Antimicrobial susceptibility of hazard analysis critical control point Escherichia coli isolates from federally inspected beef processing plants in Alberta, Saskatchewan, and Ontario

Abstract

A survey to estimate the prevalence of antimicrobial resistance in Escherichia coli was conducted in 7 Canadian federally inspected processing plants during 2001. Escherichia coli isolates were recovered during routine hazard analysis critical control point sampling from beef carcasses and trim and subsequently tested for their antimicrobial susceptibility by using susceptibility panels.

Of the 2653 isolates analyzed, 68% were sensitive to all 18 antimicrobials tested. For 14 of the 18 antimicrobials evaluated, the percentage of resistant isolates was ≤ 1. Twenty-five percent of the isolates were resistant to tetracycline, 9% to sulfamethoxazole, 7% to streptomycin, and 3% to ampicillin. Multiple resistance was found in 12% of the isolates, with 7% showing resistance to 2 antimicrobials, 2% to 3 antimicrobials, 2% to 4 antimicrobials, and 1% to 5 or more antimicrobials. Forty-five different antimicrobial resistance patterns were observed. The reasons for the development of the antimicrobial resistance were not investigated in this study.

This study was useful as a pilot to help to develop a national antimicrobial resistance surveillance program in Canada. This study indicates that laboratory standardization is possible for consistent results across the country and that the indicator organism, E. coli, is fairly easy to obtain for surveillance but Salmonella are not, due to their low prevalence in beef.

Introduction

One of the specific food safety priorities in the primary production area is the investigation of antimicrobial resistant bacteria in livestock. The most common pathway by which humans can acquire pathogenic bacteria that are resistant to antimicrobials from livestock is presumed to be contaminated meat (1). Outbreaks of foodborne disease in humans caused by multiresistant bacteria have occurred occasionally (2). However, the frequency of these occurrences is not known with certainty.

Although the precise magnitude of human health risk associated with the use of antibiotic drugs in food animal production is unknown, scientists generally agree that resistance is a problem in both human and veterinary medicine (1). The World Health Organisation (WHO) has stated that information is limited as to the prevalence and spread of resistance in zoonotic bacteria or indicator organisms (2). Suitable indicator organisms are those that are frequently isolated from healthy animals, in food and humans. They are commensals in animals and humans and are often used as indicators of food hygiene. Examples of indicator organisms are Escherichia coli, Enterococcus faecium, and Enteroccoccus faecalis. These commensal organisms are also believed to be reservoirs of resistance determinants for pathogens of animals and humans. One of the WHO's goals for countries around the world is to improve the surveillance of antimicrobial resistance of bacteria in food animals and food of animal origin in order to: 1) promote prudent and judicious use of antimicrobials, 2) enable informed decision-making, 3) guide prescription practices, 4) encourage standardization of laboratory techniques, 5) identify areas for more detailed investigation, and 6) promote collaboration among the various sectors involved (2). Currently, the US government is monitoring the antimicrobial susceptibility of Salmonella spp. from beef cattle at processing and from clinical isolates at veterinary diagnostic laboratories as part of their antimicrobial resistance surveillance program called National Antimicrobial Resistance Monitoring System (NARMS) (3).

In the future, it is likely that surveillance of bacteria collected in this fashion will be extremely useful in detecting emerging trends in resistance in bacteria arising from agriculture and contaminated food supplies. For example, the United States Food and Drug Administration (FDA) is including post approval surveillance of anti biotic resistance as one of the criteria that it will use in determining the continuing safety of antibiotics approved for use in agriculture (4). Other countries will probably follow suit. It is critical that practical, reliable methodologies are developed for this surveillance, so that correct inferences are drawn from the data generated from these programs in the future. It is also important that pilot studies of this type are conducted to aid in the development of sustainable, ongoing resistance surveillance.

The purpose of the study reported here was to monitor the antimicrobial susceptibility of bacterial isolates cultured during hazard analysis critical control point (HACCP) monitoring in federal beef processing plants in Alberta, Saskatchewan, and Ontario. This information can be used to help develop a national antimicrobial resistance surveillance system that will to provide temporal trends of antimicrobial susceptibility, facilitate identification of resistance in bacteria in animals as it arises, evaluate interventions designed to reduce anti microbial resistance, provide timely information to veterinarians, prolong the life span of approved drugs by promoting prudent and judicious use, and identify areas for detailed investigation.

Materials and methods

Sample collection at federal beef processing plants

For 1 y from January 2001 to 2002, microbial isolates from HACCP sampling were collected from federally inspected processing plants in Alberta; Saskatchewan; and Ontario; 1 plant in Saskatchewan, 3 plants in Alberta, and 3 plants in Ontario.

