INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Among the Enterobacter genus, Enterobacter cloacae and Enterobacter aerogenes are the two species most frequently isolated from clinical samples. These organisms belong to the normal digestive flora and are opportunistic pathogens colonizing and infecting hospitalized patients (25). Until recently, E. cloacae was the species most frequently reported in nosocomial infections. Over the last 5 years, E. aerogenes has emerged as an agent of nosocomial infection, and multiresistant strains have caused outbreaks in intensive care units (ICUs) in Belgium (9, 13), France (2, 6, 7, 12, 19), Austria (1), and the United States (11). Recent national surveys have described the epidemiology of E. aerogenes in Belgium (23) and France (4, 10). Other recent works have analyzed the multiple mechanisms of resistance of this organism to various antimicrobial agents (5, 8, 15, 21, 22).

Data from the bloodstream infection component of the National Surveillance of Hospital Infections and from a retrospective study on the epidemiology of E. aerogenes isolates in Belgian hospitals showed statistically significant increases between 1994 and 1995 of the incidence of E. aerogenes isolates (from 3.2 to 5.6 per 10,000 patient-days) and of the incidence of bloodstream infections (from 2.3 to 2.6 per 10,000 patient-days in ICU) (23). A significant increase of the frequency of resistance of this organism to ceftazidime, fluoroquinolones, and imipenem was observed in 29, 23, and 14% of hospitals surveyed, respectively (23). Consequently, two national surveys were carried out, (i) to evaluate the incidence and antimicrobial resistance trends of E. aerogenes strains from Belgian hospitals in 1996 and the first half of 1997 and (ii) to determine the clonal distribution and antibiotic susceptibility patterns of isolates collected during 1997 and 1998.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Descriptive epidemiologic survey. Both of the surveys were organized by the GDEPIH-GOSPIZ. Data were collected by mailing a retrospective questionnaire to 39 hospital laboratories, which were selected either on the basis of their willingness to participate voluntarily or because they already took part in a similar survey carried out in 1994 and 1995 (23). This questionnaire aimed to evaluate local prevalence and incidence data for E. aerogenes clinical isolates recovered from routine culture during 1996 and the first half (from January to June) of 1997. No attempt was made to differentiate infecting from colonizing organisms. To obtain comparable data, each participating laboratory was asked to specify the method used for data collection, and only those able to analyze all isolates with exclusion of screening cultures and duplicate isolates from the same body site per episode of hospitalization were included. Questions on the methods used in clinical laboratories for species identification and antibiotic susceptibility testing of E. aerogenes were part of the questionnaire. Aggregation of numerators and denominators from all hospitals provided a pooled mean. The relative frequency of E. aerogenes isolation was calculated as the proportion of patients with E. aerogenes among patients with a culture yielding Enterobacter spp. or other Enterobacteriaceae. Incidence was defined as the number of patients colonized or infected with E. aerogenes per thousand admissions. No attempt was made to ascertain the clinical significance of E. aerogenes isolates.

Prospective microbiologic survey. All participants were asked to send 10 consecutive nonduplicate E. aerogenes isolates from any site from patients hospitalized during the period including 1997 and the first half of 1998. Identification results were confirmed by conventional phenotypic tests and amplification by PCR of tRNA intergenic spacers (tRNA-PCR). The following information was recorded on the origin of strains: patient age, gender, hospital location and size, type of ward, and origin of specimen.

Molecular identification of E. aerogenes isolates. The genotypic identification technique used was tRNA-PCR, which was carried out as described by Welsh and McClelland (28). The tRNA-PCR profiles of the unknown strains were compared with those of reference strains of different Enterobacter species using the GeneScan software.

