4 Bacterial Antibiotic Resistance and Human Health

4.1 Summary

To compare antibiotic resistant bacterial strains it is important that strains are thoroughly characterised. It is recommended to use multiple phenotypic and genotypic characterisation methods. Many articles presenting antibiotic resistance data just contain prevalence data and can only be used to compose a view of the presence of resistant bacteria in animals, humans or meat. Data about resistance (MICs) cannot always be compared, because different methods of isolation and testing of bacteria are being used. The given percentages of resistant bacteria are not always very accurate. Only articles that use multiple genetic methods to examine strains are reliable when claims of transfer are being put forward. In summary, the following points put forward in this chapter are:

• The acquiring of resistant bacteria by humans in general consists of two distinct routes (the human therapeutic antibiotic use being of dominant importance):
  • Use of antibiotics by humans can cause resistant bacteria -present in the intestines- to emerge
  • Resistant bacteria or resistant genes can be acquired by contact with exogenic sources containing resistant bacteria. (This route incorporates the dissemination route from animals harbouring antibiotic resistant bacterial strains)
• The animal-human link might comprise of several dissemina-tion routes:
  • Direct contact with an animal or animal faeces
  • The consumption of meat or fish
  • The consumption of vegetables or fruit
  • Human to human spread
  • Contact with water containing faeces
• Two scenario's can be drawn when the spread of bacterial resistance from animals to humans is discussed:
  • When a resistant bacteria of animal origin is able to colonise the human gut the resistance in effect has been transferred from the animal to the human. However, a bacterium has to survive the stomach. When it enters the intestines it has to be able to multiply in sufficient amounts before it truly colonises the human. The time the specific bacterial strain is able to stay in the intestines determines if it is a transient passenger or a permanent resident. Resistance genes present on the bacterial chromosome, on plasmids or on transposons can be expressed during transient or permanent colonisation. Chances increase, however, when the stay is prolonged.
  • The second possibility is that resistance genes are being transferred from bacteria present in meat or animals to bacteria that are commonly found in human intestines. Transfer of genes can take place in the gut or prior to ingestion, after which the resistant bacteria may be able to colonise the human gut. Resistance traits present on plasmids or on transposons have a chance of being transferred to another bacterium. Genes present on the bacterial chromosome, but not on a transposon, have a much lower chance of being transferred.
• The categories of people at higher risk of being infected and circumstances that increase risk are as follows:
  • immunocompromised persons (elderly patients, ill neonates, etc.)
  • patients subjected to surgical operations, people with burns
  • patients with breathing devices, catheters and drains
  • type of ward: Intensive Care, renal units, hematology ward, surgical ward
  • transfer of patients between wards or between hospitals
  • use of antibiotics (cephalosporines)
  • prolonged stay in the hospital
  • hygienic measures taken (or not taken) in the hospital
• The following criteria in the elucidation of possible bacterial antibiotic resistance transfer from animals to humans are essential:
  • The bacterial strain and/or the resistance trait present in the human should be identical to a bacterial strain and/or resistance trait present in the meat consumed. (In many reports meat samples and human samples are compared that might be totally unrelated.)
  • The exact source of the resistant bacteria needs to be elucidated. Otherwise a possible relationship between antibiotic usage in animal feed and resistant bacteria in humans cannot be confirmed.
  • Recent use of antibiotics by the people concerned needs to be established and documented.
  • When meat samples are examined, one needs to be sure that resistant bacteria found are not the result of contamination during processing, preparing or transport of the meat.
  • Typing methods for identifying bacteria have to be specific enough to detect small differences between bacterial strains and their resistance traits.
• The elucidation of factual bacterial resistance transfer from animals to humans (and from humans to humans) requires fenotypic and genotypic methods that are the most discriminatory, so that even small differences between strains and resistance traits can be distinguished.
• For a complete analysis also the presence of plasmids and/or other resistance traits present should be determined.
• Below the techniques for isolation and characterisation of resistant bacterial strains are listed.
  • Isolation of resistant strains and phenotypic characterisation
    • Enrichment cultures (different concentrations of the antibiotic used in selection)
    • Determination of species and sub-species: biochemical methods an ready-to-use kits
    • Determination of Minimal Inhibitory Concentrations (MIC; different methods and media)
    • Determination of MICs of single or multiple antibiotics
  • Genotypic characterisation
    • Pulsed Field Gel Electrophoresis (PFGE) of chromosomal DNA (digested by SmaI)
    • Region amplification within a (resistance) gene by PCR; regions within In: intergenic amplification by PCR
    • Ribotyping
    • Long PCRs of transposons
    • Conjugational studies

4.2 Introduction

The acquiring of resistant bacteria by humans in general consists of two distinct routes:

Scheme 4.2.1. Modes of acquiring resistant bacteria.

In essence the different proportions (if there are more than one) adding to the total human bacterial resistance needs to be determined which can be depicted as follows:

Figure 4.2.1. Possible sources of human bacterial antibiotics resistance.

It is important to know which sources cause a risk for human health and to compare the risks of these different sources. A pool of resistant bacteria in animals might in theory be a risk when these bacteria or their resistance traits are transferred to humans or human bacteria. The question that now rises is whether resistance genes easily spread to other hosts in the same environment or to populations in other environments.

