Introduction

Temocillin is the 6-α-methoxy derivative of ticarcillin. This structural modification increases stability to AmpC β-lactamases as well as to extended-spectrum beta-lactamase (ESBLs) enzymes, but reduces its activity against Pseudomonas aeruginosa, Acinetobacter spp. and almost completely removes its affinity for the essential penicillin-binding proteins of Gram-positive bacteria and anaerobes (Bacteroides fragilis and others) [1]. Overall temocillin appears as a narrow spectrum antimicrobial agent with very consistent activity against Enterobacteriaceae [2, 3] and selectively against some glucose non-fermenting Gram-negative bacilli (e.g. Burkholderia cepacia, Delftia acidovorans and Comamonas spp.) [4, 5] that may be involved in lung infection among patients with cystic fibrosis.

Temocillin marketed in Belgium and in the UK in the late eighties by Beecham Pharmaceuticals and presently licensed to Eumedica has been used on a limited scale in these two countries in microbiologically documented therapy of infections known to be caused by pathogens with potent β-lactamases. In Belgium, temocillin is currently recommended for parenteral treatment of acute pyelonephritis (Belgium Antibiotic Politic Coordination Committee guidelines: www.health.fgov.be/vesalius) and it has been approved recently in the US for treatment of Burkholderia cepacia lung infection in cystic fibrosis [5]. Recent studies on the in vitro activity of temocillin are limited [6–8]. One multicentric survey in Belgium showed susceptibility rates of 79–100% among clinical isolates of Enterobacteriaceae collected from hospitalized patients [6]. Another local survey indicated good activity of temocillin against multiresistant ESBL-producing Escherichia coli isolates (92% of the strains were found susceptible to temocillin) [7], while a recent study from the UK showed that nearly 90% of the AmpC- and ESBL-producing isolates Enterobacteriaceae isolates were susceptible to temocillin [8].

Owing to the increasing incidence of infections caused by multi-drug resistant Gram-negative bacteria and also because of the limited number of clinically active drugs available against such organisms, we aimed to determine the prevalence of ESBL-producing Enterobacteriaceae in several Belgian intensive care units (ICUs) and to assess the in vitro activity of temocillin against these isolates.

Materials and methods

Bacterial strains

Non-duplicate, consecutive clinical isolates of various Enterobacteriaceae were prospectively obtained from patients hospitalized in ICUs for more than 72 h in order to enrich the collection of organisms as much as possible on nosocomial pathogens. The survey took place at seven different university hospitals distributed throughout the country during a 6-month period in 2005. Each laboratory collected up to 100 clinical isolates. Origins of the specimens included lower respiratory tract (sputum, bronchial or endotracheal aspirates, broncho-alveolar lavage), pus, wounds and blood samples. Routine screening cultures (i.e. rectal swabs and stools) as well as screening samples from environmental sources were excluded. Urine specimens were not retained because nosocomial urinary tract infections seldom occur in patients hospitalized in ICUs and the clinical significance of positive urine cultures in patients with bladder catheters is most often uncertain.

Identification of the isolates was performed at each participating laboratory following its routine methodology. All strains were stored at −80°C and were centralized at one laboratory (UCL Mont-Godinne University Hospital) for susceptibility testing, confirmation of ESBL production and characterization of the enzymes involved in resistance.

Susceptibility testing

MICs were determined by E-test method (AB BIODISK, Solna, Sweden) for amikacin, ciprofloxacin, ceftazidime, meropenem, temocillin and piperacillin/tazobactam (only for ESBL-producing isolates) according to the manufacturer’s instructions and results were categorized using breakpoints and interpretive criteria provided by the CLSI [9]. Susceptibility to temocillin was determined according to the breakpoint values provided by Fuchs et al. [10]: susceptible if MIC ≤ 16 μg/ml and resistant if MIC ≥ 32 μg/ml. The Escherichia coli ATCC 25922 and Klebsiella pneumoniae ATCC 700603 (SHV-18 ESBL-producing) strains were used at least twice as controls to validate the MIC results and ESBL detection procedures.

Detection of ESBL production

Screening of ESBL production was carried out using a species-adapted algorithm. All Escherichia coli and K. pneumoniae isolates were subcultured on two different selective Mueller-Hinton agar plates (Oxoid, Basingstoke, UK) containing cefotaxime (0.5 μg/ml) and ceftazidime (0.5 μg/ml). Colonies growing on one or both ESBL screening plates were further tested by ESBL E-tests (ceftazidime/ceftazidime-clavulanate and cefotaxime/cefotaxime-clavulanate) (AB BIODISK, Solna, Sweden) for confirmation of ESBL production.