Antimicrobial susceptibility testing. MICs of 10 antimicrobial agents were determined by the agar dilution method, and results were categorized according to the National Committee for Clinical Laboratory Standards breakpoints (18), except for isepamicin MIC results, which were interpreted according to the guidelines of the Comité de l'Antibiogramme de la Société Française de Microbiologie, and for temocillin MICs, which were categorized according to the recommendations of the manufacturer (SmithKline Beecham Pharma) (breakpoints [in micrograms per milliliter] for susceptible and resistant categories: isepamicin, 16 and 64; temocillin, 16 and 32). Pure powder for analysis was kindly provided by the manufacturers. A multiresistant E. aerogenes (MREA) isolate was defined as one that was resistant to ceftazidime and ciprofloxacin.

Double-disk synergy test. Screening for extended-spectrum beta-lactamase (ESBL) production was performed by determining potentiation of the zones of inhibition around ceftazidime (30 µg) and/or cefepime (30 µg) disks by placing clavulanic acid disks (20 µg of amoxicillin and 10 µg of clavulanic acid) 30 mm (ceftazidime) or 20 mm (cefepime) apart (19, 26).

Beta-lactamase detection and identification. Isoelectric focusing (IEF) was performed on polyacrylamide gels as previously described (16). Enzyme activity was detected by the chromogenic nitrocefin test. Beta-lactamases with known isoelectric points (TEM-1, 5.4; TEM-2, 5.6; TEM-3, 6.3; TEM-24, 6.5; and SHV-1, 7.6) were focused in parallel with the extracts.

AP-PCR fingerprinting. DNA was extracted from a whole-cell lysate of a single colony by boiling for 15 min. Two sets of primers were used, primer 5 (Pharmacia Biotech) and primer M13 (5'-GAGGGTGGCGGTTCT). Amplifications were performed using Ready-To-Go PCR Beads (Pharmacia Biotech) on a Bio-Med Thermocycler 60 thermal cycler as follows: for primer 5, 94°C for 120 s followed by 35 cycles of 94°C for 30 s, 50°C for 60 s, and 72°C for 30 s; for primer M13, 95°C for 300 s followed by 45 cycles of 95°C for 60 s, 36°C for 60 s, and 72°C for 60 s followed by 72°C for 120 s. DNA fragments were separated at 180 V for 3 h on 2% agarose gels containing ethidium bromide. Gels were analyzed visually and by using BioNumerics software (Applied Maths, Ghent, Belgium). Normalized arbitrarily primed PCR (AP-PCR) profiles were compared using the Pearson correlation product moment coefficient, and those coefficients produced a matrix of similarity between strain pairs for each primer. A combined matrix of similarity was constructed by averaging the two primer-specific matrices and used for generating a dendrogram by the unweighted pair group method using arithmetic averages.

Performance of AP-PCR typing. Typeability and discriminatory power were determined as previously reported (9) for 25 epidemiologically unrelated E. aerogenes strains, which were previously characterized into different pulsed-field gel electrophoresis (PFGE) (XbaI) types. The reproducibility was evaluated as follows. Duplicate bacterial lysates of each of 10 epidemiologically unrelated E. aerogenes strains were coamplified in the same PCR experiment and in separate PCR experiments. These repeat samples were analyzed in the same gel for interrun and interextract pattern reproducibility. Intergel reproducibility was assessed by analyzing multiple amplicons of a single strain in 15 different gels.

PFGE analysis. To validate results of PCR type clustering of E. aerogenes strains in two groups, PFGE analysis was performed on two samples of 13 and 15 isolates originating from these groups. DNA preparation and cleavage with XbaI were performed as previously described (9). The gels were stained with ethidium bromide (0.25 mg/liter) and photographed using a Bio-Rad GelDoc 1000 camera. DNA patterns were analyzed visually and by using BioNumerics software (Applied Maths). Normalized PFGE profiles were compared using the Pearson correlation product moment coefficient.

Epidemiologic analysis of E. aerogenes isolates. An epidemic AP-PCR type of E. aerogenes was defined as a type recovered in more than one hospital, and a sporadic type was defined as a unique pattern found in a single strain. The geographic distribution of isolates by AP-PCR type was mapped by hospital location.

