ABSTRACT
The results of prevalence studies on glycopeptide-resistant enterococci (GRE) in the intestine may be influenced by the detection methods applied. In most studies different media, different concentrations of antibiotics, and different methods are used, and these differences result in differences in recovery rates. In this cross-sectional study on the carrier state of GRE among patients at the University Hospital Antwerp, Antwerp, Belgium, performed on 21 May 1996, direct plating and broth enrichment were compared by using the same media. Stool samples (n = 213) or rectal swabs (n = 122) were plated directly on Enterococcosel agar (bioMérieux) and after enrichment in Enterococcosel broth. The prevalence of GRE was 12.8%. Direct plating recovered 53.4% of the GRE isolates, and broth enrichment recovered an additional 46.5% of them; in the latter test the isolates were thus present at less than 103 CFU per g of feces. The prevalence of GRE among dialysis patients was higher than among the other patients, but the difference was not significant (P = 0.06), possibly as a result of the small numbers of dialysis patients examined. The GRE species isolated included 19 E. gallinarum (44.2%), 13E. faecium (30.2%), 6 E. faecalis (13.9%), and 5 E. casseliflavus (11.6%) isolates. All E. faecalis and E. faecium strains isolated carried thevanA gene, and E. gallinarum and E. casseliflavus carried the vanC1 and vanC2gene, respectively. The majority of isolates were polyclonal. Our data indicate that the rate of detection of GRE from both stool samples and rectal swabs is significantly increased with enrichment cultures.
During recent years glycopeptide-resistant enterococci (GRE), more commonly designated vancomycin-resistant enterococci (VRE), have attracted much interest (4, 7, 13, 19, 20, 21, 25, 27, 29, 33, 38). They may be part of the intestinal microflora of humans (8, 18, 19, 22, 39, 42) and animals (10).
While enterococci are easily cultivated on common laboratory media, the isolation of these organisms and, more specifically, the isolation of GRE from heavily contaminated specimens such as feces may be problematic. At present, there is no uniformly accepted screening method for GRE. Numerous types of commercially available and in-house-prepared selective agar and broth formulations have been used, and recently, their sensitivities for the isolation of enterococci from various specimens have been compared (12, 15, 26, 40). Although solid media are often used in screening studies (17), Enterococcosel broth may be preferred over other media for routine surveillance (26, 40).
Concentrations of vancomycin varying from 4 to 64 μg/ml have been added to these selective media. Whereas a high concentration is likely to detect high-level vancomycin resistance, a concentration of 6 μg of vancomycin per ml has been shown to be reliable for the detection of enterococcal strains with low-level glycopeptide resistance (34).
In May 1996, we determined the prevalence of gastrointestinal colonization with GRE among hospitalized and hemodialysis patients admitted to the University Hospital Antwerp.
In order to compare rates of isolation from both solid and enrichment broths, the sensitivities of direct plating and broth enrichment were compared by using Enterococcosel medium (40) to which vancomycin at 6 μg/ml was added (34).
MATERIALS AND METHODS
Prevalence study.The study was performed in the 600-bed University Hospital Antwerp on 21 May 1996. The study protocol was approved by the hospital’s ethical committee. After providing verbal consent, 306 (56.6%) of the 541 hospitalized patients and 29 hemodialysis patients participated in the study. A fecal sample was obtained. If this was not available on the day of the survey, a rectal swab was collected.
Enrichment cultures and identification of GRE.Rectal swabs were suspended in 1 ml of physiological saline; stool specimens were prepared as 10% suspensions. From these, a drop was streaked on ad-Enterococcosel agar plate (bioMérieux, Marcy-l’Etoile, France) containing 6 μg of vancomycin per ml and 0.8 ml was added to 5 ml of d-Enterococcosel broth supplemented with 6 μg of vancomycin per ml. After 24 and 48 h, black colonies with a brown halo appearing on the plates were subcultured for further study.
