THE CONTINUING CHALLENGE OF ESBLS

Summary of recent advances

Since their first description more than twenty years ago, Escherichia coli and Klebsiella pneumoniae possessing extended-spectrum class A beta-lactamases (ESBLs) continue to thwart our best clinical efforts. In the “early years” the most common beta-lactamases were of the TEM and SHV varieties. Now, CTX-M enzymes are being discovered though out the world and are becoming the most prevalent beta-lactamases found in clinical isolates. The Klebsiella pneumoniae carbapenemases (KPC) (ESBL type enzymes that confer resistance to extended spectrum cephalosporins and carbapenems) present the most significant challenge to date. Structural studies of ESBLs indicate that active site expansion and remodeling are responsible for this extended hydrolytic activity. Continuing questions still exist regarding the optimal detection method for ESBLs. Most relevant are the increasing concerns regarding the status of carbapenems as “best therapy” for ESBL producing bacteria in light of the emergence of carbapenemases.

Introduction

The development of extended-spectrum cephalosporins in the early 1980s was regarded as a major addition to our therapeutic armamentarium in the fight against beta-lactamase-mediated bacterial resistance [1–3]. Regrettably, the emergence of Escherichia coli and Klebsiella pneumoniae resistant to ceftazidime and other cephalosporins seriously compromised the efficacy of these life-saving antibiotics. The new bacterial beta-lactamases present in these common enteric bacilli (the parent TEM-1 and SHV-1 enzymes) demonstrated unique hydrolytic properties. Point mutations in the bla_SHV and bla_TEM genes that resulted in single amino acid changes (Gly238→Ser, Glu240→Lys, Arg164→Ser, Arg164→His, Asp179→Asn, and Glu(Asp)104→Lys) formed the basis of this remarkable resistance phenotype [4,5]. Currently, ESBLs are becoming a major threat for patients in the hospital, long-term care facilities, and community. It is our goal in this analysis to:

1. Review the classification and global epidemiology of ESBL-producing enteric bacteria; highlighting the emergence of community-acquired ESBLs and KPC enzymes as the next major public health dangers.

2. Summarize the atomic features of ESBLs so the reader may appreciate “structure-function relationships” in these enzymes;

3. Present the current “state of the art” regarding the laboratory detection of ESBLs.

4. Discuss the clinical impact and current therapy of infections caused by ESBL-producing E. coli, Proteus spp. and Klebsiella spp. Although Enterobacter spp. are not routinely tested for ESBL production, our discussion will include mention of this genus when appropriate.

What is an ESBL and why are they important?

Beta-lactamases are among the most heterogeneous group of resistance enzymes. More than 700 distinct beta-lactamases are described (personal communication, Dr. Karen Bush). These globular proteins are composed of alpha–helices and beta–pleated sheets [6]. Despite a significant amount of amino acid sequence variability, beta-lactamases share a common overall topology (Figure 1).

Figure 1.

In general, ESBL are capable of hydrolyzing penicillins (e.g, ampicillin and piperacillin), cephalosporins of the first-, second-, third- and fourth-generations, and the monobactam aztreonam (but not the cephamycins or carbapenems) [7,8]. In contrast, ESBLs (particularly TEM and SHV family derivatives) are readily inhibited by the commercially available beta-lactamase inhibitors (i.e., clavulanic acid, tazobactam, or sulbactam). This unique property serves as an important phenotypic test that is conveniently exploited to identify ESBLs in bacteria [vide infra].

Two systems are commonly used to classify beta-lactamases: the Ambler scheme and the Bush-Medeiros-Jacoby system [7,9]. They are summarized in Table 1. Both systems are used interchangeably in the literature: ESBLs belong to group 2be in the Bush-Medeiros-Jacoby system and to class A in the Ambler system.

Table 1.

Classification of beta-lactamases.