Isolates of K. oxytoca were assessed for ESBL production by double ESBL E-test using ceftazidime/ceftazidime-clavulanate when ceftazidime MIC was ≥ 4 μg/ml.

For AmpC inducible Enterobacteriaceae isolates (Citrobacter freundii, Enterobacter spp., Serratia spp., Providencia spp. and Morganella morganii), strains displaying ceftazidime MICs ≥ 4 μg/ml were also evaluated for ESBL production by cefepime/cefepime-clavulanate ESBL E-test strips as recommended by Stürenburg et al. [11].

An ESBL MIC ratio  ≥ 8 of any of the tested cephalosporin substrates alone versus its combination with clavulanic acid was used as an indicator of ESBL production. Isolates with ceftazidime MIC > 2 μg/ml and for which no ESBL production could be detected by phenotypic tests were considered as probable AmpC hyperproducers.

Characterization of ESBLs

Isoelectric focusing was performed on crude bacterial extracts frozen/thawed several times from bacterial growth in exponential phase in brain heart infusion medium using the Phastsystem apparatus (Pharmacia AB, Uppsala, Sweden). Isoelectric points (PIs) were determined by comparing PI values of selected β-lactamases with known PIs (5.4, 6.3, 6.5, 7.6, 8.1) after staining of gels with nitrocefin (500 μg/ml).

ESBLs were further characterized by PCR for bla genes of TEM, SHV, and CTX-M families for all isolates. Detection of bla TEM, bla SHV and bla CTX-M gene families was performed with PCR using specific primers designed in house as described previously [12]. PCR products were analysed on a 1% agarose gel. Identification of TEM, SHV and CTX-M types and subgroups was performed on selected strains using DNA sequencing of full-length ESBL gene PCR products (ABI 3100, Perkin Elmer). Sequence homology was determined using BLAST for comparison in gene Bank databases. Selected strains of the same species showing homology in the resistance profiles and resistance mechanisms were analysed.

Results

In total, 652 isolates (mean number of isolates per hospital: 93; range: 65–100) were tested. Fifty percent (330/652) of the isolates originated from the lower respiratory tract, 28% (182/652) from pus or wounds, 7% (45/652) from blood and 15% (95/652) from other body sites.

Escherichia coli was the most prevalent species (186 isolates [29%]), followed by Enterobacter cloacae (115 isolates; 18%), Klebsiella pneumoniae (76 isolates; 12%) Enterobacter aerogenes (75 isolates; 11.5%) and Klebsiella oxytoca (9.5%). Concerning groups of organisms, the non-inducible Enterobacteriaceae spp (Escherichia coli, Klebsiella spp. and Proteus spp.) were more prevalent (57%) than the inducible AmpC-producing species (Enterobacter spp., Citrobacter spp., Serratia spp., M. morganii and Providencia spp.) which represented 43% of all isolates. Overall, meropenem displayed the highest activity (MIC50 of 0.064 μg/ml and less than 1% of the strains being non-susceptible) followed by amikacin (MIC50 of 2 μg/ml; 5% non-susceptible) and temocillin (MIC50 of 4 μg/ml; 8% non-susceptible isolates). Ceftazidime and ciprofloxacin were found to be less active (20% non-susceptible strains). The proportion of non-susceptible strains by species is shown in Table 1. Enterobacter aerogenes accounted as the most resistant species (66.7 and 57.0% of isolates non-susceptible to ceftazidime and to ciprofloxacin, respectively) while high resistance rates (around 30%) to ceftazidime were observed in Enterobacter cloacae, C. freundii and in K. pneumoniae isolates. Among Escherichia coli, the highest level of resistance was found for ciprofloxacin (17 %). Globally, temocillin had good activity against all Enterobacteriaceae species except against Enterobacter aerogenes and Serratia marcescens which were found less susceptible (25% non susceptibility for each of these two species).

Table 1 Number and percentage of non-susceptible isolates to the tested antimicrobials, classified by species
Full size table

Table 2 shows the MIC distribution of temocillin against a subset of Enterobacteriaceae isolates with defined resistance mechanisms. In all, 86% of the Enterobacteriaceae isolates with hyperproduced or acquired AmpC were susceptible to temocillin at ≤16 μg/ml (modal MIC=16 μg/ml). Likewise, ESBL-producing isolates had a temocillin modal MIC of 16 μg/ml with the single exception of ESBL and AmpC co-producing Enterobacter aerogenes which were less susceptible (MIC50 of 32 μg/ml).

