Extended-Spectrum β-Lactamase Genes of Escherichia coli in Chicken Meat and Humans, the Netherlands

We determined the prevalence and characteristics of extended-spectrum β-lactamase (ESBL) genes of Enterobacteriaceae in retail chicken meat and humans in the Netherlands. Raw meat samples were obtained, and simultaneous cross-sectional surveys of fecal carriage were performed in 4 hospitals in the same area. Human blood cultures from these hospitals that contained ESBL genes were included. A high prevalence of ESBL genes was found in chicken meat (79.8%). Genetic analysis showed that the predominant ESBL genes in chicken meat and human rectal swab specimens were identical. These genes were also frequently found in human blood culture isolates. Typing results of Escherichia coli strains showed a high degree of similarity with strains from meat and humans. These findings suggest that the abundant presence of ESBL genes in the food chain may have a profound effect on future treatment options for a wide range of infections caused by gram-negative bacteria.

I nfections with drug-resistant bacteria are associated with higher rates of illnesses and deaths, which have a serious effect on costs of health care (1,2). During the past decade, drug resistance in Enterobacteriaceae has increased dramatically worldwide. This increase has been caused mainly by an increased prevalence of extended-spectrum β-lactamase (ESBL)-producing Enterobacteriaceae (3,4) and has increased the use of last-resort antimicrobial drugs (i.e., carbapenems).
ESBL genes are located on plasmids that can be easily transferred between and within bacterial species. Some ESBL genes are mutant derivatives of established plasmid-mediated β-lactamases (e.g., bla TEM/SHV ), and others are mobilized from environmental bacteria (e.g., bla CTX-M ). During the 1990s, most reports on ESBL genes concerned bla TEM/SHV types, which were related to cross-infections in hospitals. However, the recent global increase has been caused mainly by bla CTX-M -type genes. The epidemiology of ESBL genes is changing rapidly and shows marked geographic differences in distribution of genotypes of bla CTX-M β-lactamases (5). In the United States, the most prevalent drug resistance gene in humans is currently bla CTX-M-15 , which is often associated with a widely distributed variant of Escherichia coli O:25b, sequence type 131 (ST131). Bacteria containing ESBL genes are currently a common cause of infections originating in community-dwelling persons without a history of hospitalization, and these organisms can then be introduced into hospitals (6-9).
Fecal carriage of ESBL genes has been identifi ed as the major reservoir in the environment, but the original source of this colonization has not been clearly identifi ed. Because bacterial species that carry ESBL genes are normal inhabitants of the gastrointestinal tract, food is a potential source of them. The presence of ESBL genes has been clearly documented in food-production animals, especially chickens (10,11). Drug resistance in animals is caused mainly by the large amount of antimicrobial drugs used in food production. In addition to their presence in farm animals, ESBL genes have been found in retail meat (12,13). A recent survey of broiler chickens in Great Britain found that bla CTX-M-1 was the most prevalent ESBL gene (14). Although ESBL genes in food-production animals pose a potential threat to humans, Randall et al. concluded that drug resistance genes in chickens (bla CTX-M-1 ) differed from the drug resistance genes most frequently found in humans (bla CTX-M-15 ) (14).
In the Netherlands, use of antimicrobial drugs and associated drug resistance in humans is among the lowest in Europe (15). Paradoxally, use of antimicrobial drugs in food-production animals in this country is among the highest in Europe (16). Therefore, the Netherlands provides a good setting to monitor spread of drug resistance from an animal reservoir into the human population. This spread was recently exemplifi ed by emergence of livestock-associated methicillin-resistant Staphylococcus aureus in pigs and veal calves. This was fi rst reported in the Netherlands in 2004 and has now been reported worldwide (17). The aim of our study was to determine the prevalence of ESBL genes in retail meat and hospitalized patients in the Netherlands and to compare ESBL genes and bacterial strains involved.