Statistical methods. Comparison between rates and proportions included Fisher's exact tests and Mantel-Haenzel chi-square tests. Differences showing P values of less than 0.05 were considered significant. All calculations were made by using EpiInfo software, version 6 (Division of Surveillance and Epidemiology, Epidemiology Program Office, Centers for Disease Control and Prevention, Atlanta, Ga.).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Descriptive epidemiologic survey. Thirty-nine hospitals (20% of Belgian acute care institutions) completed the questionnaire. Middle-size (400 to 599 beds) and larger (>600 beds) institutions had higher response rates (52 and 58%, respectively) than smaller hospitals (less than 400 beds) (12%). Six of the 13 university hospitals (46%) responded. The regional distribution showed 16 participants each from Wallonia and Flanders and 7 from the Brussels region.

Prospective microbiologic survey. Of 274 strains recovered from 31 participating centers, 265 (96.7%) were confirmed as E. aerogenes by phenotypic tests. The remaining nine strains reported as E. aerogenes were reidentified as E. cloacae (n = 4), Hafnia alvei (n = 2), Klebsiella oxytoca (n = 1), Klebsiella ornithinolytica (n = 1), and Serratia marcescens (n = 1).

Molecular identification of E. aerogenes isolates. Intergenic tRNA spacer fragments with lengths of 85, 86, 110, 111, 113, 114, 119, 120, 189, and 191 bp were present for all E. aerogenes strains. The presence of fragments of 104 and 105 bp and of 198 and 199 bp was variable. E. aerogenes was clearly differentiated from the seven other Enterobacter species tested (E. agglomerans, E. amnigenus, E. asburiae, E. cloacae, E. gergoviae, E. hormachei, and E. sakazakii) by the presence of peaks at 113 and 120 bp (except that the 113-bp peak was observed for E. amnigenus as well). With each of these species, several other differences were present as well.

Antimicrobial susceptibility. The MICs of 10 antimicrobial agents were determined with 249 of the 259 confirmed strains that were available after further subculture at that time of the study (Table 1). More than 76% of isolates were resistant to both ceftazidime and ciprofloxacin (Table 1). Resistance to cefepime and imipenem was observed in 7 and 2% of the strains, respectively. All imipenem-resistant strains were resistant to cefepime. No meropenem-resisant strains were observed. More than 90% of the strains were susceptible to aminoglycosides. MICs of gentamicin and isepamicin were two to four times lower than those of amikacin.

View larger version (31K): [in this window] [in a new window]   FIG. 1.   Distribution of MICs of ceftazidime (top) and cefepime (bottom) for E. aerogenes isolates (n = 236) according to the presence of plasmid ESBL. Black bars, ESBL-producing strains; white bars, non-ESBL-producing strains.

AP-PCR typing. All strains tested (n = 260) were typeable. Visual analysis of AP-PCR fingerprints from quadruplicate testing of 10 strains showed a variation of a single low-yield amplified DNA fragment in patterns obtained in two different amplification assays of the same DNA preparation. Consequently, patterns were considered by visual comparison to represent the same type if they displayed no more than one band difference. Computer analysis of the combined two primer patterns showed 78% similarity for intergel pattern reproducibility. The discriminatory power was 81% for unrelated strains.

View larger version (26K): [in this window] [in a new window]   FIG. 2.   Dendrogram of similarity of AP-PCR types of E. aerogenes Belgian isolates (n = 260). a, epidemic types; b, locally epidemic types; c, sporadic types.

View larger version (14K): [in this window] [in a new window]   FIG. 3.   Geographic distribution of E. aerogenes isolates belonging to the 2 major epidemic types. Flanders (upper part of Belgium) is separated from Wallonia (lower part) and Brussels (central position). Diagrams indicate the proportions of AP-PCR type BE 1 (black portions) and BE 2 (white portions) strains by hospitals.