Bush-Jacoby-Medeiros system	Major subgroups	Ambler System	Main attributes
Group 1 cephalosporinases	–	C (cephalosporinases)	Usually chromosomal; Resistance to all β-lactams except carbapenems; Not inhibited by clavulanate
Group 2 penicillinases	2a	A (serine β-lactamases)	Staphylococcal penicillinases
(clavulanic acid susceptible)	2b	A	Broad-spectrum – TEM-1, TEM-2, SHV-1
	2be	A	Extended-spectrum – TEM-3–160, SHV-2–101
	2br	A	Inhibitor resistant TEM (IRT)
	2c	A	Carbenicillin-hydrolyzing
	2e	A	Cephalosporinases inhibited by clavulanate
	2f	A	Carbapenemases inhibited by clavulanate
	2d	D(oxacillin-hydrolyzing)	Cloxacillin-hydrolyzing (OXA)
Group 3 metallo-b-	3a	B (metalloenzymes)	Zinc-dependent
lactamase			carbapenemases
	3b	B
	3c	B
Group 4		Not classified	Miscellaneous enzymes, most not yet sequenced

Rapid sequencing methods are enabling investigators the opportunity to identify many TEM and SHV family variants (see: www.lahey.org) as well as “non-TEM, non-SHV” type ESBLs that are relatively resistant to beta-lactamase inhibitors.

Global epidemiology of Enterobacteriaceae-producing ESBLs

First described in Germany (1983) and France (1985) among Klebsiella spp, ESBLs exist in every region of the world and in most genera of enterobacteria [8,10]. A contemporary perspective on the epidemiology of ESBLs reveals the widespread emergence of the CTX-M-type. These are now considered the most prevalent ESBLs worldwide [11–13]. PER- and OXA-type enzymes are more common in P. aeruginosa and Acinetobacter spp., but there have been sporadic reports of PER type ESBLs in Enterobacteriacea originating from France, Argentina and Italy [14–16]. Of tremendous import because of their ability to inactivate carbapenems, plasmid-borne KPC enzymes are emerging among K. pneumoniae and other enterobacteria in the Eastern seabord and the heartland of the U.S. and internationally [17]. Table 2 summarizes current global trends in beta-lactam resistance among enterobacteria.

Table 2.

Resistance rates of selected enterobacteria to third generation cephalosporins and carbapenems. Data from MYSTIC (www.mystic-data.org).

Country	Year	Resistance rate in percentage (Total number of isolates tested)
		E. coli		Klebsiella spp.
		†3GC	††Carbapenems	3GC	Carbapenems
United States	2000	2	0	6.6	2.1
		(312)	(312)	(233)	(233)
	2006	7.2	0	16.5	9
		(640)	(312)	(619)	(619)
Canada	2000	15	0	0	0
		(33)	(33)	(17)	(35)
	2006	31.5	0	20.8	1
		(73)	(73)	(120)	(120)
Mexico	2000	31	0	15.4	0
		(39)	(39)	(52)	(52)
	2006	72	1	42.1	0
		(108)	(108)	(76)	(76)
Brazil	2000	8	0	54	0
		(24)	(24)	(24)	(24)
	2006	16.5	0	54	0.6
		(139)	(139)	(170)	(170)
Australia	2000	8	0	0	0
		(13)	(13)	(19)	(19)
	2006	0	0	4	0
		(45)	(45)	(55)	(55)
Israel	2000	15.8	0	43.5	0
		(19)	(19)	(23)	(23)
	2006	21	5	56.5	13
		(19)	(19)	(23)	(23)
Turkey	2000	27.6	1.5	61	2
		(203)	(203)	(194)	(194)
	2006	45.2	0.4	54.7	4
		(272)	(272)	(245)	(246)
Russia	2000	43	0	84	0
		(16)	(16)	(19)	(19)
	2006	20	0	85	0
		(20)	(20)	(20)	(20)
Poland	2000	47.4	0	95	0
		(19)	(19)	(22)	(22)
	2006	30	0	35	0
		(20)	(20)	(40)	(40)
Czech Republic	2000	0	0	39	0
		(19)	(19)	(18)	(18)
	2006	5	0	40	0
		(20)	(20)	(20)	(20)
Belgium	2000	5	0	19.3	0
		(166)	(166)	(140)	(140)
	2006	10.4	0	21	0
		(182)	(182)	(196)	(196)
Germany	2000	3	0	8.2	0
		(134)	(134)	(102)	(102)
	2006	13.2	1	16.2	1.3
		(91)	(91)	(74)	(74)
Sweden	2000	0	0	0	0
		(20)	(20)	(10)	(10)
	2006	2.7	0	1.2	1.2
		(148)	(148)	(82)	(82)
United Kingdom	2000	6.5	0	23.6	0
		(92)	(92)	(72)	(72)
	2006	15	1.5	17	0
		(67)	(67)	(47)	(47)
Spain	2000	3.7	0	1.8	0
		(63)	(63)	(56)	(56)
	2006	16	0	12.4	0
		(113)	(111)	(137)	(137)

