Antimicrobial resistance in Campylobacter strains isolated from French broilers before and after antimicrobial growth promoter bans

Abstract

Objectives: The antimicrobial susceptibility of Campylobacter strains isolated from standard and free-range broilers in 1992–1996 and 2001–2002 was studied.

Methods: Strains were isolated from caeca or skin samples collected from standard or free-range broilers arriving in slaughterhouses. The MICs of ampicillin, nalidixic acid, enrofloxacin, tetracycline, erythromycin and gentamicin were determined by agar dilution and compared according to species (Campylobacter jejuni or Campylobacter coli), production system and sampling period.

Results: Results showed that all chickens harboured Campylobacter. An increase over time of the C. coli/C. jejuni ratio for standard chickens occurred. A wide range of MICs was observed among isolates from the same broiler or from the same farm. Strains collected on entry to the slaughterhouse and after storage showed no significant difference in their antibiotic resistance. C. coli was more resistant than C. jejuni to tetracycline and erythromycin during the first period and to all tested molecules (except gentamicin) during the second period. Strains isolated from standard chickens were also more often resistant than those isolated from free-range broilers. The percentage of C. jejuni strains resistant to ampicillin decreased from 1992–1996 to 2001–2002, whereas no change could be observed for the other antimicrobial agents. However, for C. coli the resistance to ampicillin, nalidixic acid, enrofloxacin, tetracycline and erythromycin significantly increased.

Conclusion: There was an increase in the incidence of antibiotic resistance of C. coli between 1992–1996 and 2001–2002.

Introduction

Campylobacter is one of the leading sources of human bacterial diarrhoea worldwide, with Campylobacter jejuni and Campylobacter coli representing the most frequently involved species.^1,2 Therapy is not usually required but may be necessary in severe cases or in immunocompromised patients. As one of the main sources of infection is considered to be foods of poultry origin,³ it is important to monitor the antimicrobial susceptibility of strains isolated from chickens or chicken products.⁴ Moreover, in response to the fear that continued use of growth promoters could contribute to the selection and spread of antimicrobial-resistant bacteria in food animals and thus weaken the efficacy of antibiotic treatment in human medicine, the European Union suspended successively in 1997 and 1998, the use of avoparcin and four other antimicrobial growth promoters (virginiamycin, tylosin, spiramycin and bacitracin) in animal feed.⁵ The present study aimed at comparing the antimicrobial resistance of Campylobacter strains isolated from French broilers or poultry products in 1992–1996 and 2001–2002. That is, respectively, before and after the ban of growth promoters in feed antibiotics. Additionally, for the two periods studied, strains were collected from standard and free-range broilers, two production types that differ with respect to antimicrobial use. In France, in free-range production contrary to standard production, antimicrobial growth promoters have never been authorized and antibiotic use for therapy was, and is, always limited. Other factors differentiating free-range and standard productions are flock density, access to outdoor runs and age at slaughter.⁶

Materials and methods

Campylobacter strain collections

The first collection (1992–1996) was obtained from four standard and four free-range chicken flocks slaughtered at the same place (East France). For each flock, 10 chickens were collected. Campylobacter isolates were obtained from caeca and skin samples⁷ that had been stored refrigerated for 7 days in order to mimic storage conditions of retail products. A maximum of three isolates of Campylobacter per chicken was stored in this first collection. In addition, 52 C. jejuni and seven C. coli strains collected from skin on retail standard chicken products from different French regions were added to this first set.

The second set of strains was collected in 2001–2002. Ten standard and four free-range chicken flocks were sampled in three locations in France (East, in the same slaughterhouse as in 1992–1996, Centre or West). Ten chickens per flock were randomly collected in slaughterhouses; caeca and skin samples that had been stored refrigerated for 7 days were analysed. For this second collection, in order to study the diversity of Campylobacter isolates, it was decided to keep all isolates obtained from each sample.

