INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

In the last decade, enterococci became the second most frequently reported cause of surgical wound infections and nosocomial urinary tract infections and the third most frequently reported cause of bacteremia (23, 33). Ampicillin and aminoglycosides have been considered the drugs of choice for treatment of serious enterococcal infections (5). The number of enterococci that are resistant to ampicillin and aminoglycosides has increased (4, 14, 15, 31, 32). The glycopeptide antibiotics vancomycin and teicoplanin are important substances for treatment of severe hospital infections. Diseases caused by enterococci which are resistant to the -lactam antimicrobial agent ampicillin and aminoglycoside antibiotics can be treated only with glycopeptides (5, 24). Unfortunately, resistance to vancomycin and teicoplanin (both of which are glycopeptides) has also been reported. In the United States, the Centers for Disease Control and Prevention reported that there was a 20-fold increase in the occurrence of vancomycin-resistant enterococci (VRE) associated with nosocomial infections from 1989 to 1990 (6). Therefore, the therapeutic use of glycopeptides should be restricted; otherwise, infections caused by such multiresistant enterococci will become untreatable (3, 7, 17, 35). The high level of resistance to glycopeptides (MIC, >512 mg/liter) is inducible and is encoded by the vanA gene cluster, which is carried on transposons similar to or related to Tn1546 (34). Transfer of resistance can occur via conjugative plasmids. Conjugation experiments have shown that the rates of transfer between vanA donors and vanA-negative recipient strains of enterococci range from 2 × 10⁷ to >2 × 10⁴ (20).

The source of glycopeptide-resistant enterococci is not known. One possibility is that these organisms are spread via the food chain. Some data have indicated that raw poultry and raw minced meat may harbor VRE (2, 18, 37). In this context, adding the glycopeptide avoparcin, a mycelial product of Streptomyces candidus, to animal feed was thought to be responsible for the development of glycopeptide resistance in enterococci in animals (2, 18). Because this possibility could not be ruled out, the use of avoparcin as a feed additive was banned in Germany in January 1996 and in the whole European Union (EU) later. Furthermore, the possible use of resistant Enterococcus faecium strains as human probiotic strains and as starter cultures for a number of cheese products has been discussed as a possible source of VRE (37). The increased use of glycopeptide antibiotics in hospitals and the widespread use of glycopeptide antibiotics for treatment of patients may be another source of VRE (35).

In this study the incidence of VRE in fresh minced beef and pork was investigated to determine the importance of raw minced meat as a possible vector for the transfer of VRE from animals to humans. Fresh minced pork and beef samples from an EU-admitted meat-processing plant with a broad catchment area in Germany were tested to obtain representative data. In order to assess the potential human risk through the food chain, we also investigated the occurrence and resistance of enterococci to a broad range of antimicrobial substances that are used either as therapeutic agents or as feed additives.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Pork and beef skeletal muscles (but no cheek meat or sticking or diaphragm muscles) were comminuted with a mincer and packaged for retail markets. The minced meat investigated was produced in an EU-admitted meat-processing plant in Berlin, Germany. The meat originated from different abattoirs throughout Germany. Enterococci were isolated from 275 meat samples collected on 55 days (5 days per week) between May and November 1996; all of the enterococci isolated, including VRE, were identified to the species level. In the period between November 1996 and April 1997 the occurrence of enterococci and VRE was determined, but the organisms were not identified to the species level. The antibiotic susceptibility patterns of all isolates were determined.

Hygiene control. To minimize the risk of contaminating the meat with enterococci from the environment, processing hygiene and disinfection were performed strictly according to the EU guidelines (12). The results of these measures were checked with a wet swab-dry swab technique (9, 25). Briefly, a wet swab and a dry swab were moved over a defined area of each surface in the meat-processing plant. Subsequently, the swabs were spread onto Enterococcosel agar (ECSA) (Becton Dickinson) and ECSA containing 32 mg of vancomycin (Sigma Chemical Co., St. Louis, Mo.) per liter (ECSA-VA) for quantitative detection. After enrichment by incubation in dilution broth for 24 h at 37 ± 0.5°C, a qualitative procedure to detect enterococci on ECSA and ECSA-VA was performed.