Meat Survey
During August 17-October 30, 2009, a prospective observational study was conducted in 4 hospitals in the southern part of the Netherlands (Sint Elisabeth Hospital in Tilburg, Twee Steden Hospital in Tilburg, Amphia Hospital in Breda, and Lievensberg Hospital in Bergen op Zoom). Randomly chosen packages of meat from major grocery stores in the region of the 4 participating hospitals were included. Each sample was derived from a different package containing raw and unspiced meat. All samples were incubated for 16-18 h at 37°C in 15 mL of tryptic soy broth (TSB). Subsequently, 100 μL of the initial broth sample was transferred into a second sample of TSB broth containing 8 mg/L vancomycin and 0.25 mg/L cefotaxime. After overnight incubation, 10 μL of the broth was placed on a chromogenic agar plate selective for ESBL (bioMérieux, Marcy l'Etoile, France), and the plates were incubated overnight. Colonies with distinct morphologic appearance were further characterized.
Species and resistance patterns of oxidase-negative, gram-negative rods were determined by using the Vitek2 System (bioMérieux). Phenotypic confi rmation of ESBL was performed by using Etest (bioMérieux) for all isolates. A combination of ceftazidime, cefotaxime, and cefepime with and without clavulanic acid was used. If the MIC of >1 of the cephalosporins showed an 8-fold reduction in combination with clavulanic acid, the isolate was considered to be an ESBL producer. If the Etest result was inconclusive, a combination disk diffusion test (Rosco, Taastrup, Denmark) was performed. All tests were performed and interpreted according to the National Guideline for the Laboratory Detection of ESBL (18).

Fecal Carriage Survey
In the 4 hospitals in the same area were the meat had been obtained, 2 consecutive prevalence surveys were performed 3 weeks apart (November 1-December 20, 2009) as part of each hospital's infection control program. Patients who had positive results for ESBL at the fi rst sampling were excluded from subsequent sampling. These hospitals provide care to ≈1 million persons. Rectal swabs specimens were obtained from all patients admitted to the internal medicine, surgery, urology, pulmonology, and intensive care unit wards. Patients <18 years of age or those who had a colostomy were excluded. Rectal swab specimens were incubated in TSB broth cultures by using broth enrichment as described for the meat samples. Duration of hospitalization on the day of the survey was recorded for all participating patients.

Blood Cultures
All E. coli and Klebsiella spp. resistant to cefotaxime, including all strains presumably producing ESBL on the basis of microbiologic results, and isolated from clinical blood cultures, were obtained during July 2008-December 2009 from the 4 study hospitals. Confi rmation of ESBL genes was performed as described for meat samples. Blood culture isolates were obtained from individual patients.

Genetic Characterization of Drug Resistance Genes
Characterization of drug resistance genes in all strains that were phenotypically ESBL producers was conducted 2 ways. First, we tested all isolates for bla CTX-M by using denaturing high-performance liquid chromatography as described (19). Second, we screened for ESBL genes by using a micro-array (Check-Points, Wageningen, the Netherlands) that was designed to detect single nucleotide polymorphisms in essential bla TEM and bla SHV genes, variant genes, and bla CTX-M group genes (bla CTX-M-1 , bla CTX-M-2 , bla CTX-M-9 , and bla CTX-M-8/25 ) . The procedure has been reported (20). Subsequently, sequencing was performed to further specify ESBL genotypes. On the basis of microarray results, bla SHV , bla TEM , and bla CTX-M genes were amplifi ed by PCR and specifi c primers. PCR amplicons were selected and sequenced after purifi cation (Agencourt Ampure; Beckman Coulter, Leiden, the Netherlands). Sequence analysis and alignments were performed by using Bionumerics 6.01 software (Applied Maths, Sint-Martens-Latem, Belgium), the BLAST program (http:// blast.ncbi.nlm.nih.gov/Blast.cgi), and information from the Lahey Clinic (Burlington, MA, USA) (www.lahey.org/ studies). If results of denaturing high-perfomance liquid chromatography and DNA sequence were discordant, the DNA sequence was used as the correct result.

Multilocus Sequence Typing of E. coli Strains
All E. coli isolates from meat, rectal swab specimens, and blood cultures were typed by using multilocus sequence typing (MLST) as described by Wirth et al. (21). If patient or meat samples contained >1 morphologically distinct ESBL-producing E. coli strain, all strains were included in the MLST analysis.

Statistical Analysis
Data were analyzed by using SPSS version 18 software (SPSS, IBM, Somers, NY, USA). Univariate analysis was performed for calculation of difference in prevalences by using the χ 2 test. One sample could contain >1 strain because morphologically distinct colonies with different drug resistance genes or a different MLST result were all included in the fi nal analysis.

Fecal Carriage Survey
A total of 927 rectal swab specimens were obtained from 876 patients (461 male patients and 415 female patients, mean ± SD age 65.7 ± 16.8 years). Phenotypic screening identifi ed 59 patients as possibly being infected with ESBL-producing E. coli. Confi rmatory test results for ESBL genes were inconclusive for 2 samples and excluded (no inconclusive results were obtained with meat or blood cultures). A total of 45 (4.9%) samples contained confi rmed ESBL genes.

