OD2006: Persistence of antimicrobial resistant bacteria on livestock farms.
Annex10: Do antimicrobial-resistant organisms persist on livestock farms after the withdrawal of antimicrobial drugs? A review of the literature.
Introduction
Prior to the advent of the widespread clinical use of penicillin in the 1940s, the work of Alexander Fleming had already demonstrated that if one varied either the concentration of the drug used or the conditions for bacterial growth, then one could induce the appearance of penicillin-resistant bacteria52. The first published reports of clinical isolates of penicillin-resistant Staphylococcus aureus appeared as early as 194443. For 2 decades, however, science appeared to be ahead of the game as a variety of both naturally-derived and synthetically manufactured antibacterial agents were discovered and patented. At this time the treatment of bacterial diseases was revolutionised and, during the 1950s, antibiotics also became widely used in veterinary medicine and agriculture. In addition to their use as veterinary therapeutic agents, the incorporation of sub-therapeutic doses of antimicrobial drugs into animal feed has been found to enhance livestock productivity26;30;46. However, as the isolation of multiple-resistant strains of Salmonella from diseased calves became more frequent the UK government commissioned a Joint Committee to report on the use of antibiotics in animal husbandry and veterinary medicine70. The resultant Swann Report, published in 1969, concluded that administering antibiotics to animals at sub-therapeutic levels did constitute a risk to human and animal health. However the Committee surmised that this risk could be minimised by only allowing drugs of low therapeutic value to be used as antimicrobial growth promoters42. In 1974, in response to this report, the EU banned the use of penicillin and tetracycline as agricultural growth promoting agents (EU Directive 70/524/EEU). During the 1990s European studies demonstrated that vancomycin-resistant Enterococcus faecium (VRE) could be isolated from animals on farms that were administering a related glycopeptide drug, avoparcin, as a growth enhancing agent1;45. Molecular typing of VRE isolates obtained from human, animal and environmental sources revealed a high degree of heterogeneity with, nonetheless, a low level of genetic similarities between strains taken from different origins16. However, more detailed analyses of the transposons (Tn1546) and gene clusters responsible for vancomycin-resistance in Enterococcus faecium, found that identical Tn1546-types were present in isolates from humans and those from animals, implying that horizontal gene-transfer between human and animal E. faecium was occurring and/or that there was a common vancomycin-resistance gene pool39;66;72;73. Consequently, the European Union (EU) prohibited the administration of avoparcin to livestock in 1997. Subsequent to the avoparcin ban, in-line with recommendations from the World Health Organisation (WHO), two further EU directives will result in the implementation of a total ban on the use of antimicrobial drugs as digestive enhancers or growth promoters within Europe by the year 2006.
In contrast, although avoparcin has never been licensed for use within the American agricultural industry, 19 other antibiotic drugs are currently licensed for use as growth-promoting and prophylactic agents across the USA. This list still includes therapeutic agents such as penicillin, chlortetracycline and streptomycin62. Therefore, the enforced decrease in the use of antimicrobial growth promoting agents within the EU provides the opportunity to assess whether the effect of withdrawing the use of in-feed antimicrobial agents on the patterns of resistance observed amongst bacteria on farms and within the wider community.
Resistant enterococci
Denmark and Norway prohibited the use of avoparcin in 1995, 2-years prior to the total EU ban. Concurrently, the Danish Integrated Antimicrobial Resistance Monitoring and Research Programme (DANMAP) was established in order to provide information on the association between the use of antimicrobials and the occurrence of resistance, and to compare resistance in bacteria from food, livestock and humans14. A paper from 1999 reported that the proportion of Danish poultry derived Enterococcus faecium (E. faecium) isolates that were resistant to glycopeptides had significantly reduced from ~81% in 1995 to ~5% in 1998. However, a similar trend was not observed for E. faecium isolates obtained from Danish pigs, for which throughout this period 18-30% were consistently resistant to glycopeptides13. The authors postulated that these differences could be due to differences in management of the two farm types. That is the broiler farms were largely operating all-in-all-out policies whilst the pig farms were generally continuous production systems and were rarely, if ever, depopulated. The pig farmers, therefore, may have been less able to break the cycle of resistant organisms transferring from older to younger stock. Co-resistance patterns for the E. faecium isolates in this study were obtained using a microbroth dilution technique. This study showed that majority of Vancomycin-Resistant E. faecium (VRE) obtained from pigs were also resistant to narrow-spectrum penicillins, tetracycline and the macrolide drug tylosin. In fact, these drugs were in common use in the Danish pig industry at that time, with tylosin being widely used both as a therapeutic agent and as a growth promoting feed-additive. Further work led to the observation that Pulsed-Field Gel Electrophoresis (PFGE) patterns obtained for the pig isolates pointed to the dissemination of a single E. faecium clone. Gene-transfer studies found that the transfer of vanA-mediated vancomycin-resistance between donor and recipient strains also led to the simultaneous transfer of ermB-mediated macrolide-resistance, thus indicating that the two resistances were present on the same mobile genetic element3. Therefore, it appears that the widespread use of tylosin in the Danish pig industry may have contributed to the maintenance of VRE after the withdrawal of avoparcin. Indeed, in subsequent years the proportion of E. faecium isolated from Danish pigs that were VRE fell from 20% to 6% over the same time period that the use of tylosin in the Danish pig industry fell from 62,000kg to 1,800kg per year7.