4.3 'Spreading the Disease'

Below a tentative scheme is presented with possible dissemination routes of resistant bacteria or their resistance traits from animals to humans:

Figure 4.3.1. Possible reservoirs of antibiotic resistant Gram-positive bacteria and possible transfer routes (modified from Witte (1997, 1998); McDonald et al. (1997)).

It is by no means clear that such routes are in fact a reality or will actually contribute to the total antibiotics resistance in human bacteria. This scheme is on all levels heavily debated. Until now little evidence is presented to substantiate this scheme. We will however for clarity discuss the presented dissemination routes.

• Direct contact with an animal or animal faeces
  The people most intensively in contact with animals or animal faeces are farmers, slaughterhouse workers and veterinarians. Farmers and their families are frequently in contact with animal faeces, e.g. when they are cleaning a cowshed or a sty. Also direct contact with animals might be a source of resistant bacteria transfer. Another group are slaughterhouse workers. These people can be contacted with animal faeces and intestinal contents. Veterinarians can also come in contact with animal faeces or intestinal contents.

Not only the faeces of husbandry animals can contain resistant bacteria. Also faeces from pets fed with feed containing resistant bacteria can be a risk for humans.

Another risk forms the contact with antimicrobials themselves possibly generating resistant bacteria present in the human bowl. Farmers are exposed when they administer the drug, fill the feeding trays or when antimicrobial dust is present (MAFF, 1998).

• The consumption of meat or fish
  Meat can be contaminated with faeces or the intestinal contents during slaughter. If the meat is not sufficiently cooked some bacteria can survive and enter the human intestine. Also when the used utensils (knife, cutting board) are used for preparing uncooked food this food can be contaminated (Kruse and Sorum, 1994).

Fish cultured in an aquaculture, like salmon, is sometimes treated or fed with antibiotics. The consumption of fish containing resistant bacteria might lead to the presence of resistant bacteria in the human intestine.

• The consumption of vegetables or fruit
  One way in which food can be contaminated with bacteria originating from animals is that the environment gets contaminated. If vegetables are fertilised with faeces from animals containing bacteria resistant to antibiotics, the bacteria can survive on the vegetable. Vegetables are also treated with antibiotics, so residues can be present or resistant bacteria can evolve on the plant. Bacteria that show intrinsic resistance to antibiotics, like E. casseliflavus and E. gallinarum which show low resistance to vancomycin can be present on vegetables (Van den Braak et al., 1997). Raw vegetables and salads also can contain resistant enterobacteria (Levy, 1998; Feinman, 1998).

In a number of experiments people were given sterile food (no living bacteria present) to investigate the effect on resistant bacteria (Corpet, 1988). People that ate normal food carried up to one million times more lac+ enteric bacilli resistant to tetracycline in their faeces then the group that ate sterile food.

• Contact with water that contains faeces
  People drinking contaminated water or swimming in contaminated water can be exposed to resistant bacteria.
• Human to human spread
  Human to human spread involves not only the tentative scheme of animal to human transfer with subsequent human to human spread but more importantly the rise of resistant human bacteria as a result of therapeutic antibiotic use with a subsequent spread of human resistant bacteria. People in hospitals often are treated with antibiotics. Especially people who are treated for a long time with antibiotics have a high risk of the emergence of resistant bacteria in their intestines. They are susceptible for overgrowth of their own minor intestinal flora (e.g. enterococci). Resistant bacteria are propagated when antibiotics are given for extended periods. Resistant bacteria like VRE have been shown to spread from patient to patient in, but also between hospitals (Clark et al., 1993; Sader et al., 1994). The Dutch Health Council pointed out that (Gezondheidsraad, 1998):

'... The development of resistant bacteria in hospital patients is mainly caused by antibiotics used in the course of treatment. However, the cause of resistance development among the general public is less clear. ...'

The reason for a 'division' between the hospital environment and the community is not substantiated in the Dutch Health Council report. Patients treated and cured enter the community possibly carrying small numbers of resistant bacteria, which might easily spread within the community. Resistant bacteria can be transferred (transmit) indirectly through the personnel, when hand washing and disinfecting of equipment is not carried out properly (Gopal Roa, 1998). This might also be a dissemination route towards the community. The hygienic measures taken in different hospitals vary highly, as well as the prevalence of resistant strains (Verhoef, 1998, personal communication). A less frequent way of transmission is by the presence of airborne droplets. The moving of patients between wards in one hospital or between different hospitals increases the risk of spread of resistance.

Outside hospitals bacteria can be transferred from one person to another. Increased international travelling contributes to a global spread of resistant bacteria and resistance gene cassettes. The hospital might therefore in general be a source of resistance within human bacteria.

