β-Lactamase/inhibitor combinations have variable activity: most isolates with CTX-M-15 enzyme are resistant because of concurrent production of OXA-1 penicillinases, but sporadic isolates with group 9 enzymes mostly are susceptible.30 Drugs generally remaining active include carbapenems, nitrofurantoin, fosfomycin, temocillin and tigecycline. A few K. pneumoniae isolates with CTX-M-15 enzyme are carbapenem-resistant because of porin loss, with ertapenem more compromised than imipenem or meropenem.35 These are widely scattered, but nowhere frequent.

Most (77%) producers are from urinary infections,31 with minorities from bacteraemias or other sites. Many (45% in the 2004 survey31) are indicated as coming from community patients, but diligent enquiry often reveals that these individuals are elderly, have multiple underlying health problems (e.g. diabetes or dementia), and have been hospitalized in the preceding 1–3 years.

In Shropshire, which was an early epicentre of infections with E. coli strain A, there were 25 deaths among the first 108 cases, with infection deemed contributory to death in 10 of the first 54 deaths, and possibly contributory in a further four.24 Stool cultures detected ESBL producers, including strain A, in 4.5% and 2.5% of diarrhoeal in- and outpatients, respectively in Shropshire.24 Isolates with CTX-M enzymes were also found in 2% of diarrhoeal patients around York,36 although these mostly had group 9 enzymes, as did those from diarrhoeal calves in Wales,37 implying little link to the major clinical problem. There is no suggestion that the CTX-M⁺E. coli were related to the diarrhoea; diarrhoeal faeces were screened solely because of laboratory availability. Nevertheless, gut carriage is potentially significant since most urinary and intra-abdominal infections with E. coli arise endogenously.

France

French ESBL surveys during the 1990s showed that an E. aerogenes clone with TEM-24 β-lactamase and several K. pneumoniae clones with SHV-4 enzyme had become widespread.2–4,38,39 The E. aerogenes clone is also widespread in Belgium, and has been found in Italy, Spain and Portugal, with the earliest representatives dating from 1988.40

One of the first CTX-M enzymes described, CTX-M-1, was discovered in France in 1989.41 A decade later CTX-M-3 was found in an Enterobacter cloacae isolate recovered in the suburbs of Paris, from a patient without overseas travel.42 Shortly afterwards, Dutour et al.43 reported Enterobacteriaceae with CTX-M-1, -3 and -14 enzymes from several Parisian hospitals, and demonstrated that the encoding plasmids were related to elements identified a decade earlier. Retrospectively examining enterobacteria recovered between 1989 and 2000 in three Parisian hospitals, Saladin et al.44 identified nine E. coli and Proteus mirabilis that produced CTX-M-1, -2, -9, -14, -20 or -21 enzymes. Despite these early discoveries, producers of CTX-M enzymes remained rare during the 1990s. A survey in Aquitaine (south-west France) in 1999 found ESBLs in only 1.5% of Enterobacteriaceae from private laboratories serving community and healthcare centre patients;45 the TEM-24⁺ clone of E. aerogenes was widely represented, but only one CTX-M enzyme producer was identified, an E. coli isolate with CTX-M-1.

As elsewhere in Europe, CTX-M enzymes are accumulating with the new century, especially in the north of France. An E. coli strain with CTX-M-15 β-lactamase caused an outbreak from October 2001 to March 2003 in a 35-bed long-term care facility in a suburb of Paris,46 involving 26/47 residents. By 2002–2003 E. coli was the main host species for ESBLs at multiple hospitals in and around Paris, with many of the producers clonally related, at least within sites, and with CTX-M-15 the dominant type.47–50 Other CTX-M types (CTX-M-3, -10 and -14) continued to be seen, as did non-CTX-M types, predominantly TEM-3 and, in E. aerogenes, TEM-24. Many of the isolates belonged to phylo-group B2, which is associated with extra-intestinal infections; they were variably resistant to aminoglycosides and tetracyclines. By contrast to these studies in northern France, surveys in 2002–2003 in the Auvergne (south-east France)51 and southern France52 found that E. aerogenes remained the most frequent ESBL producer, with TEM-3 and TEM-24 accounting for 90% of all ESBLs. This north–south divide may now be breaking down, and a 2004 study identified CTX-M enzymes in southern France, including in hospitals at Perpignan, Toulouse and Montpellier53 whereas further work shows that CTX-M-15 is widespread around Reims in the east (C. De Champs and P. Nordmann, unpublished data).