All isolates were stored at −80°C before confirmation of identification by multiplex PCR.⁸ Only isolates identified as C. jejuni and C. coli were subjected to antimicrobial susceptibility testing. These isolates were further characterized by pulsed-field gel electrophoresis (PFGE) after digestion with SmaI or KpnI endonucleases⁹ (data not shown).¹⁰ For the first collection, one isolate per chicken was selected, possibly more in the case of different PFGE patterns, whereas for the second set, all singular strains isolated from each broiler (or three isolates in the case of identical PFGE patterns) were included in the study.

Antimicrobial susceptibility testing

Antimicrobial susceptibility testing was conducted by agar dilution, according to the NCCLS document M7-A4.¹¹ Six antimicrobials were tested: ampicillin (0.25–64 mg/L) (Sigma, Saint-Quentin-Fallavier, France), nalidixic acid (1 to 256 mg/L) (Sigma), enrofloxacin (0.03 to 32 mg/L) (Bayer, Puteaux, France), tetracycline (0.125 to 128 mg/L) (Sigma), erythromycin (0.25 to 4 mg/L) (Sigma) and gentamicin (0.03 to 16 mg/L) (Sigma). The following reference strains were used: Escherichia coli (ATCC 25922), Staphylococcus aureus (ATCC 29213), Pseudomonas aeruginosa (ATCC 27853), Enteroccocus faecalis (ATCC 29212) and Campylobacter jejuni (ATCC 33560). Breakpoints for Campylobacter resistance were >16 mg/L for ampicillin and nalidixic acid, >8 mg/L for tetracycline and gentamicin, >4 mg/L for erythromycin and >2 mg/L for enrofloxacin (breakpoint given for ciprofloxacin) according to the statement 2000–2001 of the Antibiogram Committee of the French Society for Microbiology.¹²

Statistical analysis

One hundred and eighty-three different strains isolated during 1992–1996 and 596 isolates obtained in 2001–2002 were included in the statistical analysis. χ² and Fisher's exact two-tailed tests were used to compare, for each antibiotic, the distributions of resistant and non-resistant strains according to sample type, species, flock production type and period. A significance level of 5% was used.

Results

Prevalence and identification of Campylobacter species from broilers

For both collections, all chickens from all studied flocks yielded Campylobacter from caeca and skin samples. For the 2001–2002 collection, in three flocks, more than 100 Campylobacter isolates were obtained and 60–100 isolates were obtained from six other flocks. A total of 962 Campylobacter isolates were obtained from the 14 chicken flocks. However, after freezing, a number of isolates could not be recovered. Finally, 661 isolates were analysed by multiplex PCR and after analysis by PFGE, no more than three isolates per chicken, sharing identical patterns, were included in the study. The distribution of species according to period and production type is given in Table 1.

Diversity within a flock

The large number of isolates collected during the second period, and the molecular typing, enabled the differentiation of strains and showed the wide dispersion of MIC classes within a farm. Results showed that different strains from a single chicken could exhibit very different resistance profiles. Thus, for example, the no. 7 chicken of the no. 9 standard flock had four different strains, with MICs of ampicillin, nalidixic acid, enrofloxacin, tetracycline, erythromycin and gentamicin of 4–16, 2–256, 0.0625–8, 0.125–0.25, 1–4 and 0.25 mg/L, respectively. Using as an example the 10 chickens sampled from the no. 9 flock, the MIC intervals were 1–32, 2–>256, <0.03125–16, <0.125–>64, 0.25–>64 and 0.0625–4 mg/L, respectively. The mean dispersions of different MIC classes per farm were 3.4 doubling concentrations for gentamicin, 6–6.4 for ampicillin, erythromycin, nalidixic acid and enrofloxacin and 7.6 for tetracycline.

Antimicrobial resistance

Antimicrobial susceptibility according to sample type. Strains collected at entry into the slaughterhouse (i.e. caeca isolates) and after storage (i.e. skin isolates) showed no significant difference in their antibiotic resistance. Thus, the results of both sample types were considered for further analysis.

The MICs and percentages of resistant strains of 183 isolates of 1992–1996 and 595 isolates of 2001–2002 for C. jejuni and C. coli strains are shown in Tables 2 and 3.