Two different events that in theory might lead to animal to human transfer of resistance can be distinguished:

• When a resistant bacteria of animal origin is able to colonise the human gut the resistance in effect has been transferred from the animal to the human. However, a bacterium has to survive the stomach. When it enters the intestines it has to be able to multiply in sufficient amounts before it truly colonises the human. The time the specific bacterial strain is able to stay in the intestines determines if it is a transient passenger or a permanent resident. Bacteria like enterococci and enterobacteria are capable of permanently colonising the large intestine (Drasar and Barrow, 1985). Resistance genes present on the bacterial chromosome, on plasmids or on transposons can be expressed during transient or permanent colonisation. Chances increase, however, when the stay is prolonged. In short:

• The second possibility is that resistance genes are being transferred from bacteria present in meat or animals to bacteria that are commonly found in human intestines. Transfer of genes can take place in the gut or prior to ingestion, after which the resistant bacteria may be able to colonise the human gut. Resistance traits present on plasmids or on transposons have a chance of being transferred to another bacterium. Genes present on the bacterial chromosome, but not on a transposon, have a much lower chance of being transferred. In short:

Data concerning transfer of Gram-positive bacteria resistant to AGPs from animals to humans is in essence non-existent. Van den Bogaard et al. (1997b) claimed that a turkey and a farmer had the same strain of vancomycin-resistant E. faecium. Until now this letter is the only one that describes indistinguishable strains in animals and humans. Moreover, it was not shown that this strain really colonised the human intestine and was not a transient passenger. Furthermore, reproducibility is lacking making this observation in effect open for debate and in want of thorough scientific scrutiny. Apart from these comments, extrapolation from this observation to other organisms or antimicrobial resistance traits is scientifically unsound and without foundation. The following hampers proving a transfer case:

Scheme 4.3.1. Research criteria concerning resistance transfer

4.4 Infectious Bacteria

Bacteria can cause several severe infections in humans. To be able to control the occurrence and spread of nosocomial infections (hospital acquired infections) it is useful to be aware of the risk factors for developing such an infection. In general, the condition of the patient, the use of antibiotics and the impact of hygienic measures in the hospital are important risk factors. The categories of people at higher risk and circumstances that increase risk are listed below (Bates et al., 1993; Gordts et al., 1995; Bogle and Bogle, 1997; Jones, 1996; Weinstein, 1998; Murray, 1990; 1998):

• immunocompromised persons: (elderly patients, ill neonates, patients with underlying disease)
• patients subjected to surgical operations, people with burns
• patients with breathing devices, catheters and drains
• type of ward: Intensive Care, renal units, hematology ward, surgical ward
• transfer of patients between wards or between hospitals
• use of antibiotics
• prolonged stay in the hospital
• the impact of hygienic measures taken (or not taken) in the hospital

The use of high amounts of antibiotics in hospitals contributes to the emergence of nosocomial infections. An example of this is the use of cephalosporins to treat infections caused by Gram-negative bacteria. The Gram-negative bacteria are killed, but resistant Gram-positive bacteria already present, like enterococci, can substantially increase in numbers and subsequently cause a superinfection. Bacteria containing multiple resistance genes are the most dangerous to humans, because of the higher change of resistance to the antibiotic(s) used for treatment. Some of the resistance traits present probably emerged and are maintained because of (high) prescription of antibiotics to humans in hospitals and the community.

The Gram-positive bacteria usually involved in (nosocomial) infections are described below. The genera Enterococcus, Staphylococcus, Streptococcus and Clostridium often cause nosocomial infections. For example, 34 % of hospital acquired infections are caused by enterococci, S. aureus and coagulase-negative staphylococci (Weinstein, 1998). The infections usually associated with the above-mentioned bacteria are:

Table 4.4.1. Some infectious diseases caused by bacteria.

• Enterococci
  Enterococci can play a role in urinary tract infections and infective endocarditis (Murray, 1990). Other nosocomial infections caused by enterococci are surgical wound infections and bacteraemia. Around 11 % of the bacteraemias are caused by VRE. When blood infection occurs often more then one pathogen is present (Jumaa et al., 1997).

Enterococci can also be found in infections located at the intra-abdomen and pelvis. Also ill neonates (premature babies) can be infected. Patients, who have undergone surgery, like a liver transplant or surgery of the central nervous system have a higher risk of becoming infected (Murray, 1990, 1998). Other factors that increase the chance that a patient develops a VRE infection are the treatment with (third generation) cephalosporins or aminoglycosides for a long time as well as a long stay in the hospital (Bogle and Bogle, 1997).

In the majority of people, however, enterococci are part of the microflora. Van den Bogaard (1997a) reported that 91 % of human faecal samples (province of Limburg, NL) contained enterococci (106/117). Schouten et al. (1997) found lower percentages, around 78 % in elderly people in the proximity of Nijmegen (NL). Whether these findings portray a 'normal' community situation remains questionable. More data are needed from comprehensive epidemiological studies.

Normally, the amount of enterococci found in the human microflora is quite low compared to other bacteria; 0.5 % of all bacteria are enterococci. When antibiotics kill part of the intestinal flora, resistant enterococci can cause overgrowth, which might lead to infection. In hospitals mainly E. faecalis is isolated; E. faecium accounts for 10-15 % of the isolates. E. faecium is more often carrying resistance genes (Murray, 1990).