Total aerobic mesophilic colony counts for the minced meat samples. The minced meat samples were stored at 4 ± 0.5°C immediately after they were produced, and they were analyzed 3 to 6 h after processing. The number of aerobic mesophilic CFU per gram in each meat sample was determined by the surface plating method (drop plating) as described in the German national regulations (8). The resulting values were transformed into log values, and statistical parameters were calculated with a computer program (BIAS 5.0; Ackermann, Frankfurt am Main, Germany).

Isolation of enterococci. A 25-g portion of minced meat from each sample (the total weight of each sample was 125 g) was placed in 225 ml of buffered peptone water (BPW) and homogenized with a stomacher for 1 min. Then 1 ml of the diluted sample was plated onto ECSA. The agar plates were incubated for 24 h at 37 ± 0.5°C aerobically. From each ECSA plate, which represented one meat sample, two typical black colonies on the underlying black agar with colony diameters of ca. 1 mm were isolated randomly and used for further investigation. Up to 10 colonies per production day (batch) were isolated (five samples per day). The total enterococcal colony count was also determined.

Isolation of VRE. A direct culturing method was used for quantitative detection of VRE in the meat samples. The procedure used was the procedure used to determine the total enterococcal counts with ECSA-VA instead of ECSA (see above).

Species identification. The Enterococcus species were identified with a Rapid ID 32 Strep identification kit (bioMérieux, Marcy l'Etoile, France). As recommended by Nusser (28) and Reuter (30), growth at 10 ± 0.5, 45 ± 0.5, and 50 ± 0.5°C and growth in the presence of 6.5% NaCl were examined, and the potassium tellurite reaction was performed to confirm the results. The catalase reaction test and an experiment to determine the Lancefield serotype (Slidex StreptoD; bioMérieux) were also performed (26). To distinguish Enterococcus gallinarum and Enterococcus casseliflavus from the E. faecium group and from Enterococcus faecalis, acidification of methyl--D-glucopyranoside (Sigma) was examined for all isolates (10).

Determination of the MICs. The MICs of 19 antibiotics and three feed additives were determined by using the recommendations of the National Committee for Clinical Laboratory Standards (NCCLS) (27) and the broth microdilution method. Microtiter plates containing the test substances in cation-adjusted Mueller-Hinton broth supplemented with 3% lysed horse blood (PML Microbiologicals, Portland, Oreg.) were used. The microtiter plates were allowed to reach room temperature before inoculation. A 0.5-McFarland unit suspension, prepared as recommended by the NCCLS (27), was diluted 1:30 and homogenized, and 10 µl was inoculated into each well, which resulted in a final inoculum of 5 × 10⁵ CFU/ml. The test results were determined visually after incubation for 16 to 20 h at 37 ± 0.5°C by using the instructions given by the NCCLS (27). The MIC ranges described by the NCCLS (27) were suitable for enterococci in most cases. When no specific ranges for enterococci were mentioned, the ranges for gram-positive bacteria were used. To interpret the results obtained with the three feed additives tested and the new antibiotic substance LY 333328, the interpretive guidelines for related substances were used (i.e., the vancomycin guidelines were used for avoparcin and LY 333328, and the erythromycin guidelines were used for the macrolides tylosin and virginiamycin).

-Lactamase test. Production of -lactamase by isolated enterococci was investigated with the BR66 identification stick test, which is a nitrocefin-based test (Oxoid). S. aureus ATCC 29213 was used as a positive control, and E. faecalis ATCC 29212 was used as a negative control.