Multiresistance

For the first period (1992–1996), 2/132 (1.5%) and 6/132 (4.5%) C. jejuni isolates were resistant to four and three antibiotics, respectively, and 2/51 (3.9%) C. coli isolates were resistant to three agents. For the second period (2001–2002), 2/326 (0.6%) and 6/326 (1.8%) C.jejuni isolates were resistant to four and three molecules, respectively, whereas 19/269 (7%), 82/269 (30%) and 48/269 (18%) C. coli isolates exhibited resistance to five, four and three antibiotics, respectively. All enrofloxacin-resistant isolates were also resistant to nalidixic acid (the majority being C. coli). All strains were susceptible to gentamicin.

Resistance according to production type

For each species and each period, the distributions of resistant and non-resistant isolates were compared between standard and free-range broilers. The C. jejuni isolates of the second studied period were more often resistant to ampicillin (P < 0.001) and tetracycline (P < 0.001) in standard broilers compared with free-range broilers. The C. coli isolates from standard broilers of 2001–2002 were more resistant to all molecules (except gentamicin) than isolates from free-range chickens (P < 0.001). This observation was also true for strains of 1992–1996 for tetracycline (P = 0.045) and erythromycin (P = 0.017).

Resistance according to species

For the first period, C. jejuni isolates were more often resistant to ampicillin, but less resistant to tetracycline and erythromycin than C. coli (P < 0.001). For the second period, C. coli isolates were more resistant to all molecules (except gentamicin) than C. jejuni (P < 0.001).

Resistance according to period

The percentage of C. jejuni strains resistant to ampicillin decreased from 1992–1996 to 2001–2002 (P < 0.001), whereas no change could be observed for the other antimicrobial agents. However, for C. coli, the resistance to ampicillin, nalidixic acid, enrofloxacin, tetracycline and erythromycin significantly increased (P < 0.001).

Discussion

Prevalence and distribution of Campylobacter species

All broilers from all studied flocks yielded positive Campylobacter cultures. Such a high prevalence is in accordance with previous observations from other investigators.^13,14 The lower prevalence previously reported in a French study⁶ reflected different sampling and culture conditions (one instead of 10 sampled chickens, direct isolation instead of enrichment phase).

The C. jejuni species represented 72% for the 1992–1996 collection and 55% for the 2001–2002 collection, although the standard/free-range proportion was superior in 2001–2002 (68% of strains from standard chickens for the first collection, 77% for the 2001–2002 collection, P = 0.02). However, the comparison between caeca and skin samples collected from broilers arriving in slaughterhouses (retail products excluded) in 1992–1996 and 2001–2002 showed that the C. jejuni percentage significantly decreased in standard production from 67.6%–52.5% (P < 0.05). Indeed, a high proportion of C. coli in the caeca of standard broilers in 2001–2002 (63%) was observed, compared with the skin of the same animals (34%), and may be related to differences in the digestive flora of standard chickens.

The relatively high percentage of C. coli in free-range chickens (39% in 1992–1996) was in accordance with previous observations.⁶ However, the observed decrease over time of the C. jejuni percentage in French standard broilers is noteworthy and confirms results of national surveillance programmes for zoonotic bacteria reported by Kempf et al.¹⁵ According to these authors, in 2002, the percentage of C. coli in standard chicken production was 60%. The increase in the percentage of C. coli may be due to modifications of poultry breeding conditions between the two periods. Apart from the ban on antimicrobial growth promoters, another major factor occurred during this period, i.e. the ban on animal protein-based feed, which might have influenced the digestive microflora equilibrium. Indeed, Udayamputhoor etal.¹⁶ showed that the caeca of birds receiving plant protein-based feed were colonized less significantly with C. jejuni than the caeca of birds receiving the other types of feed. It will be important, in the future, to analyse the effect of the diet formulation on C. coli colonization and to search for other, as yet, unsuspected factors to explain the species ratio changes. Furthermore, it is of utmost importance to monitor the variation of C. coli in human Campylobacter diseases, as suggested by Tam et al.¹⁷

Antimicrobial resistance of 2001–2002 isolates

The breakpoints used for this study were those described by the Antibiogram Committee of the French Society for Microbiology. The main differences between these breakpoints and the limits given by the NCCLS concerned ampicillin (≥4 instead of >16 mg/L) and enrofloxacin (≥2 instead of >2 mg/L).