Whilst most work has concentrated on producer strains from hospitals and long-term care facilities, the community spread of E. coli with CTX-M-15 enzyme is illustrated by two reports, one of a Salmonella Typhimurium strain with CTX-M-1 enzyme and the other of a Shigella sonnei strain with CTX-M-15;48 neither patient had travelled abroad. S. enterica Virchow with CTX-M-9 enzyme has been identified in France from poultry as well as humans.54

Spain and Portugal

Reports of ESBL-producing Enterobacteriaceae in Spain date from 198855 and, as in other European countries, early publications concerned bla_SHV or bla_TEM variants.55–57 CTX-M β-lactamases were first reported in 2000, with the publication of papers describing E. coli and Salmonella with CTX-M-9 recovered in Barcelona and Murcia in 1996–9858,59 and an E. coli isolate with CTX-M-10 enzyme recovered in Madrid in 1997.60 Subsequent re-examination of ESBL-producing K. pneumoniae and Enterobacter collected in Madrid from 1988 onwards revealed that CTX-M enzymes had emerged in Spain at least by the early 1990s,5,61 only a few years after their first discovery in Germany, Japan and Argentina. The earliest producers found were K. pneumoniae from 1990—one with CTX-M-10 enzyme from an outpatient urine sample,61 the other, with CTX-M-9, from a gastroenterology inpatient; also an Enterobacter gergoviae with CTX-M-10 was recovered from a catheter urine in 1991.5 CTX-M-10 β-lactamase was also found in clinical E. coli isolates and in faecal Citrobacter freundii and K. pneumoniae isolates from outpatients from 1991.62,63

Despite these early examples, CTX-M β-lactamases did not become prevalent in Spain in the 1990s. A study in Barcelona between 1994 and 1996 found that only 7% of E. coli isolates with decreased susceptibilities to broad-spectrum cephalosporins had CTX-M-9-like enzymes64 and a long-term (1995–2003) surveillance of K. pneumoniae and E. coli in Seville first recorded CTX-M producers only from 1998.65 As in other European countries, this pattern has since changed dramatically, with multiple reports of CTX-M β-lactamase-producing isolates, most of them from community patients.62,65–70 One multicentre study examined all E. coli and K. pneumoniae isolates with an ESBL phenotype from March to June 2000 from 40 hospitals across Spain.67,68 ESBLs were found in 0.5% and 2.7% of E. coli and K. pneumoniae, respectively and it was notable that 90% of participating centres recovered producers. The most prevalent ESBLs in E. coli were CTX-M-9 (27.3%), SHV-12 (23.9%) and CTX-M-14 (16.7%). CTX-M-10 enzyme, though longer established, was present in only 4.5% of producer isolates. In the case of K. pneumoniae, CTX-M-10 (12.5%) was the only CTX-M enzyme type found, whilst TEM-4 (25%) and TEM-3 (16.7%) were the most common ESBLs. E. coli isolates with CTX-M-9 and -14 enzymes were encountered widely, whereas CTX-M-10 was concentrated in the central region, with just a few producers from the north.68,71 Fifty-one per cent of E. coli with ESBLs (but only 7% of the K. pneumoniae) were from non-hospitalized patients. There was little clonality among isolates with CTX-M enzymes,71 although a few strains broke this pattern, as in the case of an E. coli lineage with a unique CTX-M-9-encoding plasmid, observed in faecal samples of multiple attendees at a summer camp in 2002.72

The high prevalence of CTX-M-9 enzyme found in these large-scale surveys is corroborated by local studies of clinical and faecal E. coli, K. pneumoniae and Salmonella spp. isolates73 from hospitalized and community patients, and from healthy and sick chickens.74 CTX-M-14 enzyme, which also belongs to CTX-M group 9, also occurs widely,67 including from Salmonella spp.75 and in faecal E. coli from healthy volunteers with little or no hospital contact;75 it was the most frequent ESBL in a long-term study from Seville.65 Most recently, group 1 enzymes, including CTX-M-1, -3, -15, -28 and -32 have been encountered.23,70,74,76 As in the UK, Italy and France, CTX-M-15 seems to be spreading rapidly76 with producers also reported from Portugal;25 once again it is genetically linked with bla_OXA-1. CTX-M-1 and -32 enzymes were also recovered from healthy and sick animals in 2003.74 Group 2 enzymes, which are widely distributed in South America and Israel, have recently been detected in clinical E. coli isolates from Barcelona76 and in a faecal E. coli isolate from a healthy volunteer in Madrid;62 enzymes belonging to the CTX-M-8 and -25 groups have not yet been found in Spain.