^†3^rd Generation Cephalosporins (e.g., cefotaxime, ceftazidime, ceftriaxone and aztreonam). Resistant phenotype defined as MIC ≥2 µg/ml.

^††Carbapenems (i.e., ertapenem, meropenem and ertapenem). Resistant phenotype defined as MIC ≥8 µg/ml.

The emerging threat of CTX-M beta-lactamases in the community

A recent development after the year 2000 is the identification of infections caused by bacteria harboring ESBLs in community dwellers. These are typically urinary tract infections (UTIs) caused by E. coli expressing CTX-M. These E. coli are also resistant to quinolones, aminoglycosides, and sulfonamides. In the Calgary Health Region of Canada, Pitout et al described the clonal spread of two closely related strains harboring CTX-M-14, isolated most often from urine samples [18,19]. Nationwide surveillance carried out in the United Kingdom saw the rise of several types of ESBLs in the community, among which clonally-spread E. coli carrying CTX-M-15 predominated, although CTX-M 9 was also represented [20].

Similarly, CTX-M-15 producing isolates appear with increased in frequency among clonally related E. coli strains across Italy, Portugal and France [21–24]. In Spain, the clonal spread of E. coli harboring CTX-M-15 has occurred as well, but the situation appears more complex [25]. Rodriguez-Bano et al, [26], described 49 patients in the region of Sevilla with community-acquired ESBL producing E. coli, 64% of which carried CTX-M-9. All the strains were clonally unrelated. A larger national study confirmed the high prevalence of CTX-M-9 and CTX-M-14, without evidence of clonal dissemination [27]. Further investigation suggested the dissemination of closely related plasmids harboring CTX-M-14 among clonally unrelated E. coli isolates from the community [28].

Outside of Europe, the occurrence of CTX-M ESBLs in the community has been reported as well. In Hong-Kong, 42/600 (7%) of community isolates of urinary E. coli were ESBL producers [29]. A low prevalence (1.8%) and great diversity of enzymes and bacterial species were found among community isolates in Brazil, in contrast with Bolivian and Peruvian isolates among which E. coli harboring CTX-M-15 and CTX-M-9 are predominant [30,31].

The initial observation of infections caused by bacteria harboring ESBLs in hospitals would suggest that CTX-Ms arose in the nosocomial setting and spread to the community. In fact, hospitals are the arena where the selective pressure of broad-spectrum antimicrobials and suboptimal infection control practices best conspire to foster the emergence and transmission of multidrug-resistant organisms. Nursing homes, in turn, may serve as reservoirs from which colonized and infected patients transfer to the community or back to the hospitals [8,32]. Prior to the rise of CTX-M, such a model seemed to have corresponded with the available data [33]. Even after the noted spread of bacteria producing CTX-M into the community, recent hospitalization, along with age and exposure to cephalosporins and/or quinolones, have consistently been identified as risk factors for infection with these organisms [8,34,35]. Furthermore, the above mentioned reports from Spain, Portugal and the United Kingdom demonstrate the clonal dissemination of E. coli producing CTX-M-15 in the community, with the concurrent spread of these clones into hospitals and long term care facilities [23,25,36].