Broth tubes whose contents turned black after 1 or 2 days of incubation at 37°C were subcultured on vancomycin-supplementedd-Enterococcosel agar and onto Trypticase soy–5% horse blood agar.
Suspected enterococcal colonies were identified at the genus level by the following characteristics: cellular morphology, reaction on Gram staining, bile esculin hydrolysis, and catalase and pyrrolidonyl arylamidase (Rosco Diagnostica, Taestrup, Denmark) activity.
Species identification was based on the conventional method of Facklam and Collins (14): tolerance to tellurite for E. faecalis, motility and pigment production for E. gallinarum and E. casseliflavus, and acid formation from arabinose, lactose, mannitol, raffinose, ribose, sorbitol, sorbose, and sucrose and arginine hydrolysis for E. faecium. Identification was confirmed by oligonucleotide D11344-primed PCR as described previously (9). PCR for the species-specific genesvanC1 and vanC2 was used to confirm the identification of E. gallinarum and E. casseliflavus.
Susceptibility testing.The vancomycin resistance of colonies appearing on vancomycin-supplemented medium was first confirmed by subculturing the colonies on brain heart infusion (BHI) agar plates containing 6 μg of vancomycin per ml as described by Swenson et al. (34).
For enterococci growing on this medium, disk susceptibility tests were performed for vancomycin (30 μg; bioMérieux) and teicoplanin (30 μg; bioMérieux). The criteria recommended by the National Committee for Clinical Laboratory Standards were used (30, 31). The MICs of vancomycin and teicoplanin were determined by the E-test method (AB Biodisk, Solna, Sweden) according to the instructions of the manufacturer.
High-level resistance to aminoglycosides was determined on Mueller-Hinton agar (BBL Microbiology Systems) supplemented with streptomycin (2,000 μg/ml) and gentamicin (500 and 2,000 μg/ml) as described by Sahm and Torres (32).
Detection of van genes by multiplex PCR.The oligonucleotide primers for vanA, vanB, andvanC published by Clark et al. (6) were used in a multiplex PCR in a 50-μl volume. Five to 10 colonies from an overnight culture on blood agar were suspended in the reaction mixture containing 10 mM Tris HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCl2, each nucleotide at a concentration of 200 mM, 0.1 mM each primer, and 1 U of Taq polymerase (Goldstar; Eurogentec, Seraing, Belgium). The cycling was 94°C for 30 s, 58°C for 30 s, and 72°C for 30 s for 30 cycles, with a first step at 95°C for 10 min and a final step at 72°C for 10 min. For the amplification of the C2 gene the primers of Dutka-Malen et al. (11) were used in the same mixture with the following cycling: 94°C for 2 min for the first cycle, 94°C for 1 min, 54°C for 1 min, and 72°C for 1 min for the next 30 cycles and a final step of 72°C for 10 min. The amplicons were revealed by agarose gel electrophoresis on a 2% agarose (Hispanagar, Burgos, Spain) gel.
Genotyping.The clonal distribution among the vancomycin-resistant strains was studied by pulsed-field gel electrophoresis (PFGE) as described previously (9).
Epidemiological investigation.For each hospitalized patient colonized with a vanA-positive strain of GRE, two patients in the same ward at the time of the study were chosen as controls. In total, the data for 28 controls could be analyzed. The following factors were registered: age, sex, hospital ward at the time of sampling, principal diagnosis, length of stay in the hospital, and antibiotic treatment at the time of sampling.
RESULTS
Prevalence of colonization with GRE.Three hundred six (56.6%) of the 541 hospitalized patients and 29 hemodialysis patients agreed to participate in the study.
GRE were isolated from 43 (12.8%) of the 335 specimens examined: 36 from 306 hospitalized patients (prevalence, 11.8%) and 7 from 29 hemodialysis patients (prevalence, 24.1%) (Table1). The difference is not significant (P = 0.06).
GRE in fecal specimens
Direct plating yielded another 42 (12.5%) suspected enterococcal colonies that were not identified as GRE.