Among hospitalized patients, prior oxyimino-β-lactam use, diabetes, and underlying diseases were independent risk factors for infection or colonization with non-clonal E. coli that had CTX-M enzymes, whereas previous fluoroquinolone use was associated with infection or colonization with those with SHV and TEM enzymes.77 Among non-hospitalized patients, most ESBL-positive E. coli isolates (64%) had bla_CTX-M-9 and were from patients with high scores for Charlson's co-morbidity index for severity of underlying disease.66 Many were old and reported fluoroquinolone use in the preceding 2 months. Most infections were of the urinary tract and almost half of the cases appeared truly community-acquired, as the patients lacked recent hospitalization and were not in nursing homes. Similar risk factors were reported for infections with ESBL producers outside the hospital in Israel, where CTX-M-2 predominates, although more patients had recent hospitalization.78

Dissemination of bla_CTX-M-9 is associated with a few large conjugative plasmids of long-established incompatibility groups; the gene is located within class 1 integrons with a low, but increasing, diversity of resistance cassettes79,80 often also conferring resistance to trimethoprim and aminoglycosides. Fluoroquinolone resistance is common among isolates with CTX-M-9 and -14, but is not associated with qnr (P. Nordmann and L. Poirel, unpublished data). Dissemination of epidemic plasmids carrying bla_CTX-M-14 or bla_CTX-M-32 among E. coli has been described,73 as was transmission of a plasmid coding for CTX-M-10 between E. cloacae and various E. coli and K. pneumoniae strains.5,61,81

Italy

CTX-M-type enzymes were not sought in the first nationwide Italian survey of ESBL production, carried out in 1999.82 However, only 8% of collected Enterobacteriaceae isolates with ESBL phenotypes lacked TEM- and/or SHV-ESBLs, implying that any diffusion of CTX-M-type enzymes was limited.

Reports of Italian isolates with CTX-M ESBLs only began to appear in 2003, and then included: (i) several E. coli, E. aerogenes and C. freundii isolates with CTX-M-1 enzyme, but with no information on sources or clonal relationships;83 (ii) a few isolates with CTX-M enzymes (one K. pneumoniae with CTX-M-15; two related Proteus vulgaris isolates with CTX-M-2, and three unrelated E. coli isolates with CTX-M-1) from a hospital in northern Italy.84 In 2001–2002, only 2.6% of all ESBL producers had CTX-M phenotypes, with cefotaxime MIC > ceftazidime MIC.84

A second nationwide ESBL survey, in 2003, sought CTX-M enzymes specifically, and found them to be widespread85 Producers—mostly E. coli and, in lower numbers, K. pneumoniae—were detected in 10/11 participating centres and accounted for ∼20% of all ESBL-producing isolates, though this proportion varied from <2% to ∼50%. All the CTX-M enzymes were of group 1, with CTX-M-1 and -15 predominant and CTX-M-32 rarer. Producers were detected from inpatients and outpatients at similar rates. Genotyping revealed multiple lineages of E. coli and K. pneumoniae irrespective of whether the enzyme present was CTX-M-1 or -15. This diversity, and the ability to transfer ESBL determinants conjugatively, was greater among E. coli isolates with CTX-M-1 enzyme than among those with CTX-M-15.86

The rapid and recent dissemination of CTX-M-β-lactamase-producing E. coli was further documented by a longitudinal surveillance examining 20 000 isolates collected over 5 years in northern Italy.87 The prevalence of ESBLs increased steadily from 0.2% in 1999 to 1.6% in 2003, with this change largely attributable to CTX-M-β-lactamase-positive isolates, which comprised 38% of all ESBL producers in 2003 compared with 12% in 1999. Most had CTX-M-1 enzyme, which was also found in Citrobacter amalonaticus and Morganella morganii—unusual hosts for ESBLs. In both instances the isolates were from patients co-infected with CTX-M-1⁺E. coli, suggesting in vivo transfer.88