Seemingly in-contrast to the situation in Denmark, a Norwegian study reported a continuing high prevalence of VRE on broiler and turkey farms in 1998, 3-years after the withdrawal of avoparcin from the Norwegian livestock industry19. However, the two studies are not directly comparable due to the utilisation of markedly different sampling techniques. The DANMAP programme collects one caecal or cloacal sample from randomly selected herds or flocks at slaughter, the samples are plated onto enterococci-selective media and resultant colonies showing typical E. faecium morphology are submitted for identification confirmation and susceptibility testing8. In contrast, in the Norwegian study, the poultry farmers participating collected and submitted a sample of fresh faecal material taken from the poultry-house floors. In the laboratory, the samples were plated onto selective media containing 50mg/l vancomycin. Therefore, the Norwegian methods detect the presence of VRE within a sample, whilst the Danish methods measure the proportion of E. faecium that are VRE. Consequently, the Danish results are likely to reflect a decrease in the proportion of the intestinal E. faecium population of poultry that were glycopeptide-resistant. Whereas the Norwegian direct-plating methods revealed that VRE were still present within the Norwegian broiler farms, but supplied limited information about the level of VRE persistence on these farms. In fact, other research carried out in Denmark that have used direct-plating methods showed that 5-years after the withdrawal of avoparcin, VRE were still present in 104 of 140 broiler flocks previously exposed to avoparcin. Furthermore, on the VRE-positive farms on average 65% of birds were excreting VRE35. Recent work in the UK has also found VRE persisting on some broiler farms 5-years after the EU ban on avoparcin use31.
The important question raised by these observations is, why are VRE persisting on some farms long after the cessation of use of, what was assumed to be, the primary-selective antimicrobial agent?
One hypothesis is that the genes coding for these resistances occur on multiple-resistance plasmids and therefore the use of other antibacterial agents, such as disinfectants, may indirectly select for resistances to pharmacologically unrelated compounds? In Norway, avoparcin was the only antimicrobial growth promoter that was ever widely used in the poultry industry, and after the ban in 1995 other agents were not generally substituted (with the exception of a small quantity of zinc bacitracin)19. Norwegian poultry farms also tend to operate all-in-all-out systems whereby the houses remain empty for 10-30 days after cleaning, prior to the arrival of the next flock of birds. Studies of empty and cleaned broiler houses have demonstrated that VRE can survive some cleaning and disinfecting routines20;36 (Garcia-Migura, unpublished data). In a Norwegian study the feed storage bins, feeding machinery and feed troughs yielded VRE-positive samples. Suggesting that contaminated feed equipment may have been responsible for the colonisation of birds by VRE soon after their arrival on to the farms20. A study in Denmark found that VRE could be isolated from broiler house floors after cleaning and disinfecting. PFGE profiles were obtained for isolates collected from the cleaned environment and from faecal samples from the preceding and subsequent flocks. The results showed that, within a given house, highly similar profiles were present between faecal and environmental isolates, indicating that resistant-clones were persisting between flocks36. Interestingly, isolates collected from different houses on the same farm did not appear to be related, therefore the persistence of resistant clones appears to be at the house-level, not the farm-level.
The withdrawal of avoparcin across the EU provided an opportunity to investigate the potential effects of withdrawing an antimicrobial drug from agricultural use across a wider sector of the community than the farms themselves. Several studies have looked at levels of VRE isolated from poultry meat after the avoparcin ban, however the variety of sampling and isolation techniques that have been employed by the different research groups act to complicate direct comparisons between their findings. A single Norwegian study using direct plating methods, as detailed above, found that 3-years after the withdrawal of avoparcin 30% of the broiler and turkey carcasses were still positive for VRE21. A series of comparative studies conducted in Germany, showed a fall in VRE-contaminated thawing liquid of frozen broiler carcasses from 100% in 1994 to 25% in 1997, less than 2-years after the removal of avoparcin44. An Italian study also suggests a fall from 15% VRE-positive samples before the ban, to 8% 18-months after the ban59. In the German study, a survey of the faecal carriage of VRE in non-hospitalised residents was also undertaken. In 1994,12% of people sampled were positive for faecal-VRE. In 1996, 6-months after the German withdrawal of avoparcin, this fell to 6%, followed by a further fall to 3% in 199744. A short communication from The Netherlands also details a fall in faecal-VRE carriage by healthy suburban human volunteers from 12% in 1997 to 6% in 199968.