Enrichment cultures and identification of GRE.Of all specimens tested, 213 were stool specimens (63.6%) and 122 were rectal swab specimens (36.4%). The 213 fecal samples yielded 36 (16.9%) isolates of GRE, while 122 rectal swabs yielded 7 (5.7%) isolates of GRE. This difference is significant (P = 0.003). Broth enrichment increased the number of GRE by 46.5% (Table2). All isolates recovered after direct plating were recovered after broth enrichment. An additional 20 isolates were found only after broth enrichment. This holds particularly for E. casseliflavus, of which 4 of 5 isolates were obtained after enrichment. Broth enrichment yielded, after subculturing, another 49 isolates not identified as enterococci, resulting in a false positivity rate of 14.6%.
GRE isolated by direct plating and after broth enrichment
Among the GRE isolates, 19 were E. gallinarum, 13 wereE. faecium, 6 were E. faecalis, and 5 wereE. casseliflavus (Table 2). E. gallinarum andE. casseliflavus carried the vanC1 and thevanC2 genes, respectively, while E. faecium andE. faecalis carried the vanA gene.
Results of susceptibility tests.All E. faecalisisolates and all except one E. faecium isolate were resistant to vancomycin (MICs, >256 μg/ml) and teicoplanin (MICs, 48 to 128 μg/ml). One E. faecium isolate was borderline resistant. Although the strain was resistant to vancomycin and teicoplanin by the disc diffusion test, the teicoplanin MIC for the isolate was 2 μg/ml by the E test. Two of the E. gallinarum isolates were resistant to vancomycin (MICs, >256 μg/ml) and teicoplanin (MICs, 48 and 64 μg/ml, respectively). All the remaining E. gallinarum isolates were moderately vancomycin resistant (MICs, 6 to 8 μg/ml) and were teicoplanin susceptible. For the E. casseliflavus isolates the vancomycin MICs were 6 μg/ml and the teicoplanin MICs were 0.75 to 1.5 μg/ml. For 32.5% of the isolates (E. gallinarum, 3 of 19; E. faecium, 6 of 13; E. faecalis, 4 of 6;E. casseliflavus, 1 of 5) streptomycin resistance was more widespread than gentamicin resistance (E. gallinarum, 1 of 19; E. casseliflavus, 1 of 5).
Genotyping results.Cluster analysis and visual inspection of the restriction profiles obtained by PFGE revealed four E. faecium isolates with identical patterns. Two of these were from patients in the renal dialysis unit, and two were from patients admitted to the hospital the day before the survey for GRE. There was no epidemiologic relationship between these two patients or with the renal dialysis unit.
Another pair of identical E. faecium isolates that were different from the previous cluster were isolated from two patients hospitalized in the same ward for 48 and 54 days, respectively.
Two genotypically identical E. faecalis isolates were isolated from patients admitted to different wards for only 1 or 2 days prior to the survey.
Distribution of carriers of GRE in the hospital.The distribution of carriers of GRE was not homogeneous throughout the hospital. Six of 14 (40%) vanA carriers were hospitalized in five surgical wards; the other 9 (60%) were distributed over eight different medical wards. The prevalence among patients admitted to the hematology department was not higher than that among patients admitted to other departments. GRE carrying vanC1 or vanC2genes were identified in several wards. Compared to the overall prevalence in the hospital, carriers were found more often in the intensive care unit (3 carriers among 24 patients; 12.5%), the gastroenterology unit (3 carriers among 24 patients; 12.5%), the neonatology unit (3 carriers among 21 patients; 14.2%), and the pediatric unit (6 carriers among 28 patients; 21.4%).
Risk factors for colonization with GRE.No significant differences were found regarding sex, mean age, mean duration of present hospital stay, or antibiotic treatment for case patients and controls or for the number of days in the hospital during the past year (Table 3).