CTX-M-1 β-lactamase was present in 76% of ESBL-positive E. coli isolates from sick and healthy companion or stray dogs and cats, based on a survey carried out in Rome from 2001 to 2003. Perhaps even more strikingly this study found ESBL producers in specimens (mostly faeces) from 21 of the 298 animals examined.89

Poland

It is impossible to determine when CTX-M β-lactamase producers first appeared in Poland and whether they first emerged in nosocomial pathogens or in the community because, until the mid-1990s, Polish laboratories did not screen for ESBLs and there were no epidemiological surveys of ESBLs. ESBL surveys only commenced in 1996, when the then Sera and Vaccines Central Research Laboratory also began to promulgate the value and methodology of ESBL detection to clinical laboratories.

Among the first ESBL producers detected and analysed there were four clonal C. freundii isolates and one E. coli isolate from the Praski Hospital in Warsaw, one of the first centres to introduce routine ESBL detection.90 A then novel CTX-M variant, CTX-M-3, was identified in these isolates, and proved similar to CTX-M-1, previously observed in Germany and France.91,92,bla_CTX-M-3 was located on a readily conjugative plasmid. Subsequent examination, from November 1996 to February 1997, found that CTX-M-3 was the predominant ESBL in the Praski Hospital,93 with producers recovered in many wards. Their prevalence reflected clonal spread of K. pneumoniae, C. freundii and Serratia marcescens producers, along with intense plasmid dissemination. The same bla_CTX-M-3-encoding plasmid, originally identified in the C. freundii isolates from July 1996 proved widespread, and was designated plasmid A1, later pCTX-M3.90 Almost all the other encoding plasmids found were its derivatives. Hosts of pCTX-M3 family plasmids included 16 RAPD types of seven enterobacterial species (E. coli, K. pneumoniae, K. oxytoca, C. freundii, E. cloacae, S. marcescens and M. morganii). Several of the E. coli and K. pneumoniae isolates were from outpatients, though they may have had earlier hospitalization.

The first multicentre survey on ESBLs among Enterobacteriaceae in Polish hospitals was conducted in Spring 1998 and covered seven institutions (M. Gniadkowski, A. Barnaiak, J. Fiett and W. Hryniewicz, unpublished data). Isolates with CTX-M enzymes were found at six centres, and accounted for 19% of all ESBL producers collected. This frequency was below that of SHV ESBLs (60.4%, mainly SHV-5), but comparable to that of TEM ESBLs (20.8%). All the survey isolates with CTX-M enzymes, together with those from 11 other hospitals up to the end of 2000 (89 isolates altogether), were subjected to detailed analysis.94–96 Eighty-six had CTX-M-3 β-lactamase whereas three had its variant, CTX-M-15. In the vast majority of cases these enzymes were coded by conjugative plasmids of the pCTX-M3 family.97,98 A more recent (2003) survey covering 13 regional Polish hospitals further underscored the rise of CTX-M enzymes, which were found in 82% of all 264 ESBL-producing isolates (M. Gniadkowski, J. Empel, A. Barnaiak, J. Fiett, E. Literacka, A. Mrówka and W. Hryniewicz, unpublished data). Once again, the enzymes found were mostly CTX-M-3 or, less often, CTX-M-15 types and producers were strongly represented at all the participating centres, varying from 62.5% to 100% of all ESBL producers. CTX-M enzymes were found in even more species than previously, including E. aerogenes, P. mirabilis and Providencia rettgeri.

pCTX-M3 is an IncL/M, conjugative, broad-host-range plasmid (I. Kern-Zdanowicz, M. Golębiewski, M. Zienkiewicz, M. Adamczyk, M. Gniadkowski and P. Ceglowski, unpublished data) with a molecular mass of 89 468 bp (GenBank accession number AF550415). Apart from bla_CTX-M-3, it carries bla_TEM-1, dfr (dihydrofolate reductase), aac(3)-II and aadA2 (aminoglycoside-modifying enzyme) genes; dfr and aadA2 lie within a class 1 integron, which may be part of a composite transposon formed by two IS26 elements. bla_CTX-M-3 is accompanied by an ISEcp1 element, 128 bp upstream from the coding region (ref. 96 and M. Gniadkowski and A. Baraniak, unpublished data), which seems likely to have been involved in the initial mobilization.19 Recently, a 1377 bp fragment of pCTX-M3, starting just behind ISEcp1 and encompassing bla_CTX-M-3 along with most of orf477, was found to be very similar to a chromosomal DNA fragment from a K. ascorbata strain,17 suggesting a possible origin. bla_CTX-M-15 genes from Polish isolates were similarly spaced from ISEcp1 as for bla_CTX-M-3 and had the same flanking sequences;96 moreover, they are located in pCTX-M3-type plasmids, implying that they had emerged by point mutation of bla_CTX-M-3, not via separate gene escapes.