Table 2 MIC distribution of temocillin for Enterobacteriaceae isolates with characterized resistance mechanisms
Full size table

Overall, 77 (11.8%) of the tested isolates were found to be ESBL-producers. The prevalence of ESBL-producing strains ranged between 3 and 29% in the different hospitals. Regarding the species distribution, ESBLs were predominantly found in Enterobacter aerogenes (26/72 [36%]) and in K. pneumoniae (23/75 [31%]) and less commonly among Enterobacter cloacae (9/115 [8%], Escherichia coli (12/186 [6%]), K. oxytoca (4/62 [6%] and Proteus mirabilis (2/35 [6%]).

Table 3 shows the distribution of ESBL enzymes determined by the consistent results of primers specific PCR followed by DNA sequencing for most of the isolates and the β-lactamases isoelectric points as determined by isoelectric focusing. The ESBL enzymes were distributed differently between health care centres and among the different Enterobacteriaceae species which may either reflect a clonal dissemination of outbreak-associated nosocomial isolates within and between hospitals (for instance in Enterobacter aerogenes and K. pneumoniae) or alternatively a limitation of horizontal plasmid or gene transfer between bacterial hosts. For instance, among the TEM-related ESBLs, the TEM-24 enzyme largely predominated in Enterobacter aerogenes and it was also occasionally found among other species (K. pneumoniae, Enterobacter cloacae) while TEM-17 and TEM-52 β-lactamases were exclusively found in Escherichia coli. On the other hand, the SHV-derived ESBLs were essentially found in K. pneumoniae isolates (SHV-4 and SHV-5 like) and in Enterobacter cloacae (SHV-12) as well as in one single Providencia spp. isolate (SHV-4); CTX-M enzymes were predominantly found in Escherichia coli (mainly CTX-M-1 group enzymes) and to a lesser extent in other Enterobacteriaceae species. In Enterobacter cloacae, CTX-M-9 group enzymes were often found in association with SHV-12. Two Proteus mirabilis isolates (one with TEM-24 and one with a TEM-131) were susceptible or displayed low resistance to ceftazidime (MIC of 0.38 μg/ml and 2 μg/ml) which may reflect the low level expression of ESBL enzymes in some Enterobacteriaceae groups such as Proteus species. Interestingly, no ESBL could be characterized for three isolates (two K. pneumoniae and one Enterobacter cloacae; Table 3) which may possibly reflect the presence in these strains of uncommon ESBL enzymes (e.g., VEB, PER, GES, BES, TLA) that were not targeted by our PCR assays. Additionally, other non-ESBL narrow-spectrum β-lactamases were also often detected; mainly TEM-1/ TEM-2 in Escherichia coli and in Proteus mirabilis and SHV-1 in K. pneumoniae (data not shown).

Table 3 Characterization of extended-spectrum beta-lactamases in ESBL-positive Enterobacteriaceae isolates
Full size table

Concerning the activity of the different drugs, according to the mechanisms of resistances it was very striking that all agents except meropenem displayed a lower activity against ESBL-producing isolates (Table 4). In particular, the activity of ceftazidime and of ciprofloxacin was highly influenced by the presence of ESBLs. Non-susceptibility to ciprofloxacin or to ceftazidime was observed in respectively 58 and 66% of the ESBL-producing Escherichia coli isolates. In K. pneumoniae and in Enterobacter aerogenes, resistance rates were even higher and reached almost 100% for ceftazidime and more than 80% for ciprofloxacin. The activity of piperacillin-tazobactam was also markedly reduced and less than 50% of the ESBL-producing isolates were susceptible to this agent. On the other hand, temocillin displayed a very similar activity against ESBL-producing or non-ESBL producing Escherichia coli (92 vs. 98% temocillin susceptible isolates). However, a lower proportion of temocillin-susceptible isolates was found among ESBL-producing AmpC-derepressed Enterobacter aerogenes isolates (24% resistant and 40% intermediately susceptible isolates) and to a lesser extent in ESBL-producing K. pneumoniae (4% resistant and 30% intermediately susceptible isolates).