PFGE analysis. Strains belonging to AP-PCR type BE 1 clustered together and were clearly grouped apart from those belonging to type BE 2. However, heterogeneity was observed for some strains within each group, with a few strains showing up to 10 DNA fragment differences (Fig. 4).

View larger version (77K): [in this window] [in a new window]   FIG. 4.   PFGE profiles of epidemic E. aerogenes isolates grouped by AP-PCR types 1 and 2 and dendrogram of pattern similarity based on the Pearson similarity coefficient.

Correlation between AP-PCR type and resistance phenotype. ESBLs were present in 91% of AP-PCR type BE 1 strains, of which 95% were identified as TEM-24, and in 21% of AP-PCR type BE 2 strains, of which 50% were identified as TEM-24. Strains of types BE 1 and BE 2 were more frequently resistant to ceftazidime (P < 0.001) and ciprofloxacin (P < 0.001) than other strains and were more frequently resistant to piperacillin-tazobactam (P < 0.001) than the other epidemic strains. BE 1 strains were more frequently resistant to ceftazidime (99 versus 87% [P < 0.001]) and ciprofloxacin (93 versus 76% [P < 0.001]) than BE 2 strains and were more susceptible to amikacin (100 versus 85% [P < 0.001]). The 11 strains belonging to the sporadic AP-PCR types did not carry any ESBLs.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Incidence data from the present survey indicate a continuing increase of the importance of E. aerogenes as a nosocomial pathogen in Belgium between 1996 and 1997, particularly in Wallonia. The limited sample size in our survey was related to the difficulty to retrieve adjusted epidemiologic data on rates of isolation of E. aerogenes from two previous years. Indeed, data from half of the participants could not be analyzed, either because they were incomplete or because the collection methods were not acceptable for reliable comparison with other centers. National surveys from France used similar sample sizes (4, 10).

Belgian clinical laboratories identified E. aerogenes strains accurately, in agreement with an external prospective survey of the Belgian national external quality control, which also showed accurate identification of an E. aerogenes isolate by 95% of the participants in 1997.

Seventy percent of the centers which answered the questionnaire routinely test Enterobacter isolates for production of ESBLs. This could reflect awareness of these participating laboratories about E. aerogenes as a multidrug-resistant pathogen. It is likely that this proportion is representative not of all Belgian laboratories but of those with a special concern for this pathogen. The variety of susceptibility testing techniques as well as modification of methods over the years by the participants precluded the inclusion of antibiotic resistance rates in the survey. Participants were asked instead to send consecutive isolates for standardized testing in a central laboratory. The majority of E. aerogenes isolates in this study were resistant to -lactam drugs, except for cefepime and carbapenems. This resistance is compatible with derepressed cephalosporinase-hyperproducing mutants, combined or not with ESBL production (21, 22). Sanders et al. (24) reported on the efficacy of cefepime for the treatment of infections due to ceftazidime-resistant E. cloacae and E. aerogenes strains, but no data about ESBL production by these isolates were provided. Approximately half of the strains analyzed in our study were ESBL producers. Few published data on the clinical efficacy of cefepime for treatment of infections caused by ESBL-producing E. aerogenes strains are available. Péchère and Vladoianu (20) showed that ceftazidime-resistant E. cloacae strains contained a subpopulation of cefepime-resistant strains and that cefepime often failed to cure peritonitis due to ceftazidime-resistant E. cloacae strains in a murine experimental model. Furthermore, Limaye et al (14) described the emergence of cefepime-resistant E. aerogenes strains during cefepime treatment of a patient with septicemia and hepatic abscess with an E. aerogenes strain which was initially susceptible to ceftazidime but became resistant after ceftazidime use. These observations led Meideros (17) to caution against cefepime use in the treatment of high-inoculum ceftazidime-resistant Enterobacter infections. An alarming observation in our study was the finding of occasional strains that were resistant to both cefepime and imipenem. Although no information was available on use of antimicrobial agents in the patients colonized with such strains, this phenotype has been reported to emerge during imipenem therapy (2, 5, 8, 9, 14, 15). The vast majority of isolates were also resistant to ciprofloxacin, similarly to strains in France (4). The mechanisms of resistance characterized in MREA strains include overproduction of chromosomal beta-lactamases, ESBL production, modified porin expression, and active efflux (5, 8, 15).