In summary, the ESBL epidemiology in Poland is dominated by CTX-M producers, which have occurred for at least 10 years and are now widespread in hospitals and among bacterial species. The pool of bla_CTX-M variants is very homogeneous, with bla_CTX-M-3 dominant and, although a few isolates have bla_CTX-M-15, these probably are direct descendants of those with bla_CTX-M-3. Almost all the bla_CTX-M-3 genes seem to have arisen from a single mobilization event, with subsequent intense dissemination of plasmid pCTX-M3 and its derivatives. Recent unpublished data suggest spread in the community, though studies are limited.

Elsewhere in Europe

This article has concentrated on France, Italy, Poland, Spain and the UK, but the rise of CTX-M enzymes has occurred more widely. National spread of a K. pneumoniae clone with CTX-M-15 has recently been reported in Hungary99 whilst there is good evidence that E. coli and K. pneumoniae with CTX-M enzymes are becoming widely scattered, if not frequent, in many other European countries (Figure 2).25,99–105 It is notable that E. coli with these enzymes are causing concern even in Scandinavian countries with an exemplary record for infection control.106 CTX-M-15 is the most-reported CTX-M type across most of the further countries for which data are available, with CTX-M-1 and -3 also widely reported (Figure 2). Exceptionally, one Belgian study found that CTX-M-2 accounted for 14% of all CTX-M types.104

Figure 2

Dominant and frequent CTX-M β-lactamase types in different European countries. Data for France, Italy, Poland, Spain and the UK are based on the references cited in the text. Those for other countries are based on further reports, 25,99–105 also, for Croatia, B. Bedenic and M. Gniadkowski., unpublished data and, for Switzerland, A. Wenger and P. Nordmann, unpublished data, with national dominance only inferred based on large-scale surveys. Many other CTX-M enzymes have been isolated on a few occasions or (as with CTX-M-5-producing Salmonella Typhimurium in Latvia) have caused sizeable but time-defined outbreaks before apparently declining. CTX-M-type enzymes are emerging rapidly also in Greece but reports that stress prevalence do not identify the particular enzyme types.

The best continent-wide resistance surveillance (Author Webpage) does not distinguish ESBL producers from those with other modes of cephalosporin resistance, and has limited data for Enterobacteriaceae other than E. coli. Nevertheless, it shows widespread rises in cephalosporin resistance in E. coli from 2001 to 2005 (Figure 3); along with even more dramatic increases in fluoroquinolone resistance (see website).

Figure 3

Prevalence of resistance to extended-spectrum cephalosporins among E. coli isolates from bacteraemias, as reported under the European Antimicrobial Resistance Surveillance System (Author Webpage) in 2001 and 2004, supplemented with UK data based on laboratories' reports to the Health Protection Agency.

Routes of spread

Relationships among strains with CTX-M enzymes from different countries remain largely unclear although, prima facie, it is likely that the continent-wide rise of CTX-M ESBLs reflects a series of independent events.

It seems that the pCTX-M3-type plasmids, encoding CTX-M-3, began to spread in Poland a little before related enzymes, predominantly CTX-M-15, became prevalent in western Europe. It therefore is reasonable to ask, whether increased movement between east and west Europe has allowed this ‘Polish’ plasmid to spread westwards. Some insight here can be gained by examining the length of the spacer region between ISEcp1 and bla_CTX-M-3/15. On this basis it seems most likely that bla_CTX-M-15 genes in isolates from the UK (E. Karisik, D.M. Livermore and N. Woodford, unpublished data), Boulogne49 and Lisbon107—also New Delhi,108 Istanbul,109 Toronto13 and Yaounde (Cameroon)110—emerged independently from those observed in Poland, since the ISEcp1 elements are located much closer (48 bp) to bla_CTX-M. On the other hand, the ISEcp1-associated bla_CTX-M-3 gene from an E. cloacae from Versailles (France) was located on a pCTX-M3-type plasmid.43 Similarity was also suggested between CTX-M-3 producers from Poland and Taiwan.111 A representative of the UK E. coli strain A was found in Austria, but without any obvious epidemiological link to the UK.103