Is it possible that there may be dissemination of ESBL harboring organisms from the community into hospitals? A recent report from Israel described high rates of patients with bacteremia and colonization with ESBL-producing Enterobacteriaceae on admission to the hospital [37]. Clinicians are already facing the tremendous challenges posed by this situation, in terms of the detection and isolation of patients to prevent further nosocomial expansion, and of the choice of empiric antibiotic therapy [38–40]. An additional potential reservoir of resistant bacteria and genetic determinants of resistance which intersects with the community is the food supply, as illustrated by the finding of diverse ESBL-producing bacteria, including CTX-M-15, in poultry and other farm animals [41–43]. Finally, the origin of CTX-M enzymes probably lies in beta-lactamases found in environmental species, like Kluyvera spp. [12], further supporting the notion of a community reservoir for these enzymes. The existence of resistant microorganisms predating the clinical use of antibiotics has given rise to the concept of the “antibiotic resistome”[44,45].

Structural properties of ESBLs

Important insights have emerged from the study of the atomic structures of class A ESBL enzymes. Overall, one appreciates that the first common theme to emerge is that the active site is selectively “remodeled and expanded” to accommodate the bulky R₁ side chain of extended-spectrum cephalosporins [6]. This remodeling is observed in the atomic structures of Toho-1, TEM-52, TEM-64, the Gly238Ala ESBL in TEM, SHV-2, PER-1 and OXA-10 beta-lactamases [46–52]. Recently, the crystal structure of KPC-2 was determined, demonstrating modifications in the active site that allow access to carbapenems [53]. Understanding the three dimensional structure of these enzymes is essential for the future development of beta-lactams as pharmacological agents [54]. Our attention here will focus on the analysis of CTX-M ESBLs.

CTX-M ESBLs

Among the non-TEM, non-SHV ESBLs, the CTX-M beta-lactamases are the most prevalent. They can be divided into distinct clusters (see www.lahey.org). Unlike most (but not all) TEM-and SHV-derived ESBLs, CTX-M betalactamases hydrolyze cefotaxime and ceftriaxone better than they do ceftazidime. It also appears that CTX-M enzymes are more readily inhibited by tazobactam than they are by clavulanic acid.

CTX-M beta-lactamases are commonly found in K. pneumoniae, E. coli, thyphoidal and nonthyphoidal Salmonella, Shigella, Citrobacter freundii, Enterobacter, spp., and Serratia marcescens [12]. Of note, different genetic elements may be involved in the mobilization of bla_CTX-M genes. Plasmids and Insertion Sequences (e.g., ISEcp1 or ISEcp1-like insertion sequences) have repeatedly been observed upstream of ORFs encoding the CTX-M-1, CTX-M-2, CTX-M-3, CTX-M-9, CTX-M-13, CTX-M-14, CTX-M-15, CTX-M-17, CTX-M-19, CTX-M-20, and CTX-M-21 enzymes.

From comparative sequence analyses, modeling, and mutagenesis studies, key amino acids involved in the substrate specificity for hydrolysis in the CTX-M beta-lactamases are idenitifed: Asn104, Asn132, Ser237, and Asp240 [12]. Unlike what is seen with TEM and SHV (enlargement of the active site), the crystal structures of CTX-M-14, CTX-M-27, CTX-M-9 and CTX-M-16 indicate that point substitutions leading to specific interactions may be responsible for the improved activity against ceftazidime and cefotaxime, rather than active site expansion.

In general, the remodeling of a “broad spectrum” beta-lactamase to an extended spectrum enzyme comes at a “price”. Many of these TEM and SHV family ESBLs are not as catalytically efficient as wild-type progenitors against natural substrates (especially against penicillins). One uniformly observes a decrease in k_cat/K_m ratio and decrease in penicillin MICs. In addition, these enzymes are less stable [55–58]. It is sobering to ponder all the amino acid changes that are permitted in a beta-lactamase that permit the evolution of the ESBL phenotype [59–61]. The implications for the novel design of future generations of cephalosporins are far reaching.