Comparison of risk factors among 15 patients colonized with GRE and 28 controls
The mean durations of hospital stay for carriers of vanAisolates of GRE and for controls were 18.7 and 12.1 days, respectively. Nine of 15 (60.0%) carriers of vanA GRE had received antibiotic treatment at the time of sampling. Among the control group, 19 of 28 (67.8%) were treated with antibiotics (P = 0.6). Six of 15 carriers were hospitalized for 1 or 2 days. Four of these had not received any antibiotics. Three of the five patients hospitalized for only 1 day had not been hospitalized during the past year; one patient had been hospitalized for 1 day during the previous year and one other patient had been hospitalized for 35 days during the previous year. These patients most probably were carrying GRE before their present hospitalization.
DISCUSSION
The prevalence of gastrointestinal colonization with GRE was 12.8% among the patients studied. This is more than 10-fold higher than the incidence of infections with these organisms in Belgium (35). Although the main reservoir for enterococci in humans is the gastrointestinal tract, the percentage of gastrointestinal tract carriage of GRE varies widely, depending on the geographic location, the setting, and the detection methods used. In the United States, the overall incidence of infection with GRE is rather high and nosocomial outbreaks are commonly reported (3, 5, 20, 28). High colonization rates have been found in hospitalized patients (28). The presence of GRE in the community and in animals in the United States is low, but because only a small number of samples have been examined, the prevalence is still largely unknown. In contrast, in Europe, despite a low prevalence of infection, GRE have been isolated at different rates from healthy volunteers, animals, and environmental sources (10, 37). The results of these studies are difficult to compare since in most of them, different media, different concentrations of antibiotics, and either direct plating or enrichment methods were used, and these differences may result in differences in recovery rates. Although the use of a direct plating method on antibiotic-containing medium has been shown to increase the rate of recovery of resistant enterococci (8), Klare et al. (23, 24) have recently reported on the isolation of GRE only after broth enrichment in antibiotic-free medium, suggesting that small numbers of organisms might be missed in selective media. Therefore, in the present study, both direct plating and the enrichment method were compared by using the same media in order to establish an optimal medium and method for the recovery of GRE.
The broth enrichment step used in the present study almost doubled the number of vanA- and vanC1-carrying isolates that were detected and was responsible for the detection of four of the fivevanC2-carrying E. casseliflavus isolates. The 19vanA-carrying isolates were identified as E. faecium (n = 13) and E. faecalis(n = 6). All specimens positive on direct plating were also positive after enrichment; no GRE were recovered exclusively on agar. A comparable increase in the numbers of isolates detected after broth enrichment was obtained by Ford et al. (16). Since the sensitivity of the plating procedure can be estimated to be 103 CFU/g of feces, the large proportion of isolates obtained only by broth enrichment illustrates the high frequency of carriers who excrete GRE in small numbers. A straightforward comparison by inoculation of both swabs and fecal samples both on agar plates and in enrichment broths was not realized because paired specimens were not obtained. Since the enrichment procedure involves the examination of a larger specimen volume, however, it is logical that it detects more carriers.
In contrast to Van der Auwera et al. (39), we found a significant difference between the rates of isolation of GRE from stool specimens (36 of 213; 16.9%) and from rectal swabs (7 of 122; 5.7%). A difference in the procedures used to collect rectal swabs cannot be excluded to explain the differences in the detection rates between these studies. However, in the study by Van der Auwera et al. (39), stool specimens were cultured without an enrichment step, whereas swabs were cultured with an enrichment step. This difference in procedures may also be a reason for the differences in the results. One of the objectives of the present study was to investigate whether broth enrichment cultures detect more GRE than direct plating to estimate more closely their real prevalence. Enrichment of cultures entails a greater amount of work and may not be readily feasible for routine surveillance; about 14% of broth cultures produced false-positive results; i.e., isolates were not identified as enterococci. However, it should be realized that direct plating fails to detect a considerable number of carriers of GRE, even though they carry GRE in lower numbers.