Many enterococci are intrinsically resistant to cephalosporins, penicillins, aminoglycosides, clindamycin and Zn-bacitracin. Also resistance caused by acquired genes has been found. This concerns resistance against tetracycline, erythromycin, clindamycin, chloramphenicol, vancomycin and trimethoprim (Baquero, 1997; De Neeling et al., 1997a, c; Hoeffler and Zimmerman, 1997; Alpharma, 1998).

Vancomycin-resistant enterococci usually contain the mobile Tn1546 or Tn1547, containing the 7 genes responsible for vancomycin and teicoplanin resistance (Arthur et al., 1993, Quintiliani and Courvalin, 1996). The fear exists that VRE may not only transfer their resistance genes to other enterococci, but also to (methicillin resistant) Staphylococcus aureus strains. MRSA can still be treated with vancomycin. When multiple resistant MRSA is also resistant to this antibiotic, very few or no antibiotics are available for treatment. This tentative scenario is a powerful motive to ban antibiotics within livestock rearing.

• Staphylococcus aureus
  Staphylococcus aureus is involved in infections in hospitals (Baquero, 1997); 12 % of the bacteriaemias is caused by S. aureus, while this organism is responsible for surgical wound infections and skin infections in 28 and 21 % of the cases. Also osteomyelitis and endocarditis can be caused by S. aureus. The mortality rate can increase when resistant strains are involved, as shown by Romero-Vivas et al. (1995) who compared the fatality of bacteraemias caused by MRSA and MSSA (methicillin susceptible SA).

Between 1946 and 1950 penicillins, tetracyclines and macrolides were used to treat infections caused by staphylococci. As early as in 1950, penicillin was not successful in 80 % of hospital-acquired S. aureus infections because these strains produced _-lactamases. This high resistance percentage is still valid for the S. aureus strains isolated in (Dutch) hospitals.

In the 1960s methicillin was used instead of penicillin to treat S. aureus infections (Jones, 1996). This antibiotic is a semi-synthetic penicillinase-resistant penicillin (De Neeling et al., 1997c). Later, methicillin resistant strains were isolated, as well as cephalosporin resistant strains. Gentamicin was subsequently used to treat MRSA. However, during the 1970s MRSA strains resistant to gentamin arose.

MRSA can be found globally. Resistance to this nosocomial pathogen is especially high in Japan, where around 60 % of the S. aureus strains are methicillin-resistant. In Australia and Southern Europe this percentage is approximately 15, and in the USA hospitals 29 % of the strains are resistant (Voss and Doebbeling, 1995; Panlilio et al., 1992).

In the Netherlands this percentage is much lower, around 0.3 % between 1989 to 1995. (De Neeling et al., 1997). In nursing homes the prevalence of MRSA was also very low, 0.15 % in 1992 - 1993 (Frenay et al., 1993). In Denmark and the United Kingdom comparable prevalence levels of MRSA are found, 0.5 and 0.6 % (Jones, 1996).

Vancomycin is one of the few antibiotics that can be used to treat infections caused by MRSA. For some MRSA strains (especially in the USA) this is the only antibiotic that is still effective. In the lab the conjugational transfer of vancomycin resistance from E. faecium to S. aureus has been observed (Noble et al., 1992). The arising organism carrying this resistance trait might not be treatable in vivo.

In 1997 three patients geographically separated S aureus resistant to vancomycin was detected (Levy, 1998). Antibiotics were still effective to treat these infections. A strain isolated in Japan was intermediately resistant to vancomycin (MIC= 8 _g/ml). No genes seemed to be transferred from Enterococcus. Cell wall synthesis was increased making vancomycin less effective (Bogle and Bogle, 1997).

• Streptococcus pneumoniae
  Streptococcus species often are present in the mouth and intestines of animals and humans. Infections caused by S. pneumoniae can be dangerous to young children, elderly and people being immunosuppressed (De Neeling, 1996).

S. pneumonia is the major cause for pneumonia and is also involved in a large number of cases of meningitis, bacteraemia, sinusitis and otitis media (Baquero, 1997; De Neeling, 1996). The number of sick people (morbidity) and the mortality (amounts of deaths) caused by these diseases is relatively high.

Penicillin used to be the antibiotic of choice for treatment of above mentioned infections. Since the end of the 1960s S. pneumonia strains resistant to penicillin have been isolated. Especially in Spain, Hungary and Iceland the emergence of resistance is probably related to the high use of _-lactam antibiotics (De Neeling, 1996).

• Streptococcus bovis
  This is a D group Streptococcus, frequently present in the intestines of humans and animals. This bacterium can cause endocarditis, bacteraemia, neonatal infections and meningitis (Horaud and Bougu‚nec; 1987). Strains resistant to erythromycin, vancomycin, kanamycin, streptomycin and tetracycline, although not predominant, have been isolated (Poyart et al., 1997).
• Coagulase negative staphylococci

Infections by coagulase-negative staphylococci increasingly occur in intensive care patients, as well as in people being immunosuppressed or having prosthetic devices.

Staphylococcus epidermis and S. haemolyticus often are resistant to methicillin, oxacillin, aminoglycosides, macrolides, lincosamides, teicoplanin and exceptionally to vancomycin (Swartz, 1994; Baquero, 1997). Usually, coagulase-negative staphylococci can still be treated with vancomycin (De Neeling et al., 1997c).