Detection of ESBLs

As ESBLs are found though out the world, numerous detection strategies have been developed. According to the Clinical Laboratory Standard Institute (CLSI) criteria, enterobacterial resistance to ceftriaxone, cefotaxime, ceftazidime, cefepime, and aztreonam is defined by MICs ≥ 16 µg/ml [62]. However, since several ESBL producers have MIC values for extended-spectrum cephalosporins and aztreonam below the standard breakpoints for resistance (e.g., between 2 and 8 µg/ml), the real prevalence of these organisms may be unappreciated.

Highly revealing studies performed in the United States and Europe by Tenover et al and Livermore et al, respectively, reported that errors in the detection of ESBL mediated resistance are frequently encountered with both automated and disk diffusion methods [63,64]. A contemporary analysis from Italy involving 60 independent clinical microbiology laboratories showed that only 25/60 (41.7%) correctly recognized and reported all the 5 ESBL-producing enterobacteria presented for analysis. Nearly 56% of the isolates were incorrectly characterized when testing cephalosporins and aztreonam; this was most worrisome in the case of CTX-M-1-producing E. coli and TEM-52-producing P. mirabilis [65].

Since the inaccurate identification of ESBL producers bears important clinical implications for antibiotic therapy and infection control measures, specific reporting guidelines are issued [[62] Health Protection Agency 2005, SFM 2007]. For all confirmed ESBL producers, the general consensus states that ESBL producers should be reported as resistant to all penicillins, cephalosporins (except for the cephamycins cefoxitin and cefotetan), and aztreonam irrespective of routine antimicrobial susceptibility results [62]. Notably, beta-lactam/beta-lactamase inhibitor combinations (e.g., piperacillin-tazobactam, amoxicillin-clavulanate, and ampicillin-sulbactam) are not affected by this rule and should be reported as obtained during routine susceptibility tests.

Screening for ESBL production

Both broth dilution and disk diffusion methods for screening for ESBL producers are advised by CLSI. It is recommended that E. coli, K. pneumoniae and K. oxytoca strains with MIC ≥ 8 µg/ml for cefpodoxime or MICs ≥ 2 µg/ml against ceftazidime, cefotaxime, ceftriaxone, or aztreonam should be investigated using specific phenotypic confirmatory tests for ESBL production. For P. mirabilis isolates, confirmatory tests should be performed if strains demonstrate MICs ≥ 2 µg/ml for ceftazidime, cefotaxime or cefpodoxime. The use of more than one of the above agents for screening improves the sensitivity of ESBL detection.

We recommend that if laboratories are using the disk diffusion method, E. coli, K. pneumoniae, K. oxytoca and P. mirabilis with a zone inhibition diameter lower than the following values should be investigated with confirmatory tests: ceftazidime (≤ 22 mm), cefotaxime and aztreonam (≤ 27 mm), ceftriaxone (≤ 25 mm). In the case of cefpodoxime, the “cutoff” for P. mirabilis is ≤ 22 mm, whereas in the remaining three species cited above it is ≤ 17 mm [62].

Criteria for screening for ESBL production in other Enterobacteriaceae have not been established by the CLSI [62]. In contrast, the British Society for Antimicrobial Chemotherapy (BSAC) maintains that all Enterobacteriaceae resistant to ceftazidime (MIC ≥ 4 µg/ml or zone inhibition ≤ 21 mm for E. coli and Klebiella spp. and ≤ 27 mm for the remaining species), cefotaxime (MIC ≥ 2 µg/ml or zone inhibition ≤ 29 mm), or cefpodoxime (MIC ≥ 2 µg/ml or zone inhibition ≤ 19 mm) should be evaluated by the ESBL confirmatory tests (Health Protection Agency, 2005). Similarly, the Société Française de Microbiologie (SFM) suggests the evaluation of all enterobacteria with the confirmatory test. However, criteria used to detect an ESBL producer are not closely related with MICs and zone diameter inhibition of cephalosporins and aztreonam. In addition, the ESBL confirmatory test should be performed when an isolate shows resistance to aminoglycosides (Comite de L’Antibiogramme de la Société Française de Microbiologie, SFM, 2007), since bla genes encoding ESBLs are frequently found in the same plasmid that encode resistance determinants to other classes of antibiotics such as aminoglycosides, tetracycline and sulfonamides [10].