vanA gene. All VRE strains were grown for 24 h at 37 ± 0.5°C in BPW. Isolation of bacterial DNA, amplification of a vanA gene fragment by PCR, and agarose gel electrophoresis were performed as described by Ausubel et al. (1) and Klare et al. (18). After the cells were lysed with 0.5% sodium dodecyl sulfate, proteins were removed by digestion with proteinase K (Boehringer, Mannheim, Germany). A cetyltrimethylammonium bromide solution (Sigma) was used to remove cell wall debris, polysaccharides, and the remaining proteins. To extract the bacterial DNA for amplification of the vanA gene, the DNA was precipitated with isopropanol and transferred to a fresh tube containing 70% ethanol. The DNA concentration was determined with a model TKO 100 DNA fluorometer (Hoefer). The following oligonucleotides were used as primers for amplification of the 377-bp fragment of the vanA gene: vanA 1 (5'-TCT GCA ATA GAG ATA GCC GC-3'; vanA sequence positions 443 to 462 [11]) and vanA 2 (5'-GG AGT AGC TAT CCC AGC ATT-3'; vanA sequence positions 819 to 800) (primers were obtained from TIB MOL BIOL, Berlin, Germany). Each PCR mixture contained 10 µl of Mg²⁺-free buffer, 3 mM MgCl₂, each deoxynucleoside triphosphate at a concentration of 200 µM, 0.4 µl of primer vanA 1 (50 pmol/100 µl), 0.44 µl of primer vanA 2 (50 pmol/100 µl), 1.25 µl of polymerase (2.5 U), 50 µl of mineral oil, 20 ng of template DNA, and enough water to bring the volume to 100 µl. The solutions were obtained from a PrimeZyme polymerase kit (catalog no. 100-652; Biometra, Göttingen, Germany). A thermal cycler (model TRIO Thermoblock; Biometra) was programmed for 30 cycles; cycle 1 consisted of 94°C for 3 min, 55°C for 1 min, and 72°C for 1 min, cycles 2 through 29 each consisted of 94°C for 1 min, 55°C for 30 s, and 72°C for 30 s, and cycle 30 consisted of 94°C for 1 min, 55°C for 1 min, and 72°C for 4 min. Gel electrophoresis was performed for 90 min in a 1.4% agarose gel (Low EEO; Appligene) at 100 V. E. faecium 64/3, a vanA-negative strain, and vanA-positive strain E. faecium 70/90 were used as reference strains.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Hygiene and disinfection in the meat-processing plant. The surfaces in the meat-processing plant were examined to determine if they were contaminated with enterococci immediately before the minced meat was processed. Enterococci were detected in 7 of 48 surface samples only when the enrichment procedure was used. No VRE were detected on the premises after the disinfection procedure was performed. The disinfection measures were performed as recommended in the EU guidelines (25) (48 surfaces were tested).

Microecological results. The total bacterial counts in the minced meat samples were between 2.7 × 10³ and 9.3 × 10⁶ CFU/g. Enterococci were present at concentrations between 0.5 × 10¹ and 7.1 × 10² CFU/g. VRE at a concentration of 10 CFU per g of minced meat were found in 3 of 555 samples (0.5%) when no enrichment procedure was performed. In 46 of 555 samples (8.3%) VRE at concentrations ranging from 1 to 9 CFU per g of minced meat were found after overnight enrichment in BPW. A total of 509 meat samples (91.7%) were VRE negative (Fig. 1). The occurrence of VRE was distributed nearly equally over the investigation period.

View larger version (57K): [in this window] [in a new window]   FIG. 1.   Occurrence and distribution of VRE in fresh minced meat samples after direct plating and after a 24-h enrichment procedure.

MIC determination. The MIC data obtained in vitro with 22 antibiotic substances and the enterococci isolated from non-VRE-selective ECSA plates (n = 209) revealed that all of the isolates were susceptible to ampicillin and only one Enterococcus strain was susceptible to methicillin. The combination of amoxicillin and clavulanic acid was very effective; no strain showed resistance. A total of 134 of the 209 isolates were resistant to or exhibited intermediate reactions with ceftriaxone, and 192 of the 209 isolates were resistant to or exhibited intermediate reactions with cephalothin. The carbapenem substance imipenem was effective against 172 enterococci; 9 strains were resistant to imipenem, and 28 exhibited intermediate reactions.

vanA gene. The vanA-specific 377-bp fragment was amplified from the genomic DNA of all VRE. The results obtained for some strains are shown in Fig. 2. The vanA gene fragments of E. faecium, E. faecalis, E. durans, E. gallinarum, and E. hirae were amplified from both beef and pork isolates. The positive control strain E. faecium 70/90 contained the typical 377-bp fragment, and the negative control strain E. faecium 64/3 did not.