It is plausible too that the E. coli clones with CTX-M-15 enzyme, now spreading in the UK and France, originated in other countries with which Europe has extensive and growing population exchanges, perhaps in India, where CTX-M-15 was first described,22 or North Africa.112 Clonal relationships have recently been noted between E. coli with CTX-M-15 enzymes from Paris, Tunis and, to a lesser extent, the Central African Republic.113 Such data do not, however, prove the direction of spread and, in the UK, the early spread of CTX-M-15 was not centred in conurbations with large immigrant populations.30 Moreover—if migration is proposed as a route of spread—it remains unclear why group 9 enzymes (mainly CTX-M-9 itself) currently predominate in Spain whereas the rest of Europe largely has group 1 (CTX-M-3 and -15) types (Figure 3). Group 9 types (CTX-M-9 or -14) are otherwise dominant in East Asia, notably China, whereas South America—with substantial population flows to and from Spain—predominantly has CTX-M-2 type enzymes, at least in its Southern ‘cone’. Within Eurasia, CTX-M-2 appears frequent only in Israel114 a country with fewer obvious cultural and migrational links than Spain to South America.

Food is another possible vector of spread and this article has repeatedly noted the detection of E. coli with CTX-M enzyme in food animals. Nevertheless, there is no proven link between this carriage and human infections. In the UK, at least, the enzyme types found in both food animals and in foodstuff belong to groups 2 and 9 (as in Spain74) whereas the huge majority of clinical producers have CTX-M-15.

Routes of transmission of Klebsiella spp. with ESBLs are well understood in the context of nosocomial outbreaks and whilst nosocomial transmission of E. coli clones is less common it can occur. What is less clear is the route by which community infections arise. On one hand, many patients with ‘community’ infections with CTX-M-β-lactamase-producing E. coli have a history of recent hospitalization,78 where they may have been colonized. In support of this view the distribution of enzyme producers among urinary infections is biased towards complex cases and older age groups,66 whereas a more random distribution, with more producers from uncomplicated cystitis, would be predicted if most acquisition was via the food chain in the community. On the other hand, not all colonized individuals have a history of hospitalization and it may be that low-level gut colonization occurs in the community, via the food chain, perhaps with plasmid transfer to resident E. coli, and that the proportion of resistant E. coli with CTX-M enzymes tends to be enriched during healthcare contacts, owing to frequent antimicrobial exposure. Screening populations for faecal or rectal carriage would be the obvious way to resolve these issues, but has not yet been undertaken on any wide scale.

Whatever the precise mode of spread (and spread through plasmid dissemination, hospital cross infection, down the food chain, and via human migration are not mutually exclusive), it is clear that Europe is moving to a situation where ESBLs are more common and less confined to hospitals, as well as one where E. coli is a major host species. If so, the opportunities for control are disturbingly small. Old data show that ampicillin resistance, largely due to TEM-1 enzyme, rose first in E. coli from hospital-acquired infections and only later among those from community-acquired cases.115 History may be repeating itself with CTX-M ESBLs.

Transparency declarations

Conflicts of interest: D. M. L., grants from AstraZeneca, Merck and Wyeth; speakers' bureaux for AstraZeneca, Johnson & Johnson, Merck and Wyeth; shareholdings in AstraZeneca GlaxoSmithkline, Pfizer and Schering Plough. G. M. R., grants from Wyeth; consultancies from Wyeth, Janssen-Cilag and Nabi Pharmaceuticals; speakers' bureaux for Wyeth and Merck; R. C., grants from Sanofi-Aventis and Wyeth. Other authors: no conflicts to declare.

The work of all co-authors on emerging β-lactamase problems is supported, in part, by the EU/FP6-funded COBRA project (6-PCRD LSHM-CT-2003-503-335). We are grateful to EARSS for permission to include Figure 3.