Phenotypic confirmatory tests for ESBL production

According to the CLSI, ESBL confirmatory testing based upon phenotype requires the use of both ceftazidime and cefotaxime alone and in combination with clavulanate [62]. Several commercial vendors offer disks for use in this application (Oxoid, Becton-Dickinson and Mast). Typically, these discs contain 30 µg /disk of ceftazidime, cefotaxime with or without clavulanate (10 µg/disk). These discs have been rigorously tested and developed and are reported to have sensitivity and specificity of greater than 95% [66,67].

As mentioned above, screening and confirmatory tests for other Enterobacteriaceae, including those producing inducible AmpC chromosomal enzymes (e.g., Enterobacter spp., M. morganii, Providencia spp., Citrobacter freundii, and Serratia marcescens), have not yet been established by the CLSI [62]. Not detecting ESBLs in these pathogens may have a significant adverse impact on patients, especially those treated with extended-spectrum cephalosporins [68,69]. ESBLs are more difficult to detect in these organisms because AmpC enzymes may be induced by clavulanate (which inhibits them poorly) and may increase resistance to cephalosporins, overcoming the synergy arising from inhibition of the ESBL [70]. Therefore, the application of CLSI guidelines for ESBL detection in this group of enterobacteria is discouraged [71].

Paradoxically, the use of cefepime and cefpirome with clavulanate might improve the ability to detect ESBLs [72,73]. For this reason, the BSAC advises the use of cefpirome/clavulanate combination disks (in addition to cefpodoxime/clavulanate) for Enterobacter spp. and C. freundii as a confirmatory test (Health Protection Agency, 2005), whereas the SFM recommends to use cefepime or cefpirome with the double-disk diffusion test for all enterobacteria producing AmpC enzymes (CASFM, 2007).

Commercial methods for ESBL detection

The Etest method produced by AB Biodisk (Solna, Sweden) and Bio-Stat (Stockport, UK) is a plastic drug-impregnated strip, one end of which generates a stable concentration gradient of cephalosporin (i.e., ceftazidime 0.5–32 µg/ml, cefotaxime and cefepime 0.25–16 µg/ml) and the remaining end of which generates a gradient of cephalosporin (i.e., ceftazidime and cefepime 0.064–4 µg/ml, cefotaxime 0.016–1 µg/ml) plus a constant concentration of clavulanate (4 µg/ml). ESBL production is inferred if the MIC ratio for cephalosporin alone/cephalosporin plus clavulanate MIC is ≥ 8 (Health Protection Agency, 2005). Accurate and precise (but more expensive than combination disks) these tests are suggested by the BSAC as a confirmatory test for ESBL production. Significantly, BASC recommends the cefepime/clavulanate Etest for Enterobacter spp. (Health Protection Agency, 2005). In fact, neither ceftazidime/clavulanate nor cefotaxime/clavulanate Etest strips are able to detect the ESBLs in Enterobacter spp. In contrast, the cefepime/clavulanate strip is the only commercially available and highly reliable test that permits accurate detection of ESBLs within this group of organisms [73].

Automated method for bacterial identification and susceptibility testing are used in the detection of ESBL producing organisms. The BD Phoenix System (Becton-Dickinson Biosciences, Sparks, MD) uses its "expert software" to interpret the growth response to ceftazidime, cefotaxime, ceftriaxone and cefpodoxime, with or without clavulanate. Similarly, the Vitek 2 System (bioMérieux, Marcy L’Etoile, France) uses a "card" containing ceftazidime and cefotaxime alone and in combination with clavulanate. Ceftazidime or cefotaxime plus betalactamase inhibitors are also used in the MicroScan Walkaway-96 System (Dade Behring, Inc., West Sacramento, CA). The above three semi-automated systems were recently compared to the conventional phenotypic confirmatory tests with regard to their ability to detect ESBL production in well characterized Enterobacteriaceae including Enterobacter spp., C. freundii and S. marcescens. The system with the highest sensitivity was Phoenix (99%), followed by Vitek 2 (86%) and MicroScan (84%); however, specificity was more variable, ranging from 52% (Phoenix) to 78% (Vitek 2). In addition, the performance differed widely with the species investigated. In contrast, the three available Etest strips and four disks combination (including ceftazidime, cefotaxime, cefpodoxime and cefpirome) showed sensitivity of 94% and 93%, and specificity of 85% and 81%, respectively. The double-disk test with ceftazidime, cefotaxime, cefpodoxime and cefpirome showed the highest specificity and positive predictive value among all test methods (i.e., 97% and 98%, respectively) [67].