View larger version (48K): [in this window] [in a new window]   FIG. 2.   Detection of the vanA gene in VRE from minced beef and pork. (A) Lanes 1 and 14, DNA molecular weight marker VI (catalog no. 1062590; Boehringer Mannheim) with bands at 2,176, 1,766, 1,230, 1,033, 653, 517, 453, 394, 298, 234-220, and 154 bp; lane 2, blank; lane 3, E. faecium 70/90 (positive control); lane 4, E. faecium 64/3 (negative control); lane 5, E. faecium VRE 3; lane 6, E. faecium VRE 4; lane 7, E. faecium VRE 5; lane 8, E. faecium VRE 6; lane 9, E. faecium VRE 7; lane 10, E. faecalis VRE 9; lane 11, E. durans VRE 10; lane 12, E. faecalis VRE 11; lane 13, E. faecium VRE 12. (B) Lanes 1 and 14, DNA molecular weight marker VI; lane 2, blank; lane 3, E. faecium 70/90 (positive control); lane 4, E. faecium 64/3 (negative control); lane 5, E. faecalis VRE 13; lane 6, E. gallinarum VRE 14; lane 7, E. gallinarum VRE 15; lane 8, E. faecium VRE 16; lane 9, E. gallinarum VRE 17; lane 10, E. faecium VRE 18; lane 11, E. hirae VRE 19; lane 12, E. faecalis VRE 20; lane 13, E. faecalis VRE 21. The main band of the positive control strain and the VRE strains represents the vanA-specific 377-bp fragment.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Minced meat was used in this investigation because the production process creates a risk that microbiological contamination will occur. The incidence of VRE in minced meat should therefore be greater than the incidence of VRE in other meat products. As the meat investigated originated from different counties in Germany, we assumed that our conclusions were representative. The disinfection of the meat plant was assessed strictly according to the EU guidelines. The possibility that the meat was contaminated with enterococci, especially VRE, due to insufficient hygiene practices could be excluded. The use of plastic gloves by workers avoided contamination of the meat with enterococci and VRE from humans. Other investigations of meat obtained from retail outlets (38) did not differentiate between human and environmental origins of the VRE detected.

The use of ECSA and BPW resulted in greater recovery of enterococci than the use of other media (29); therefore, these media were considered suitable for isolation and enrichment of enterococci and VRE.

Enterococci were detected in nearly all samples of minced meat at levels ranging from 0.5 × 10¹ to 7.1 × 10² CFU/g. VRE were detected directly in only 3 of the 555 samples investigated (0.5%). VRE were found after overnight enrichment in 48 of the 555 samples (8.3%). These findings suggest that VRE occurred in very low numbers and only sporadically in minced meat. The use of an enrichment procedure is essential for detection of VRE. Using pork samples obtained from retail outlets, Wegener et al. (38) found that up to 27% of the samples were VRE positive after enrichment. The considerable difference between these results and our results may be explained by possible human contamination of the meat in the retail outlets. However, the difference may be due to less effective disinfection in a meat-processing plant. The enterococci may have persisted in the factory, and they could be of environmental or animal origin. Therefore, the difference between the results of Wegener and our results cannot be satisfactorily explained.

Enterococci isolated from non-VRE-selective ECSA plates were identified to the species level. The results showed that E. faecalis was the dominant species in minced meat (87% of the isolates). The percentage of E. faecium was about 4%, and the percentages of the other species identified were less than 4%. The VRE isolates identified to the species level belonged mainly to E. faecium or the E. faecium species group. Wegener et al. (38) found that food contaminated with vancomycin-resistant E. faecium strains was an important possible source of infections in the community, based on their study in Denmark and a study performed in Belgium (13). Thus, the low incidence of E. faecium in minced meat in our study may indicate that food contamination is not an important source of VRE in Germany.