Although most strains were susceptible to aminoglycosides at breakpoint, MICs of gentamicin and isepamicin were lower than those of amikacin, suggesting the production of the AAC(6')-I enzyme. These data are in agreement with the description of French E. aerogenes strains (4) and with a Belgian study showing that blood isolates of aminoglycoside-resistant E. aerogenes harbor the AAC(6')-I gene (27).

American ESBL-producing E. aerogenes strains described to date produce a variety of SHV-type enzymes, such as SHV-3, -4, and -5 (21, 22). Most ESBL-producing strains examined here possessed TEM-24 or, less commonly, TEM-3, a finding similar to the predominance of TEM-24 among French isolates (3, 4, 10, 19). At present, there are no guidelines available for the detection of ESBLs in Enterobacter spp. Neuwirth et al. (19) and Tzelepi et al. (26) have suggested using ceftazidime and cefepime disks placed 20 mm apart from amoxicillin-clavulanic acid disks. In our experience, this synergy test correlated well with IEF results. Occasional strains showing discrepant results should be further characterized at the genetic level. In countries like Belgium and France, ESBL-producing E. aerogenes strains have become common, exhibit epidemic potential, and can transfer their resistance plasmid to other species of Enterobacteriaceae (19). Because of this clinical and epidemiologic significance of ESBL-producing Enterobacter spp., all clinical Enterobacter isolates should, in our opinion, be routinely tested for ESBL production, using the double-disk synergy test.

Computer-assisted analysis of AP-PCR profiles using two sets of primers was shown to be useful for typing Belgian isolates of E. aerogenes. It correlated well with visual analysis and was more rapid. PFGE analysis of a sample of strains confirmed the overall relatedness but also revealed further heterogeneity within E. aerogenes AP-PCR genotypes. This observation is in contrast with previous studies showing the spread of clonally related strains with homogeneous PFGE profiles (9, 13, 19) in local outbreaks. E. aerogenes strains recovered from 12 French hospitals had closely related PFGE profiles (10). Most strains examined here were also closely related by PFGE typing, whereas a few more distantly related variants were observed among the larger number of isolates examined here, indicating that a progressive genomic shift may have occurred over the course of dissemination for several years.

In contrast to previous studies, our results show that MREA did not occur only among patients admitted to ICUs but also occurred among patients cared for in general medical or surgical wards. Such a dissemination can be related to the use of extended-spectrum -lactams in all wards and to the transfer of colonized patients from the ICU to other wards of the same hospital or other hospitals. The potential of MREA to spread was also illustrated by the number of small clusters of epidemic MREA that we observed. In contrast to the one TEM-24-producing epidemic strain in France (4), the spread of Belgian isolates was distributed predominantly in two major AP-PCR types, BE 1 and BE 2. BE 1 strains, which were more prevalent in Wallonia, also produced a TEM-24 beta-lactamase and were closely related to or indistinguishable from the French epidemic strain by AP-PCR typing. These data suggest that cross-border dissemination has occurred between France and Belgium, although the sequence of spread is not clear.

In conclusion, this survey shows that a diversity of antibiotic-resistant E. aerogenes strains have become endemic in Belgian hospitals and that two clusters of genotypically related strains are widespread in this country. These data support the need for further routine laboratory detection of ESBL-producing strains and for isolation of patients carrying these strains to control their spread. We plan to monitor further the epidemiology and nosocomial incidence of MREA through a national surveillance program.