Clinical impact and implications for therapy

Patients with infection due to ESBL-producing enterobacteria tended to have less satisfactory outcomes than those infected by pathogens that do not produce ESBLs [74–79]. In a prospective multinational study analyzing bloodstream infections (BSIs) due to ESBL-producing K. pneumoniae isolates, cephalosporin monotherapy was associated with a 40% 14-day mortality rate [80]. A comparable mortality rate among patients treated empirically with cephalosporin monotherapy was observed in another study concerning BSIs due to CTX-M-type-producing E.coli isolates [39]. These data force us to conclude that extended-spectrum cephalosporin treatment is associated with high rate of treatment failure (>80%) and mortality (>35%) when the susceptibility of infecting strains is close to the CLSI breakpoints (i.e., MIC ≥ 16 µg/ml). Data regarding cephalosporin use in infections due to ESBL-producing Enterobacteriaceae other than E. coli and K. pneumoniae are insufficient [75].

One of the determinant factors in the outcome of infected patients is choosing the appropriate empiric therapy within the first 24–48 hours of presentation. In our opinion optimal therapy of infections due to ESBL producers should be based on studies of in vitro activity of antimicrobial agents, carefully performed pharmacokinetic/pharmacodynamic (PK/PD) studies, and clinical investigations conducted in a prospective fashion. In reality, these studies are logistically challenging and rare. The clinician is faced with the fact that many ESBL producing bacteria also are resistant to quinolones (e.g., ciprofloxacin), aminoglycosides (e.g, gentamicin and tobramycin), and sulfamethoxazole [10]. A “magic bullet” is indeed hard to find.

Cefepime

Many in vitro studies demonstrate that ESBL-producing enterobacteria are frequently susceptible to cefepime (MIC ≤ 8 µg/L). However, it is important to note that these isolates should be reported as resistant to cefepime, according to CLSI and BSAC criteria. Several studies analyze the outcomes of patients treated with cefepime therapy for infection with ESBL-producing, cefepime-susceptible E. coli and K. pneumoniae [81–83]. From these retrospective analyses, cefepime was associated with a failure rate of 23–83% especially when MICs against ESBL producing Gram-negative bacteria were > 1 µg/ml. A randomized, evaluator-blind, multicenter trial found imipenem/cilistatin (0.5g q6h by i.v. route) to be substantially better than cefepime (2g q8h by i.v. route) for treatment of nosocomial pneumonia among ICU-patients[84]. These results suggest that cefepime is not the optimal therapy in the treatment of ESBL-producing enterobacteria, particularly during serious infections (e.g., bacteremia and pneumonia). Is a higher dose regimen of cefepime more effective? In combination with amikacin and ciprofloxacin, 2 g of cefepime q8h is comparable to carbapenems in patients infected by E. aerogenes containing the TEM-24 ESBLs [85]. The administration of higher doses (i.e., 4–6g administered as a continuous infusion or 2g q6–8h with prolonged infusion) results in adequate time above the MIC (%T > MIC) for the infecting organism achieving clinical success [86].

Beta-lactam/beta-lactamases inhibitor combinations

Since ESBL-producing enterobacteria are frequently susceptible in vitro to beta-lactam/beta-lactamases inhibitor combinations, it is logical to assume these combinations would also be clinically effective. Clinicians must keep in mind that presence of a chromosomal AmpC enzymes that is normally resistant to inactivation by a beta-lactamase inhibitor may be present [70]. Based on this and the data obtained from microbiological and clinical observations, we do not regard beta-lactam/beta-lactamases inhibitor combinations as a suitable option for serious infections due to ESBL-producing enterobacteria [87,88].