The prevalence of fecal carriers of GRE in different population groups varies considerably. It was 3.5% among 636 patients whose samples were screened without enrichment in another Belgian hospital in 1993 (18), 5% among 97 hospitalized patients in Oxford, United Kingdom, in 1992 (21), 2% among 184 general practice patients in the United Kingdom, and 28% among 40 healthy subjects never exposed to glycopeptides in Charleroi, Belgium, in 1992 (39). Most of the latter were also present in small numbers: <50 CFU/g of feces. The higher prevalence of GRE among dialysis patients is striking, although not statistically significant. However, this may result from the small number of dialysis patients examined. The higher prevalence of carriers of GRE among patients in a renal unit has been mentioned previously by Jordens et al. (22), who found 15% carriers among 73 patients. It should be stressed that in all these studies different methodologies were followed; in particular, the concentration of vancomycin incorporated into the screening medium varied from 8 to 16 to 20 μg/ml.
Our finding of E. gallinarum as the most prevalent species is in contrast to the findings of most studies, in which E. faecium is the most common enterococcal species in human feces (16, 18, 28). The high prevalence of E. gallinarum does not result from the introduction of the broth enrichment since approximately half of all enterococci were obtained by this procedure. The prevalence of the different enterococcal species in the present study is also in contrast to that of the 472 enterococci responsible for infections in Belgian hospitals, among which E. faecalis represented 89.4% of the isolates (38), illustrating the higher virulence of this species. The E. faecium and E. faecalis isolates detected in the present study all carried the vanA gene, as did the clinical isolates of GRE from Belgium (38).
The results of the MIC determinations by the E test are in line with the detection of the genotypes found by Tenover et al. (35).
High-level resistance to streptomycin was common (14 [34.8%] of the 43 VRE), but only two isolates were highly resistant to gentamicin. This is in agreement with the results from a previous study, in which more than 50% of the E. faecalis isolates from a Belgian multicenter study showed high-level resistance to streptomycin, whereas only about 10% showed high-level resistance to gentamicin (38), reflecting the distribution of aminoglycoside resistance worldwide (19).
Application of DNA fingerprinting techniques gave insight into the degree of genetic diversity. Two groups of E. faeciumcomposed of four and two isolates, respectively, were clonally related, as were a pair of isolates of E. faecalis. From the available evidence, it appears that the four-member cluster of E. faecium and the E. faecalis cluster were probably community acquired, while a two-member cluster of E. faeciummay have been the result of nosocomial spread. All other GRE produced different PFGE patterns, suggesting polyclonality.
Among the hospitalized patients, no risk factors for colonization with GRE could be identified. In contrast to many studies (1, 2, 17, 28, 36, 41, 43), there was no difference between carriers and controls in terms of the length of the present or a previous hospital stay and antibiotic use. Colonized patients had been admitted to the hospital for longer periods than the average patient, but the lengths of the hospitalizations were not significantly different from those for the control patients, in the same wards, who were free of GRE. There was no obvious relation between colonization with GRE among hospitalized patients and previous antimicrobial therapy, and vancomycin was administered to only one patient and one control from whom no GRE were cultured. These data are in agreement with the results of the study by Gordts et al. (18). Because, in addition, a number of carriers of vanA GRE isolates were hospitalized for only 1 or 2 days before the screening, our study suggests that the GRE in these patients were probably part of the intestinal flora before hospitalization and acquired in the community; this was also already suggested by Jordens et al. (22).
In conclusion, our data show that the rate of isolation of GRE may be seriously underestimated in the absence of a broth enrichment step. Most likely, strains detected after broth enrichment are present in low numbers and may be missed if the enrichment step is omited. Our study suggests that prevalence studies in which only direct culture techniques on rectal swabs are used may be inappropriate.
FOOTNOTES
- Received 25 September 1998.
- Returned for modification 23 November 1998.
- Accepted 28 January 1999.
- Copyright © 1999 American Society for Microbiology