Each plant was visited twice each month, 2 wk apart, to collect culture plates (E. coli Count Petrifilm; 3M Canada Company, London, Ontario) for E. coli. The culture plates did not contain any antimicrobials. Samples for culture were taken from beef carcasses and trim, and collected as part of the plants' routine HACCP program. Plants did not record data on culture plates to identify whether the samples were from carcasses or trim, feeder cattle or cull cows and bulls, beef or dairy carcasses, and farm of origin. Thus, these data were not known.

Based on the budget and manpower available for the study, the sample size was restricted to a maximum of 100 bacterial isolates per laboratory per month. The number of study samples (culture plates) collected from a particular plant was proportional to the percentage of cattle slaughtered by that plant relative to the total number slaughtered by all plants included in the study. A collection schedule was established to ensure that the target number of culture plates from each plant was drawn from the most recent HACCP samples.

Laboratory analysis

Samples — Refrigerated culture containing bacterial colonies from federally inspected processing plants were delivered within 24 h to the respective laboratories in Saskatchewan, Alberta, and Ontario. The plates were sorted, labeled, and stored at 4°C until bacteriological analyses were conducted (24 to 48 h).

All 3 laboratories standardized procedures by conducting a comparison of 10 common isolates from Saskatchewan for quality control to try to ensure that the laboratories' testing procedures were comparable.

Bacteriology: Alberta and Saskatchewan — A representative number of presumptive E. coli colonies (exhibiting a purple/blue color with gas production) were picked from the culture plates and streaked onto MacConkey agar plates to verify lactose fermentation and onto blood agar plates (BAP) to ensure purity. One isolated colony from the BAP that demonstrated a single colonial morphology was screened by using E. coli medium with 4-methylumbelliferyl-β-D-glucuronide, (EC Medium with MUG; Difco Laboratories, Becton Dickinson Microbiology Systems, Sparks, Maryland, USA) and indole, methyl red Voges Proskauer, and Simmons citrate (IMViC) tests according to protocols outlined by Health Protection Branch Methods of Microbiological Analysis of Foods (5). Colonies exhibiting reactions indicative of E. coli were subsequently streaked onto fresh BAP to ensure purity, prior to storage at -70°C. More extensive biochemical reactions were determined by using analytical profile index (API) test strips (bioMerieux Canada, St-Laurent, Quebec) for those colonies exhibiting atypical reactions. Colonies exhibiting atypical reactions were not used for subsequent antibiotic sensitivity testing. Colonies exhibiting typical reactions and confirmed as E. coli were inoculated into sheep blood and stored at -70°C. Prior to antibiotic sensitivity testing, frozen cultures were thawed, streaked onto BAP, and incubated at 35°C ± 1°C for 24 h.

Antimicrobial sensitivities: Alberta and Saskatchewan — Antibiotic sensitivities were determined by using the technical product information protocols (Sensititre 18 Hour Susceptibility System; AccuMed International, East Grinstead, West Sussex, United Kingdom) supplied by the manufacturer. Briefly, isolated E. coli colonies from BAP were suspended in 2 mL of sterile purified water to obtain a homogeneous solution that was brought to a 0.5 McFarland turbidity standard. Ten microliters of the bacterial suspension were inoculated into Mueller-Hinton (MH) broth (non cation adjusted). Fifty microliters of the inoculated MH broth were manually transferred into each well of the antimicrobial susceptibility plate by using an 8-channel multipipetter. The MH broth was subsequently streaked onto fresh BAP to verify purity of the culture. After incubating the susceptibility plates at 35°C ± 1°C for 18 to 24 h, test results were manually read by using a mirror reader (API UniScept Plus; Plainview, New York, USA). The minimal inhibitory concentration (MIC) was recorded as the lowest concentration of antimicrobial that inhibited visible growth, except for sulphonamides, in which case the MIC was read as an 80% to 90% decrease in growth compared with the growth of the microorganism in the control wells. The MIC was recorded as less than, or equal to, the lowest concentration of antimicrobial on the plate for those wells where no growth was observed and greater than the highest concentration of antimicrobial if all wells exhibited visible growth.

Bacteriology: Ontario — Presumptive E. coli colonies were picked from the culture plates and streaked for isolated colonies onto MacConkey agar. Lactose fermenting colonies were picked and inoculated into tryptic soy broth (TSB). After incubation for 1 h at 35°C, enough TSB culture was moved to fill the wells of the repli- analyzer. Twenty-three plates with biochemical tests, plus 1 control plate, were run through the repli-analyzer. The biochemical reactions included bile tolerance, colistin, glucose, lactose, lysine, ornithine, citrate, hydrogen sulfide, rhamnose, mannitol, arabinose, sorbitol, sucrose, xylose, trehalose, maltose, cellobiose, esculin, adonitol, maltose, arginine, cetrimide, and colistin/nalidixic acid. Colonies identified as E. coli were sub-cultured onto MacConkey agar plates, mixed with skimmed milk plus glucose, dextrose, sucrose, and bovine serum, and stored at -70°C.