Blood Cultures
Thirty-one clinical blood cultures suspected of containing ESBL genes on the basis of phenotypic screening were available for further analysis. Genetic characterization confi rmed that 23 (74.2%) samples contained ESBL genes.

Epidemiology of Patients and Drug Resistance Genes
When we compared the most prevalent drug resistance genes in 346 patients who had been hospitalized <48 h at the time of screening with 581 patients who had been hospitalized >48 h, prevalence was similar for bla CTX-M-1 (2.3% and 2.4%, respectively) and for TEM-52 (0.6% and 0.7%, respectively) in the 2 groups. Prevalence of bla CTX-M-15 was 4× higher in the group that was hospitalized longer (0.3% and 1.2%, respectively; p = 0.27). There was only 1 possible cluster of cases (same resistance gene at the same time at the same ward), which involved 3 patients with bla CTX-M-15 . One patient was infected with E. coli and 2 were infected with K. pneumoniae.

MLST of E. coli
MLST results of 158 ESBL-positive E. coli strains isolated from chicken meat, other meat types, rectal swab specimens, and blood cultures are shown in Figure 2. E. coli containing ESBL genes showed a heterogeneous population that contained several clusters. Most clusters contained strains isolated from meat and humans. All but 1 of the ESBL-producing strains from other meat types clustered with strains from chicken meat. Twenty-fi ve (56.8%) of 44 strains from rectal swab specimens and 9 (56.3%) of 16 strains from blood cultures clustered with strains from chicken meat.
MLST results for strains with bla CTX-M-1 , bla CTX-M-15 , and bla TEM-52 are shown in Figure 2. Genotypes bla CTX-M-1 and bla TEM-52 showed a heterogeneous population and clusters containing strains from humans and meat. E. coli harboring bla CTX-M-15 was found less frequently, and no clusters with human-derived and meat-derived strains were observed. The widely disseminated ST131 clone was found in human samples only: 4 times in combination with bla CTX-M-15 , twice with bla CTX-M-14 , and once with bla SHV-12 .