4.5 Resistance Selection through Antibiotics

4.5.1 Glycopeptides as human medicine and AGP

• Glycopeptides used in humans select for vancomycin-resistant enterococci (VRE)
  Van der Auwera et al. (1996) tested the effect of the glycopeptides teicoplanin and vancomycin on human enterococcal flora. Twenty-two volunteers that were not positive for VRE were administered vancomycin or teicoplanin. After three weeks of antibiotics exposure, 64 % of the faecal samples contained VRE. The use of glycopeptides was probably responsible for the emergence of VRE in the human intestine. Another possibility is that VRE were already present in very small numbers (that is below the detection level) before treatment and that they have been selected for as a result of the use of glycopeptides.

Other, less likely possibilities are that the people were exposed to contaminated food or that their stay in a hospital (during the experiment) favoured colonisation with VRE.

• Glycopeptides (avoparcin) used as feed additive select for VRE
  In Denmark avoparcin (an AGP) was used in feed consumed by poultry and pigs, but has not been used for calves (Aarestrup et al., 1996). Faecal samples from poultry flocks, pig herds and calves were checked for the presence of VRE before the ban on avoparcin. It was shown that the E. faecium and E. faecalis present in calves were susceptible to avoparcin. On the other hand, 72 % of the poultry flocks and 20 % of the pig herds contained vancomycin-resistant E. faecium.

Bager et al. (1997) also determined the prevalence of VRE in faecal samples taken from 12 pig farms where avoparcin was used and from 10 pig farms where avoparcin was not used recently. At the farms using avoparcin 8 of the 12 herds contained VREfm, while these bacteria were isolated only from 2 out of 10 herds on the farms where no avoparcin was used. Data concerning the use of avoparcin in pig herds were obtained from feed-mills that were the exclusive suppliers of feed.

Klare et al. (1995) showed the presence of vancomycin resistant enterococci in manure of a pig and a poultry farm in the German county Saxony-Anhalt where avoparcin was used. In the manure of an egg-harvesting hen farm in the same region where no avoparcin was mixed with the feed, no glycopeptide resistant enterococci were found. In isolates obtained from poultry the number of vancomycin-resistant enterococci out of total enterococci was higher than in isolates from pigs.

Van den Bogaard et al. (1996; 1997a) determined the prevalence of vancomycin-resistant enterococci in Dutch turkey flocks having avoparcin mixed in their food and in flocks not receiving this antibiotic. Of the 12 turkey flocks that were fed without avoparcin, 8 % of faecal samples contained vancomycin-resistant enterococci, while 60 % of flocks fed with avoparcin contained these resistant bacteria.

The above mentioned examples both in human medicine and in animal farming show that the use of glycopeptides such as vancomycin and avoparcin selects for resistant strains of enterococci. It remains to be seen if this resistance is of a permanent nature.

4.5.2 Virginiamycin used as AGP

In many (European) countries virginiamycin (a streptogramin) is added to the feed of broilers and pigs (Witte, 1998; Aarestrup et al., 1998). In Denmark the prevalence of resistant bacteria amongst enterococcal isolates of pigs, broilers (Aarestrup et al., 1998; DVL, 1998) and cattle (Aarestrup et al., 1998) was studied. It was found that 40 - 68 % of the E. faecium isolated from Danish pigs and broilers (probably) fed with virginiamycin show resistance against this antibiotic. In E. faecium from cattle (which not receive virginiamycin) just 8 % was resistant to virginiamycin. Some years after the Finnish ban of virginiamycin in 1990, resistance amongst broiler and pig isolates was much lower than in Denmark (20 % respectively 2 % contained virginiamycin E. faecium; DVL, 1998). In staphylococci isolated from diseased animals (pigs, cattle from Denmark), very low resistance percentages were found (0-1 %).

SCAN noted that the causal relationship between the use of virginiamycin and the development of resistance to this antibiotic is not as clear cut as presented by the Danish studies mentioned above (SCAN, 1998b). SCAN's criticism is mainly of a methodological nature leaving the observations done in Denmark open for debate.

4.5.3 AGPs also acting against Gram-negative bacteria

• Tylosin
  In Denmark, pigs and broilers consume feed containing tylosin, but cattle do not. Aarestrup et al. (1997) analysed faecal samples of swine, cattle and broilers on the presence of antibiotic resistant Campylobacter species (a Gram-negative bacterial species). The activity of 16 antibiotics including tylosin, spiramycin (growth promoters) and erythromycin (used in human and animal medicine) was determined. The Campylobacter species most isolated from pigs was C. coli. Out of 99 C. coli strains derived from pigs, 73 and 74 % was resistant to tylosin and erythromycin respectively. In broilers the prevalence of resistance was lower: 18 % out of 17 strains was resistant to tylosin/erythromycin. In cattle and broilers C. jejuni was the most isolated strain. Out of 29 C. jenuni strains isolated from cattle, 3 % was resistant to tylosin/erythromycin. Of broiler derived C. jejuni strains, 6 % was resistant to tylosin/erythromycin.
• Nourseothricin
  Hummel et al. (1986) studied the effect of the usage of nourseothricin as a swine feed additive in former East-Germany. No structural analogues are used as medicine for animals and humans. Resistance to this antibiotic was found in E. coli strains isolated from pigs, as well as with farm related people, healthy people and people with urinary tract infections. The emergence of resistance in animals was clearly due to the use of nourseothricin. Also it was shown that resistance (or resistant bacteria) to this antibiotic was spread from animals to humans. This example is frequently referred to as an example of resistance transfer from animals to humans disregarding the fact that E. coli is a zoonotic Gram-negative organism.6 This example therefore bears no relation to the AGP issue described in this report.