Antimicrobial sensitivities: Ontario — Cultures were streaked from freezer cultures onto Mueller Hinton agar (MHA) and incubated overnight at 35°C to 37°C. From the above cultures, a single isolated colony was picked and streaked onto MHA for growth (growth plate) and incubated for 18 to 24 h at 37°C.

After incubation, a small amount of culture from the growth plate was selected using a sterile cotton swab, and suspended in 5 mL of demineralized water to obtain a homogeneous solution, which was brought to a 0.5 McFarland turbidity standard. After mixing the sample, 10 μl from the water preparation was transferred to 10 mL of cation-adjusted Mueller Hinton broth with TES buffer (MHB, Sensititre) and mixed, using a vortex mixer. A MacConkey agar plate was inoculated for a purity check. The cap on the MHB was replaced with a disposable dosing head, the sample was mixed, using a vortex mixer, and the tube was loaded onto the autoinoculator.

The custom panel (CMV6CNCD) was placed into the plate holder on the autoinoculator. The autoinoculator dispensed 50 μL/well. After inoculation, the plate was covered with a plate sealer and placed into the automated reading and incubation system (ARIS). Each plate was incubated at 37°C for 18 h and then read by the autoreader. All data and results were sent to the computer for calculation and interpretation

Data analyses

Data from all plants were entered into a spreadsheet and the proportion of resistance calculated by antimicrobial. Data were visually crosschecked among the different laboratories to see if similar MIC values were used to determine the antimicrobial resistance classification.

The federal slaughter plants participating in the study were exporting product to the United States and were implementing HACCP plans verified by the Canadian Food Inspection Agency and United States Department of Agriculture, suggesting statistically sound sampling procedures. Culture plates for the study were collected from the federal plant systematically every 2 wk during the 1-year period of the study. Given this sampling method, it was assumed that the sampling was random for all practical purposes and confidence intervals for the point estimates of antimicrobial resistance were calculated.

The data were not stratified and analyzed by province (laboratory) in order to maintain the confidentiality of the packing plants, which was a prerequisite for plant participation in the study. Since the plants did not record the source of the culture plate (carcass or trim, beef or dairy carcass, farm), the data could not be analyzed by these factors. It is highly unlikely that any single farm was sampled repeatedly during the year, given the sampling procedures described above.

Results

During the 1 y of sampling, 2653 isolates from Alberta, Saskatchewan, and Ontario were tested for antimicrobial sensitivity. The resistance breakpoints are presented in Table 1 and follow the NARMS 1999 methodology (3).

Table 1.

When 10 E. coli isolates were compared for antimicrobial sensitivity among the 3 laboratories at the end of the study, it was found that the laboratory in 1 province had read the sulfamethoxazole and cephalothin wells differently than the other 2 laboratories. Following discussion and agreement on reading and interpretation of results among the 3 laboratories, the sulfamethoxazole and cephalothin data from the 1 laboratory were removed from the analysis. As well, the results from this laboratory were not used to determine the number of antimicrobials to which resistance was found or in determining resistance patterns.

The proportion of resistant isolates is described in Table 2. The proportion of resistant isolates was ≤ 1% for 14 of the 18 antimicrobials tested. The 4 antimicrobials to which the greatest resistance was observed were tetracycline (25%), sulfamethoxozole (9%), streptomycin (7%), and ampicillin (3%).

Table 2.

The number of antimicrobials to which resistance was shown is described in Table 3. Sixty-eight percent of the isolates were sensitive to all 18 antimicrobials tested. Multiple resistance was found in 12% of the isolates, with 7% showing resistance to 2 antimicrobials, 2% to 3 antimicrobials, 2% to 4 antimicrobials, and 1% to 5 or more antimicrobials. Of the 18 isolates that were resistant to 5 or more antimicrobials, 2 isolates were resistant to 6 antimicrobials, 2 isolates were resistant to 7 antimicrobials, and 1 isolate was resistant to 11 antimicrobials.

Table 3.

The antimicrobial resistance patterns that were observed are listed in Table 4. In total, 45 different antimicrobial resistance patterns were observed in 1817 isolates.

Table 4.