The DANMAP surveillance programmes have enabled Denmark to follow resistance trends in relation to the consumption of specific antibiotics. Using the DANMAP sampling techniques, as described above, changes in the annual quantity of drugs that have been administered have been followed by an alteration in the percentage of resistant enterococci isolated by the program. For instance, the amount of virginiamycin, a growth-enhancing streptogramin drug, that food animals consumed increased four-fold to more than 10,000kg between 1995 & 1997. During this period the proportion of E. faecium isolated from broiler chickens that were resistant to virginiamycin (VMRE) also increased from 27% to 66%. In January 1998, Denmark prohibited the use of all antimicrobial growth-promoting agents, and subsequently the proportion of VMRE that were isolated in 2000 fell to 34%7. As explained previously, this is likely to reflect a fall in the proportion of enterococci that are virginiamycin-resistant, and may not be synonymous with a decrease in the prevalence of VMRE at the farm-level. In fact, Danish studies of avilamycin-resistant E. faecium (ARE) have illustrated this point. In 1998 isolates were obtained from 8 broiler farms that had administered the growth-promoter avilamycin during 1996/97 and 10 farms that had not administered avilamycin at that time. ARE were found on all exposed farms, and 7 of the 10 unexposed farms. However 72% of isolates from the exposed farms were ARE compared with 23% from the unexposed farms4. Unfortunately the authors do not describe the use, or otherwise, of avilamycin on the unexposed farms prior to 1996. However these studies have shown field-evidence for the selection of resistant E. faecium by the extremely low doses of drugs used for growth promoting purposes4.
The DANMAP results for erythromycin-resistant E. faecium (ERE) on Danish broiler and pig farms are also interesting. In 1997, 76% of E. faecium isolates from broilers were ERE, this decreased to 28% in 1999 and 13% in 2000. It is plausible that the decrease observed in ERE isolations was associated with the decrease in the use of virginiamycin by the poultry industry. Streptogramin drugs, such as virginiamycin, are composed of 2 molecules, termed group A and group B. Whilst the group B molecules are structurally unrelated to the macrolide and lincosamide drug families, all 3 groups do share the same mechanism of action. Therefore the macrolide-lincosamide-streptogramin resistance gene ermB jointly facilitates resistance to erythromycin and virginiamycin, and thus use of virginiamycin may co-select for ERE. The percentage ERE for pigs fell from 93% in 1996 to 47% in 1999 and then remained at this figure in 2000. In the late 1990s macrolide use was extremely high in Denmark, with peak use occurring during 1996-97 when 60-70,000kg active drug/year were administered to food animals. However during 1998/99, the Danish livestock industry voluntarily decreased the use of growth promoters and the use of macrolides, largely tylosin, dropped steeply to under 20,000kg/year. Since this time macrolide use in swine has been stable at ~10,000kg/year. Interestingly the annual proportion of isolated E. faecium from pigsthat are ERE continues to drop and in 2003 was ~20%10.
Further work in Denmark has used generalised linear mixed effects models to fit the DANMAP microbiological resistance results with farm-level information collected from the ante-mortem health certificates of broiler flocks28. This showed that the probability of selecting an E. faecium isolate resistant to avilamycin, erythromycin or virginiamycin was between 0.84 and 0.92 from flocks where avilamycin or virginiamycin was incorporated into the food. After the withdrawal of avilamycin the probability of selecting an avilamycin-resistant E. faecium fell steadily with time reaching 0.09 after 3+ years. For erythromycin and virginiamycin-resistant E. faecium the probabilities fell for the first 3-years post-removal to 0.31 and 0.35 respectively. After 4-years, the probabilities for a large number of the farms continued to fall to less than 0.2 for both drugs. However, this was not the case for all farms. On a sizeable number of farms little or no further decreases in probability were seen after 3-years, thus implying that factors other than growth promoter use were selecting for the persistence of growth-promoter resistant E. faecium on these farms. The authors report that at this time a change also occurred in the resistance-patterns of virginiamycin-resistant E. faecium. In 1999, 48% VMRE isolates were simultaneously resistant to erythromycin; in 2000 this fell to just 8% whilst the percentage of isolates that were concurrently virginiamycin-penicillin resistant rose from 50% to 84%. The total kilograms of active penicillin-drugs administered to all species of food animal stayed steady over that period at ~20,000-23,000kg/yr9. The models of Emborg et al. also suggested that the number of times that a growth promoter had been administered on a farm prior to its withdrawal had no effect on the probability of selecting resistant E. faecium isolates.
Other hypotheses for the persistence of antimicrobial resistance have included the co-selection of resistance-plasmids due to the incorporation of heavy metals in animal feed. Work investigating resistance in enterococci from 3 different countries, demonstrated that copper-resistance was common in E. faecium and E. faecalis from Danish and Spanish pigs in 1998-1999. At that time, it was common practice within these countries to feed diets containing up to 175ppm of copper to young pigs. In comparison enterococci isolates collected from slaughterhouses in Sweden, where the maximum acceptable level of copper sulphate in diets for pigs of all ages was 35ppm, showed very little copper-resistance. In Sweden, 5. These results imply that incorporating high levels of copper compounds in diets may select for copper resistant bacteria. All copper-resistant enterococci in this study were carrying a transferable tcrB gene. This gene has previously been associated with copper resistance in E. faecium, where it was found to be linked to macrolide and glycopeptide resistance34. Thus it is plausible that high levels of copper in the diet of pigs could also select for the persistence of vancomycin-resistant enterococci.