The susceptibility of enterococci isolated from non-VRE-selective ECSA to the glycopeptides vancomycin, teicoplanin, and avoparcin is consistent with the findings of Knudtson and Hartmann (21). In 50 pork samples Knudtson and Hartmann did not find any enterococcal isolates which exhibited multiple resistance to these antibiotics. However, Klare et al. (18) isolated five vancomycin-resistant E. faecium strains with vanA resistance from meat samples obtained from 13 different butcher shops. In view to our findings, the possibility that microbiological contamination from the environment occurred could not be ruled out. These findings (18) cannot represent the microbiological situation in fresh meat in general. In an earlier study Klare et al. (19) isolated E. faecium strains from stools of patients in intensive care units in a hospital in Berlin, Germany. They found that 49 of 52 VRE strains exhibited intermediate resistance or were resistant to ampicillin. This indicates that VRE from meat have different antibiotic resistance patterns than VRE responsible for untreatable infections in hospitals. These findings are supported by the results of other authors who detected 11% vancomycin-resistant E. faecium in stools of healthy volunteers (37) and found about 3.5% VRE-positive patients in a hospital in which no VRE had been isolated previously (13). This indicates a natural reservoir for VRE.

All Enterococcus strains, including VRE isolated from raw minced meat in this study, exhibited susceptibility to the antibiotic ampicillin, which is currently used to treat human enterococcal infections. In the case of infections caused by enterococci from meat, patients can be treated with ampicillin as the antibiotic of choice. The use of glycopeptides is not necessary. For the future, a new antibiotic, LY 333328, seems to be a substance that could be used to treat infections caused by enterococci resistant to ampicillin, aminoglycosides, and the "old" glycopeptides vancomycin and teicoplanin. The results of other investigations support our finding that enterococci isolated from clinical samples exhibit different resistance patterns than enterococci isolated from meat (16). Furthermore, a recent study of the genetic similarity of clinical and food enterococci isolated in Germany demonstrated with pulsed-field gel electrophoresis that these strains are not genetically related (22). van den Braak et al. (36) found VRE in fecal samples from vegetarians more often than they found VRE in fecal samples from nonvegetarians (9.7 versus 4.7%). These authors indicated that there may be no association between meat consumption and intestinal colonization with VRE. We concluded, therefore, that enterococci isolated from raw minced meat cannot be considered the main source of untreatable nosocomial infections of humans. However, these strains can acquire additional resistance in hospitals and can have the same antibiotic resistance patterns as the clinical strains, even when they are not genetically related. In addition, there are a lot of other virulence factors sensu stricto, including hemolysin, adhesin, aggregation substances, and proteases, that should be considered. In light of the observations mentioned above, we concluded that it is very likely that clinical strains in Germany originate mainly from sources other than meat.

The results of the direct culturing method and the enrichment procedure indicate that low levels of contamination of minced meat with enterococci and with VRE occur when minced meat is processed under prescribed hygienic rules according to EU regulations. The enterococcal flora of raw minced meat is dominated by E. faecalis. E. faecium, which harbors the vanA resistance gene in most cases, was detected only in a small percentage of samples. The level of VRE detected by direct culturing (0.5%) is very low compared with the levels of other food-borne pathogens, such as S. aureus. The level of VRE isolates after enrichment (8.3%) was the worst case observed in this study. In the enrichment procedure even one colony in 25 g of minced meat would have resulted in a positive test.

This study started a few months after avoparcin was banned in Germany. Our data could represent the baseline incidence of VRE occurring in the absence of avoparcin. However, in a previous study performed with a smaller number of samples we found VRE in 1 of 45 samples from nine batches and no VRE in 180 samples from 36 batches (22). The study was performed in 1995 before avoparcin was banned in Germany (22). Therefore, the results of this study performed within 1 year after the German avoparcin ban suggest that the use of avoparcin may not have been responsible for the development of vancomycin resistance in animals. Hence, it seems that glycopeptide resistance is ubiquitous and may not be connected with the use of avoparcin.