4.6 Research Efforts and Data Compatibility

4.6.1 General

To be able to answer the question whether resistance transfer occurs, data should be compared and analysed thoroughly. In the box below some questions are listed:

To be able to show a relation between bacteria found in animals and in humans and to be able to compare resistance data between different research groups these questions need to be kept in mind. Useful evidence for gene transfer or bacterial colonisation is only provided when these criteria are met.

4.6.2 Research methods, data compatibility and reproducibility

• Source of samples
  In general, for comparison of data, it is essential that the source of the samples be known. When trying to find out whether a relation exists between the use of antibiotics and the prevalence of antibiotic resistant bacteria, the origin of samples is imperative. For samples taken from animals it is essential to know if this flock or group of animals has received feed containing antibiotics. Moreover, it should be known which type of antibiotics was administered to the animals. In studies where meat (pork, beef) is examined for the presence of resistant bacteria, the sampled meat can not always be traced (back) to the farm of origin. The same holds true for chickens bought in shops. Usually their growing conditions are not known.

The source of resistant bacteria found in meat or poultry is not always evident. These bacteria could have been selected by the use of antibiotics in the feed of the animal or the meat could be contaminated during transport, slaughter or in the butcher shop.

Data found in the scientific literature can roughly be divided into two groups:

• Data directly related to the use of antibiotics in feed
• Other data that give an impression of the prevalence of resistant bacteria in the environment
  • Isolation and phenotypic/genotypic characterisation of strains

  In the articles that provide characterisation data, a wide variation in methods to isolate and characterise strains and to determine resistance patterns can be found.

  The use of multiple techniques by different research groups can be a problem in comparing data. For example, the use of different isolation techniques (media, enrichment procedures) can lead to different numbers of resistant bacteria. Another problem is the criteria used to state if strains or resistance traits in animals and humans are related. Which techniques are suitable to detect relationships and when is a resistance trait in animal  nd human bacteria considered identical?

  Below the techniques for isolation and characterisation of resistant bacterial strains are listed. This is basically a step-wise procedure resulting in information about the specific micro-organism studied and its resistance profile.

  • Isolation of resistant strains and phenotypic characterisation

    • Enrichment cultures (different concentrations of the antibiotic used in selection)
    • Determination of species and sub-species: biochemical methods an ready-to-use kits
    • Determination of Minimal Inhibitory Concentrations (MIC; different methods and media)
    • Determination of MICs of single or multiple antibiotics

  • Genotypic characterisation
    • Pulsed Field Gel Electrophoresis (PFGE) of chromosomal DNA (digested by SmaI)
    • Region amplification within a (resistance) gene by PCR; regions within In: intergenic amplification by PCR
    • Ribotyping
    • Long PCRs of transposons
    • Conjugational studies
      

    Back to top

  4.6.3 Isolation of resistant strains and phenotypic characterisation

  • Vancomycin-resistant enterococci
    A problem is that resistant enterococci/streptococci are only present in small numbers in the large intestine (compared with Bacteroides and E. coli). In general, when enterococci are present in human faeces (90 % of the people), their total number is around 1 _ 107 per gram. When VRE are present they form around 0.5 % of the total number of enterococci (Drasar and Barrow, 1985; Van den Bogaard et al., 1997a).

  To be able to detect vancomycin-resistant enterococci, enrichment cultures are often needed. This means samples are streaked on (agar) plates supporting growth of the desired organism (Schlegel, 1992). For isolation of enterococci a few media are suitable:

  • Enterococcosel selective agar plates (Klein et al., 1998; Van den Braak et al., 1997)
  • Enterococcosel broth combined with kanamycin esculin azide agar (Van den Braak et al., 1998)
  • Bile esculin azide agar (Sahm et al., 1997)
  • Iso-sensitest broth followed by blood agar (Kirk et al., 1997a)
  • Streptococcus agar plates (Van den Bogaard et al., 1997a)

  It is evident that not all research groups use the same media for enrichment, which makes resistance percentages found difficult or even impossible to compare. An extra enrichment procedure for vancomycin resistant enterococcal strains can be achieved by supplementing plates with vancomycin. The concentration of vancomycin used may vary between different research groups. For example, bile esculin azide agar containing 6 _g/ml vancomycin is used (Sahm et al., 1997) as well as Enterococcosel agar containing 32 _g/ml vancomycin (Klein et al., 1998).

  Klein et al. (1998) showed the effect different isolation procedures can have on the resistance percentages found. They examined meat samples for the presence of VRE. Two different methods for culturing VRE were used:

  • the use of Enterococcosel selective agar plates supplemented with vancomycin (32 _g/ml)
  • an overnight pre-enrichment method with buffered peptone water followed by inoculation of Enterococcosel selective agar plates supplemented with vancomycin.