Discussion
ESBL genes were found in a high (79.8%) proportion of retail chicken meat samples in the Netherlands. A comparison of ESBL-producing Enterobacteriaceae derived from meat and hospitalized patients showed a high degree of similarity of resistance genes and MLST patterns. Genotype bla CTX-M-1 was the most frequent drug resistance gene in chicken meat and humans and the second most frequent in blood cultures. Other meat types contained similar drug resistance genes, but the prevalence of ESBL genes was much lower. It is unclear whether ESBL genes in other meat types are related to a reservoir in food-production animals or contamination at meatprocessing facilities. An extensive reservoir of ESBL genes on farms was repeatedly shown in poultry (10,11). Our fi ndings suggest a relationship between contamination of chicken meat with drug-resistant bacteria and appearance of ESBL genes in humans in the Netherlands. This relationship was further supported by genomic comparison of strains detected in chicken meat with those detected in human fecal specimens. MLST showed that most E. coli strains harboring bla CTX-M-1 or bla TEM-52 from humans and meat belong to clusters containing strains from both sources. These fi ndings suggest a relationship between contamination of chicken meat and appearance of ESBL genes in humans in the Netherlands.
The high prevalence of ESBL genes in chicken meat is consistent with fi ndings of other investigators. Doi et al. reported that 67% of retail meat samples in Seville, Spain, contained ESBL or ESBL-like resistance genes (12). A survey of imported raw chicken in the United Kingdom reported ESBL genes in 10 of 27 samples (13). The authors concluded that ESBL genes in meat pose a potential threat to humans, but that the most prevalent ESBL genotype in humans in the United Kingdom (bla CTX-M-15 ) was not found in imported meat.
In our study, we found a high degree of similarity between drug resistance genes in humans and retail meat. A possible explanation for this fi nding is that in the Netherlands, where drug resistance in bacterial isolates in humans is less frequent (15) and cross-transmission in hospitals is controlled effectively (22), the role of acquiring drug-resistant strains from food is more easily detected. In addition, the Netherlands is one of the highest users of antimicrobial agents in food-production animals (16), which results in high rates of drug resistance among these animals. A report by the Veterinary Antibiotic Usage and Resistance Surveillance Working Group showed that cefotaximeresistant E. coli in poultry meat in the Netherlands has emerged since 2005 (23). This fi nding coincides with the increase in third-generation cephalosporin-resistance in E. coli and K. pneumonia bacteria in invasive infections in humans reported by the European Antimicrobial Resistance Surveillance Network (24).
We performed a prevalence survey among hospitalized patients to determine the size and nature of the reservoir of ESBL genes. Nearly 5% of all hospitalized patients were carriers of ESBL genes. It is diffi cult to put this rate into context because no screening studies have been conducted in the Netherlands. A large study in Chicago, Illinois, USA, found that during 2000-2005, the rate of ESBL gene carriage among high-risk, hospitalized patients increased from 1.3% to 3.2%, and bacteremia developed in 8.5% of all previously identifi ed ESBL gene carriers during hospitalization (25). In Spain, an increase was also observed in fecal carriage in hospitalized patients from 0.3% in 1991 to 11.8% in 2003 (7).
The rate we observed in 2009 in the Netherlands was lower than rates in Spain in 2003, which would be expected because of low antimicrobial drug resistance rates in the Netherlands (24). Conversely, a prevalence of 5% extrapolated to the Dutch population in 2011 (16,700,000 inhabitants) would indicate that currently >800,000 persons in the Netherlands are colonized with ESBL-producing bacteria. Because use of antimicrobial drugs in the health care setting in the Netherlands is low and has not changed during the past decade, alternative factors for increased drug resistance are not known. An indication for the role of a community reservoir is the prevalence of drug resistance genes in patients who had been hospitalized <48 h and those hospitalized >48 h. Drug resistance genes that are associated with a proposed food reservoir (26) (bla CTX-M-1 and bla TEM-52 ) were already present at hospitalization. However, bla CTX-M-15 , which is reported to be associated with health care settings (26), had a higher, albeit not signifi cant, prevalence in the group who had been in the hospital >48 h.
Considering what is known about the epidemiology of E. coli, the abundance of ESBL genes in chicken meat is a likely explanation for current fi ndings in humans. Although there are extensive campaigns promoting safe handling of chicken meat during processing, enteric pathogens are frequently transferred to humans and pose a continuous public health threat (27). Johnson et al. studied geographically and temporally matched E. coli isolates from humans and poultry (28). Drug-susceptible E. coli isolates from humans differed from drug-resistant isolates from humans and from isolates in poultry irrespective of their drug resistance pattern. Drug-susceptible isolates from poultry were similar to drug-resistant isolates of poultry and humans. Their conclusion was that drugresistant human fecal E. coli isolates likely originate from poultry, whereas drug-resistant E. coli isolates from poultry likely originate from susceptible precursors in poultry. In vitro experimental support for our hypothesis comes from a recent study showing transfer of a bla TEM-52 -carrying plasmid from an avian E. coli strain to 2 human E. coli strains in a continuous fl ow culture model (29). Emergence of ESBL genes in poultry has been associated with use of third-generation cephalosporins (particularly ceftiofur) in chickens. In Canada, a strong correlation was found between incidence of ceftiofurresistant Samonella enterica serovar Heidelberg in humans and retail chickens (30). Use of ceftiofur in animals was stopped voluntarily, and ceftiofur resistance rates subsequently decreased in retail chicken meat and humans. After partial reintroduction of this drug, resistance rates in poultry and humans increased, providing further evidence for a zoonotic source of ESBL genes.
Overrepresentation of ST131 in blood cultures confi rms the virulent properties attributed to this clone. Typically, ST131 is found in association with bla CTX-M-15 . In our survey, ST131 strains were associated with 3 drug resistance genes. This fi nding indicates that this virulent clone also acquires other drug resistance genes. We did not fi nd ST131 in the chicken meat samples, which is reassuring at this time. However, other studies have recently identifi ed ST131 in poultry and retail meat (31,32). These fi ndings confi rm that virulent clones of E. coli are capable of crossing species barriers between humans and animals. In addition, mobile drug resistance genes also cross this barrier and are likely to accelerate dissemination of drug resistance between animals and humans.
We conclude that the high rate of ESBL contamination of retail chicken meat in the Netherlands, which involves many of the same ESBL genes present in colonized and infected humans, is a plausible source of the recent increase of ESBL genes in the Netherlands. The similarity of E. coli strains and predominant drug resistance genes in meat and humans provides circumstantial evidence for an animal reservoir for a substantial part of ESBL genes found in humans. The threat of the high rate of antimicrobial drug use in food-production animals and associated emergence of drug resistance in zoonotic pathogens has been recognized for decades. Our group and others found that most samples of retail chicken meat contain transmissible drug resistance genes in bacterial species that are part of the normal human intestinal fl ora. This fi nding may have a profound effect on future treatment options for a wide range of infections with gram-negative bacteria.