Carbapenems

Carbapenems (e.g., imipenem, meropenem, and ertapenem) have the most consistent activity against ESBL-producing Enterobacteriaceae. Large scale surveillance studies demonstrate that >98% against ESBL-producing E. coli, Klebsiella spp, and P. mirabilis isolates are susceptible to carbapenems [89–91]. Based on retrospective and prospective analyses, carbapenems should be considered as the preferred treatment for infections due to ESBL-producing enterobacteria [87,92]. In an international study conducted by Paterson et al. that includes ESBL-producing K. pneumoniae bloodstream isolates, patients who were treated with imipenem/cilistatin during the 5-day period after onset of infections had a 14-day mortality of 4.8%, compared with 27.6% when another in vitro active beta-lactam was used [80].

Quinolones

If the pathogen producing an ESBL is in vitro susceptible to ciprofloxacin, a satisfactory clinical response can be achieved using quinolones. Unfortunately, epidemiological studies reveal a strong link between fluoroquinolone resistance and ESBL production in Enterobacteriaceae [89]. For example, a multicenter prospective study of K. pneumoniae bacteremia conducted during 1996–1997, detected ESBL production in 60% of ciprofloxacin-resistant isolates, compared with 16% of ciprofloxacin-susceptible strains [93]. We discourage the use of quinolones as empiric therapy, since resistance to these agents is increasing in the United States and throughout the world.

Aminoglycosides

As in the case with quinolones, aminoglycosides are effective therapy against ESBL producing pathogens provided the organism has a MIC appreciably lower than susceptibility breakpoints. Susceptibility to amikacin seems to be preserved, in contrast to gentamicin and tobramycin, thus justifying its use as empiric therapy [89].

Tigecycline

CLSI criteria to interpret susceptibility testing of tigecycline are not yet established [62]. The Food and Drug Administration (FDA) and European Committee an Antimicrobial Susceptibility Testing (EUCAST) have created provisional resistance breakpoints (MIC ≥ 4 µg/ml and ≥ 2 µg/ml, respectively). In vitro data supports the notion that tigecycline can be considered an alternative to carbapenems for treatment of infections due to ESBL-producing Enterobacteriaceae [94,95]. However, clinical experience with tigecycline is still evolving.

Fosfomycin

The excellent in vitro activity of fosfomycin against ESBL-producing E. coli and K. pneumoniae strains has been recently reported [96]. Further studies are required to assess the efficacy of fosfomycin for the treatment of UTIs caused by ESBL-producing enterobacteria [97].

Colistin

Although once considered a toxic antibiotic, clinicians have now turned to colistin as a last resort agent for the treatment of infections caused by multidrug resistant gram-negative bacteria, against which this cationic detergent-like compound remains active [98]. As mentioned above, carbapenem resistance mediated by KPC is emerging among enterobacteria. Their implication in outbreaks, as seen in hospitals in New York City, has created a context in which the empiric use of colistin is necessary [99,100].

Conclusions

ESBLs are paradigmatic as a mechanism of resistance because of the impact they have had on the therapy of infections and the insight they have offered on the relationship between structure and function, in antibiotics and in their determinants of resistance. In TEM and SHV family betalactamases, single amino acid substitutions allow the efficient hydrolysis of extended-spectrum cephalosporins, while lowering the k_cat/K_m for penicillins. In the CTX-M family, point substitutions leading to specific interactions between enzyme and substrate are responsible for the improved activity against cefotaxime in preference to ceftazidime. Usually encoded on mobile genetic elements that accelerate their dissemination, the epidemiology of ESBLs has transformed recently with the surge of KPC and CTX-M ESBLs. This illustrates the complex interactions between antibiotic use, selection and transmission of resistance, colonization and infection in different populations. The future development of i) novel beta-lactams resistant to hydrolysis by these versatile enzymes and ii) the discovery of highly potent “second generation” beta-lactamase inhibitors are eagerly awaited.