  A much higher resistant percentage was found with the second, double enrichment method, 8 % VRE (out of 555) as compared with the direct method where 0.5 % VRE (out of 555) was detected. Resistance percentages mentioned in the literature need to be looked with caution as methods of determination vary widely.

  Vancomycin resistance is predominantly found in enterococci. Enterococci can be subdivided in different species groups and species. Of these different species, only E. faecium and E. faecalis have been shown to carry the clinical relevant vanA or vanB gene complex (VRE). These species are the major enterococcal species in poultry, pigs, cattle, dogs and humans. Enterococci like E. gallinarum and E. casseliflavus that contain vanC genes are also found. This C type of resistance is not transferable and the level of resistance obtained is much lower than resistance of the VanA and VanB types.

  To be able to determine the amount of VRE in a sample, it is important that enterococci can be identified to the species level (Devriese et al., 1991; 1993a). Some common characteristics for enterococci are growth at 45 øC, 10 øC, and at pH 9.6. These parameters can be used to distinguish enterococci from other bacteria. If the bacteria involved have been shown to be enterococci, then distinct features of the species can be checked. E. faecalis for example, can be distinguished by its tolerance to 0.04 % tellurit. Likewise, E. gallinarum can be distinguished from E. faecium, E. casseliflavus and E. flavescens because of its motility and lack of pigmentation. Also 'ready-to-use' kits are applied for the identification of enterococci. The API20 strep and the RAPID ID 32 strep are two of these kits (Kirk et al., 1997a; Hill et al., 1997). The API strep 20 combined with detection of species specific genes is very reliable (Dutka-Malen et al., 1995), while the RAPID ID 32 strep misidentifies many E. faecium strains for E. galinarum.

  • Staphylococci
    Often staphylococci normally colonising animals or humans have different properties. Staphylococcus aureus can be divided in six biotypes (A-F; Hajek and Marsalek, 1971). In humans biotype A is the major biotype, while biotype B is typical for poultry and pigs. Biotype C is typical for cattle, sheep and goats. S. intermedius is the dominant species in dogs and cats (Cox et al., 1985) and is also present in horses.

  In pigs the S. aureus biotypes A and C can also be present (Devriese, 1984), but no strains common for pigs were detected in man. A biotype closely resembling biotype B can be present in humans frequently contacted with meat or animals (Isigidi et al., 1990).

  The different biotypes can be identified in culturing tests (biochemical differences). In addition to biotyping, phagetyping can be done. Different biotypes can be lysed by different phages, but also by the same phage (Shimizu, 1977; Isigidi et al., 1990).

  • Streptococci
    Like the staphylococci, streptococci show host-adaptation. This adaptation can be species related or present at the biotype level. A method often used to classify streptococci is based on the presence of cell wall carbohydrate antigens (serogrouping according to Lancefield). This method can be combined with biotyping based on biochemical/culturing tests (Devriese, 1991). In the same serogroup however, strains that colonise different hosts can be determined (e.g. Streptococcus agalactiae has a human and bovine biotype). S. pneumoniae and S. pyogenes are strains typical in primates. S. equisimilis occurs in humans and pigs, but the strains can be distinguished by plasminogen activator tests. Strains present in human or animal, only show activity for the plasminogen in that specific host (McCoy et al., 1991).
  
  Back to top

  4.6.4 Determining resistance data: phenotypic characterisation

  • General

  Resistance of bacterial strains can be detected in vitro by determination of the MIC (minimal inhibitory concentration) of a specific antibiotic. Estimation of the MIC is an important tool for distinguishing between resistant and susceptible strains. It should be noted that MIC-values do not always closely correlate with the effect in vivo. A reliable MIC estimation in this case means that the effect of the antibiotic in vivo (e.g. the effect of the antibiotic as a therapeutic agent) closely resembles the in vitro data.

  Bacteria are grown in the presence of (different concentrations or a gradient) antibiotic. The lowest concentration where no growth occurs is defined as the MIC. In the table below the different methods to determine the MIC of an antibiotic are shown.

   

  Method	Number of bacteria
  1a. Broth macrodilution (two-fold dilution; 1 - 2 ml broth) 1b. Broth microdilution (two-fold dilution; microdilu- tion trays, 100_l) 2. Agar dilution test (two-fold dilution; agar plates) 3. Etest(r) (AB BIODISK; gradient in agar; agar plates and strips) 4. Disk diffusion test (gradient in agar; agar plates and disks)	5 * 10 log 5  cfu/ml 5 * 10 log 5 cfu/ml 1 * 104 bact./drop 1 - 2 * 108 cfu/ml 1 - 2  * 108 cfu/ml

  Table 4.6.4.1. Methods to determine antimicrobial susceptibility. After inoculating with bacteria, the broth or agar plates are inoculated for 16 - 20 hours (Devriese et al., 1993a; Woods and Washington, 1995; NCCLS, 1997; Jorgensen and Ferraro, 1998).