No significant difference in the antibiotic resistance of strains collected at entry into the slaughterhouse (i.e. caeca isolates) and after storage (i.e. skin isolates) could be detected, a feature that may indicate that our storage conditions (7 days at 4°C) and the sample type did not select strains of particular resistance profiles. However, the storage conditions used here may not reflect all environmental stresses during processing from slaughterhouse to the consumer's table.

For C. jejuni, the highest percentages of resistant strains were to tetracycline (26.7%), then ampicillin (14.2%) and nalidixic acid (4.0%). For C. coli, the highest percentages of resistant strains were 83% for tetracycline, 61.7% for erythromycin and 45.1% for nalidixic acid. No strain was resistant to gentamicin. The data obtained in this study, from chickens slaughtered in only three different locations in France, were slightly different from those published by Avrain et al.,⁶ who examined Campylobacter from a representative collection of French broilers. However, as the aim of the present study was to analyse the evolution of Campylobacter-strain susceptibility, it was essential to generate, in 2001–2002, a set of strains comparable to the group of isolates collected in 1992–1996. For this reason, a large number of the samples collected in 2001–2002 originated from the slaughterhouse (and whenever possible from the farms) already sampled in 1992–1996.

The influence of the species (C. coli or C. jejuni) on the resistance to antimicrobials has been reported by several authors.^18,19 The other factor influencing the resistance profiles was the production type: isolates collected from standard flocks were often more resistant than isolates from free-range chickens. A similar observation has been reported by Frediani-Wolf & Stephan²⁰ and Avrain et al.⁶ and probably reflects the effect of the use of antimicrobials on the selection of resistant strains.

Development of antimicrobial resistance

The increased incidence of antibiotic resistance between 1992–1996 and 2001–2002 was reported especially for C. coli strains isolated from standard chickens, whereas for C. jejuni the only change observed was a decrease in ampicillin resistance. The C. coli isolates over the two sampling periods showed the most increase in resistance to quinolones (nalidixic acid and enrofloxacin), tetracycline and the macrolide, erythromycin. Fluoroquinolones and macrolides are first-choice antibiotics for serious Campylobacter infections. As poultry is a main source of human campylobacteriosis, this increase in resistance is worrying for medical reasons.

The ban in 1999 of several macrolides as growth promoters aimed at decreasing the antimicrobial resistance of bacteria in food animals. However, our results indicate that in France, an increase in resistance of C. coli strains isolated from standard chickens was observed. It is possible that the sampling period (2001–2002) was too close to 1999 to detect an effect of the ban. Moreover, previous observations in different countries²¹ showed an increase in the usage of therapeutic antimicrobials (such as macrolides or tetracyclines) after the withdrawal of antimicrobial growth promoters. This feature could have contributed to the increase in the resistance to macrolides or to tetracyclines that have never been used as growth promoters in France. Additionally, it should be underlined that the first sampling period began around the same time as the licensing of fluoroquinolones for veterinary use and this may explain the increase in resistance to fluoroquinolones, as was reported by Luber et al.²²

We thank F. Megraud, P. Pardon and P. Beaubois for their scientific support. This study was funded by ACTIA, and was presented in part at the Twelfth International Workshop on Campylobacter, Helicobacter and Related Organisms in Aarhus, Denmark, 6–10 September 2003 and at the Sixth Congress of the French Society for Microbiology in Bordeaux, France, 10–12 May 2004.

References

Author notes

1Aérial, rue Laurent Fries, Parc d’Innovation, BP 40443, 67412 Illkirch cédex; 2Unité de Mycoplasmologie–Bactériologie, Agence Française de Sécurité Sanitaire des Aliments, BP 53, F-22440 Ploufragan, France