Table 4 Comparative activity of the tested antibiotics against ESBL-positive and ESBL-negative Enterobacteriaceae isolates
Full size table

Discussion

In the present survey, we were able to document a mean prevalence of ESBL of roughly 12% among Enterobacderiaceae isolates collected from patients hospitalised in ICUs from seven Belgian university hospitals. Wide variations were, however, observed between centres both in the prevalence rates (from 3 to 29%) as well as in the species distribution among which these enzymes were found. Enterobacter aerogenes still accounted as the most prevalent ESBL-producing species (35% of the isolates belonging to this species were found to be ESBL-producers) but it was closely followed by K. pneumoniae (31% ESBL-producers) and to a lesser extent Enterobacter cloacae and Escherichia coli. Two previous national surveys have shown higher figures of ESBL prevalence in Enterobacter aerogenes (65% in 2001; 58% in 2003) [13] and the lower figures observed here could possibly reflect a decrease in clonal spread of the epidemic strains which affected most Belgian hospitals over more than 10 years since the early nineties [14, 15]. Alternatively, our data may be biased by the small sample size of this survey and its restriction only to ICU patients while several of the former studies screened isolates originating from all hospitalisation units and wards.

Also, the high rate of ESBL-producing K. pneumoniae observed in this study (31% of the isolates) is at variance with other longitudinal Belgian multicentric surveys carried out between 1998 and 2005 in which a lower prevalence of ESBL (ranging between 5 and 15%) was observed in Klebsiella species [16]. Of note, the ESBL-producing K. pneumoniae isolates (SHV-4 like and SHV 5 like) found in the present study almost exclusively originated from two hospitals and would thus probably reflect the occurrence of local outbreaks in these institutions although the clonality of the collection of strains was not investigated in the present survey.

By contrast the high frequency of ESBL-producing Escherichia coli isolates which were found in all participating centres is consistent with the global trend towards increase in the incidence of these enzymes in Escherichia coli, and clearly reflects the recent emergence of the CTX-M ESBL enzymes which rose by a three-fold factor in Belgium between 2000 and 2004 [17].

The other major aim of this survey was to evaluate the activity of temocillin as well as selected comparators against resistant nosocomial Enterobacteriaceae isolates with defined acquired resistance mechanisms including ESBL and/or AmpC β-lactamases. We found that temocillin still retained good in vitro activity against the majority of Enterobacteriaceae isolates with a modal MIC of 4 μg/ml and a MIC90 of 16 μg/ml. Among the different comparators, meropenem appeared as the most active agent showing activity against more than 99% of the isolates regardless of the species and/or resistance mechanisms. On the other hand, ceftazidime and ciprofloxacin (89 and 72% non-susceptible isolates, respectively) were poorly active not only against ESBL-producing isolates but also against a significant proportion of AmpC overproducing species among which Enterobacter aerogenes again proved as the most resistant. Overall, amikacin retained good in vitro activity and it displayed an activity profile rather comparable to temocillin against the different Enterobacteriaceae species.

Although all mechanisms of resistance of the different isolates could not be fully determined in this survey, temocillin most notably retained activity against ESBL-producing Escherichia coli isolates. This observation agrees with the data of Rodriguez et al. [7] who found that 92% of the ESBL-producing Escherichia coli isolates identified at their hospital were susceptible to temocillin at a breakpoint value of 16 μg/ml. On the other hand, ESBL-producing Enterobacter aerogenes and K. pneumoniae isolates were found less susceptible to temocillin than ESBL-negative isolates belonging to these species.

In this study, we were not able to find a relationship between the level of resistance to temocillin and the presence of specific types of ESBLs. The highest rate of resistance was however observed in ESBL and AmpC co-producing Enterobacter aerogenes isolates as well as in S. marcescens, a species notoriously known to be intrinsically less susceptible to temocillin [1, 2, 6]. Contrary to Rodriguez-Villalobos et al. [7], we did not found higher rates of temocillin-resistant isolates among CTX-M-1 group isolates in comparison to isolates producing other types of ESBL enzymes. Acquired resistance to temocillin usually arises via combination of several mechanisms including the presence of ESBLs and/or AmpC hyperproduction with impermeability and/or upregulated efflux but these aspects were not investigated in the present study. Although the efficacy of carbapenems against ESBL-producing Enterobacteriaceae isolates has been widely documented, carbapenem resistance has emerged over the last years either by the acquisition of plasmidic metallo-β-lactamases or through the loss of porins. Hence, temocillin could be considered as a second line alternative therapeutic option to carbapenems for the treatment of infections caused by ESBL-producing strains which are causing an increasing number of infections in critically ill patients. However its use in this setting must be conditioned by microbiological investigations which should demonstrate the susceptibility to temocillin of these ESBL-positive strains and which at the same time rule out the presence of Pseudomonas aeruginosa or other temocillin naturally resistant non-fermenting organisms. Prospective clinical studies are therefore warranted in order to confirm its therapeutic efficacy.