  A test often used to determine antimicrobial susceptibility (and a MIC-value) is the agar dilution test (method 2). Agar plates containing increasing concentrations of the antibiotic are inoculated with the same amount of bacteria and the growth is monitored. A comparable test is method 1, where the bacteria are grown in broth containing different concentrations of antibiotics. This can be done in small tubes or in microtiter plates containing even smaller amounts of liquid. Another method where a MIC-value can immediately be derived from the results is the Etest(r) (method 3). Here strips with antibiotic are placed in the agar, after which the MIC can be read from the strip (border growth/no growth). In method 4 also a gradient is formed in the agar, but by placing an antibiotic disk on the agar plate. In this way an inhibition zone is observed around the disk. The size of this zone (in mm) indicates the susceptibility of the bacteria for this antibiotic. Categories of zone sizes (susceptible, resistant) have been determined for some antibiotic-bacterium combinations.

  Tests to determine the MICs are in vitro tests, so the effect of an antibiotic on the bacteria in vivo can not be guaranteed. The concentration of the antibiotic at the site of infection is important and the time period the antibiotic is present (Bryan, 1982).

  For each antibiotic used in human health MICs have been documented prescribing when a certain strain is considered resistant or susceptible. Some antibiotics are characterised as intermediate resistant at a specific concentration range; in this case the effect of the antibiotic is not that/always clear. For antibiotics only used as growth promoter criteria and methods for testing and evaluating resistant and susceptible strains are less well documented. In this way it is more difficult to provide reliable/reproducible prevalence data for growth promoters then it is for antibiotics used in human medicine. When resistance criteria are poorly documented, all MIC data of the strains examined should be presented. When showing the MIC-range it is possible to compare data obtained in other laboratories.

  • Influence of media and incubation conditions

  Comparability of MIC data is hampered by the use of different media and incubation conditions by different research groups. Not all enterococcal strains show the same susceptibility to an antibiotic when the test is repeated under different conditions. It is essential that the circumstances, which provide the most reliable MIC is determined and reproducibility can be guaranteed.

  In most countries recommendations have been formulated on how to perform antimicrobial susceptibility testing (National Committee for Clinical Laboratory Standards (NCCLS, 1997), Dutch Working Group for Antimicrobial Susceptibility Testing, British Society of Antimicrobial Chemotherapy). These recommendations are updated regularly (every few years). The methods and the media to be preferably used are described. Also susceptibility and resistance criteria are provided. The use of control strains, with known MICs under the applied test conditions, is also recommended as a quality control for the performance of a test (Gordts et al., 1995).

  The recommendations regarding susceptibility testing are no strict rules. Variations in testing methods, susceptibility criteria and the media used, occur between different research groups and different countries. Also old testing criteria are sometimes used. Therefore, it is important that world-wide recommendations are developed and fulfilled.

  • Influence of media and incubation conditions: examples

  An important quality item for MIC determinations is the number of very major errors made. A very major error is defined as a resistant strain misinterpreted as susceptible. (A major error is defined as a susceptible strain being misinterpreted as resistant.)

  According to Kohner et al. (1997) determination of the MIC on a Mueller-Hinton medium in a broth microdilution or agar dilution test is a good method to detect resistance due the vanA or vanB gene system in enterococci. On the other hand, when it concerns vancomycin resistance due to the vanC gene the Mueller-Hinton medium is far less applicable; a high percentage (65 - 90 %) of very major errors is found. The Mueller-Hinton medium is recommended by the NCCLS and therefore regarded as a standardised procedure. Kohner et al. (1997) however, recommend the use of the brain heart infusion medium (BHI) in testing vancomycin resistance (VanA, B and C phenotype) with the broth dilution and agar dilution method. A disadvantage of this medium is its complexity and the lack of standardisation in testing methods (Butaye et al., 1998a).

  Especially when regarding susceptibility testing of growth promoters it is not known which testing conditions give the best results. Butaye et al. (1998a) examined the effect of different conditions in Mueller-Hinton II agar dilution tests on the MICs of growth promoters acting against enterococci. Enterococci of E. faecalis, E. faecium, E. avium, E. gallinarum and E. cecorum species groups were tested. The effect of adding sheep blood or CO2 as well as differences between aerobic and anaerobic incubation was studied.

  Important differences could be observed between the examined strain-growth promoter-condition combinations. Nevertheless it was possible to distract general conditions which give satisfactory results in most of the cases.

  When testing E. faecium and E. faecalis only, Mueller-Hinton II medium without blood and aerobically incubated is suitable. When testing more enterococcal species the recommended conditions are Mueller-Hinton II medium containing blood, while incubation should take place under a CO2-enriched atmosphere.

  • Susceptibility testing as part of phenotypic characterisation of resistant strains

  When comparing antimicrobial susceptibility patterns the strains containing the vanA or vanB gene can be separated from the strains containing vanC, because the vanA and vanB genes give rise to higher MICs. VanA containing strains are highly resistant to vancomycin, vanB strains show intermediate resistance. Enterococci containing vanC and vanB are not resistant to teicoplanin. When a strain is of the VanC type, it is probably not E. faecium or E. faecalis, but belongs to the E. gallinarum species group. However, these phenotypic tests only lead to predictions