Discussion

The results of this study provide some baseline antimicrobial susceptibility data for Canadian beef processing cattle (Alberta, Saskatchewan, Ontario), such as those available through the NARMS program in the United States (3). The difference that exists between the data collected here and the data for NARMS is that the NARMS program monitors the antimicrobial susceptibility of Salmonella spp. and Campylobacter spp. isolates (3), while we studied generic E. coli. We chose to use the indicator organism E. coli because there were not enough Salmonella spp. isolates from beef carcasses and trim in Canada to provide reliable data to monitor antimicrobial susceptibility trends in beef (personnel communication, federal packers), and Campylobacter spp. were not monitored in HACCP programs at Canadian processing plants. Both Salmonella spp. and Campylobacter spp. are zoonotic agents and would represent a more direct link to potential food safety concerns related to antimicrobial resistant bacteria than would generic E. coli.

While generic E. coli may not be pathogenic to humans, certain subtypes are pathogens (E. coli O157:H7). The occurrence of these specific pathogenic subtypes of E. coli on beef carcasses and trim is too low to make them useful monitoring organisms for antimicrobial susceptibility.

We did not monitor antimicrobial resistance in HACCP isolates from provincial processing plants, or from federal plants in other provinces, which may affect how representative our results are for Canadian beef. It would be logistically difficult to collect microbial samples at provincial processing plants, since at the time of this study, very few provincial plants had an HACCP system in place. Federal processing plants that export products must have an HACCP system with microbiological monitoring of process controls. The federal beef processing plants in Canada where we collected samples in this study account for over 85% of beef processing (personal communication, CANFAX); thus, the data here are most likely representative of Canadian beef.

When comparing the results of the 10 isolates used for quality control among the 3 laboratories, it was observed that there were some minor differences in MIC values reported for the same isolate and antimicrobial, and, occasionally, some differences in classification for sensitive and resistant status, when using the same MIC value. For example, initially there was a difference in resistance classification for tetracycline between the 3 laboratories, where one laboratory used 8 μg/mL as the breakpoint and the other 2 laboratories used 16 μg/mL as the breakpoint. Additionally, there were some differences noted in using the <= sign when reporting MIC values. Following discussions among the 3 laboratories, agreement was reached on classifications for sensitive and resistant. These findings indicate that for a national survey to be valuable, laboratories need to standardize their laboratory testing procedures, including methodology, training, reading, reporting, and interpretation of data prior to the initiation of a study.

The occurrence of resistant E. coli was low for most of the antimicrobials tested, with ≤ 1% resistant bacteria to 14 of 18 antibiotics. However, some of the E. coli isolated from beef carcasses and trim were resistant to various antimicrobials, including tetracycline and sulfamethoxazole. Similar low resistance estimates have been observed in E. coli isolated from cattle in Denmark (6) and in Salmonella spp. isolated from slaughter cattle in NARMS studies (3). However, direct comparisons between E. coli and Salmonella spp. should not be made, because different bacteria were monitored.

It is interesting to note that resistance was observed to chloramphenicol in 1% of the samples. Chloramphenicol has been banned from livestock use in Canada for over 10 y, although florfenicol is used in beef cattle. It is not known whether the observed resistance to chloramphenicol is mediated by a gene encoding resistance to florfenicol (7), acquired by horizontal transmission of genes from other sources, such as water contaminated from human sewage, due to illegal use of chloramphenicol, or due to coselection by linkage of genes encoding resistance to chloramphenicol and some other drug that is being used in cattle.

While the overall occurrence of resistant bacteria was low, and generic E. coli are not typically pathogenic to humans, the occurrence of antimicrobial resistant bacteria on carcasses and trim may pose a risk to human health (1,2). There may have been pathogenic E. coli isolated, but determination of virulence factors was not part of this study. Beef contaminated with antimicrobial resistant bacteria may not be properly handled and cooked, and antimicrobial resistance genes from generic E. coli could be transferred to other pathogenic bacteria within the human intestinal tract, which may lead to disease that is difficult to treat (1). Resistance may also be transferred from commensal to zoonotic organisms in the bovine gut and lead to human illness (1). The magnitude of such risk to human health has not yet been determined. Further research is warranted to determine a) whether indicator organisms are indicative of resistance in pathogenic bacteria and b) the impact of transference of resistance from indicator bacteria in cattle to pathogenic bacteria in humans.

This study was not designed to determine how the antimicrobial resistance developed in the E. coli isolates. Thus, one cannot say from this study that the resistance observed is due to antimicrobial usage in cattle and that reducing antimicrobial use would reduce the level of resistance.

This study established a baseline for studying the occurrence of antimicrobial susceptibility in E. coli isolates from beef carcasses and trim in federally inspected beef processing plants in Canada. The information from this study can contribute to the development of a surveillance program in Canada for antimicrobial resistant bacteria from processed beef cattle. Additionally, this information can be used in food safety risk assessment modeling and in identifying areas for further detailed investigation, such as on-farm risk factors that lead to the development of resistance.