INTRODUCTION
Campylobacter spp. are one of the leading causes of bacterial gastroenteritis worldwide with C. jejuni and C. coli being the most common species infecting both adults and children. While most outbreaks of campylobacteriosis have been reported from developed countries, sporadic and infrequent outbreaks also occur in developing countries [Reference Friedman, Nachamkin and Blaser1, Reference Oberhelman, Taylor, Nachamkin and Blazer2]. Several studies have described Campylobacter spp. infections in Southeast Asian countries, Laos [Reference Phetsouvanh, Midorikawa and Nakamura3], Vietnam [Reference Isenbarger4], Singapore [Reference Lim and Tay5], Indonesia [Reference Tjaniadi6], Bangladesh [Reference Albert7] and Thailand [Reference Varavithya8], but limited information is available about their distribution in different population groups, susceptibility to antimicrobials and the serotypes involved.
Fluoroquinolone drugs, such as ciprofloxacin, are commonly used for the treatment of diarrhoea [Reference Engberg9]. Thus, an increased incidence of fluoroquinolone resistance among human C. jejuni and C. coli isolates in Thailand is of major public health concern [Reference Isenbarger4, Reference Hoge10, Reference Sanders11]. High levels of ciprofloxacin resistance in C. jejuni have already been reported in isolates from Thai poultry [Reference Hanson12, Reference Niyomtham and Kramomthong13]. Erythromycin is an effective drug for the treatment of campylobacteriosis and Campylobacter spp. are normally susceptible to other macrolide drugs [Reference Engberg9]. However, resistance to erythromycin amongst C. jejuni and C. coli isolates from diarrhoeal patients in Thailand has been reported [Reference Isenbarger4, Reference Hoge10, Reference Kuschner14] and it is feared that campylobacter will acquire resistance to both fluoroquinolones and macrolides.
The serotypes of C. jejuni and C. coli are traditionally determined by the detection of heat-stable or heat-labile (HL) surface antigens as described by Penner's and Lior's serotyping systems, respectively [Reference Woodward and Rodgers15]. Lior's scheme is based on direct agglutination of bacterial suspensions by antiserum while the scheme of Penner utilizes passive haemagglutination. These schemes were combined by Woodward & Rodgers [Reference Woodward and Rodgers15] for use in reference laboratories but antisera to Penner's types have recently become available commercially.
Given the importance of Campylobacter spp. as a major diarrhoeal pathogen in Thailand with increased antimicrobial resistance to public health, we describe here the development of fluoroquinolone (nalidixic acid and ciprofloxacin) and macrolide (erythromycin and azithromycin) resistance over time and the distribution of serotypes amongst C. jejuni and C. coli isolated from Bangkok and rural Thai children during 1991–2000.
METHODS
Source of Campylobacter spp. isolates
A total of 968 C. jejuni and 200 C. coli isolates was obtained from surveillance studies of diarrhoea in children aged between 0 and 10 years. These studies were conducted under protocol with informed consent by the Department of Enteric Diseases, Armed Forces Research Institute of Medical Sciences (AFRIMS) in Thailand from 1991 to 2000. The study sites were based either in Bangkok or in a rural location. Isolates of C. jejuni and C. coli from Bangkok were from the Children's Hospital and Bumrasnaradul Hospital in Nontaburi Province, located in the Bangkok Metropolitan vicinity. C. jejuni and C. coli isolates from rural settings originated from Nongkai and Tabor Hospitals located in northeastern Thailand and Nakornsrithammarat Hospital in southern Thailand. Six additional rural study sites were in Rajburi Province, 150 km west of Bangkok, and one study was conducted in Tak Province in northern Thailand.
Isolation, identification and serology
All stool specimens were processed by a modified filtration method originally described by Steele & McDermott [Reference Steele and McDermott16]. Briefly, a 47 mm, 0·45 μm sterile cellulose acetate membrane filter (Millipore, Bedford, MA, USA) was placed centrally on the surface of Brucella agar (Difco, Detroit, MI, USA) with 5% sheep blood (BAP) in 90 mm Petri dishes. Five to six drops of faecal suspension, both before and after enrichment in modified Doyle's broth [Reference Doyle and Roman17], were applied to the membrane using care to keep drops separated. The samples were allowed to filter for 30 min in ambient air before the membranes were removed. The inoculated BAPs were incubated at 37°C for up to 72 h in microaerobic conditions (6% O2, 6% CO2, 3% H2 and 85% N2) [Reference Le, Lasticova, Lasticova, Newell and Lasticova18] generated by GasPak™ (BD Diagnostic Systems, Sparks, MD, USA) with daily examination for campylobacter appearing colonies. Isolates were characterized by Gram stain, morphology, oxidase and catalase tests, and further conventional phenotypic tests including hippurate hydrolysis [Reference Morris19], nitrate reduction, formation of H2S in triple sugar iron medium, oxygen tolerance, and growth at 25, 37 and 42°C. All overnight phenotypic tests were incubated in a microaerobic environment, except for oxygen tolerance which was incubated in ambient air. C. jejuni and C. coli isolates were serotyped according to Lior's scheme with antiserum sets for 33 of the common serotypes which were obtained from the National Laboratory for Enteric Pathogens, formerly located in Ottawa, Ontario, Canada [Reference Lior20].
Antimicrobial susceptibility testing
C. jejuni and C. coli were tested for antimicrobial susceptibility by disk diffusion [Reference Bauer21] with modifications described herein. Eighteen to 48 h BAP subcultures of C. jejuni and C. coli isolates were suspended in Mueller–Hinton broth (BD Diagnostic Systems) to obtain a turbidity equivalent to a 1.0 McFarland standard, and inoculated onto Mueller–Hinton II agar (BD Diagnostic Systems) supplemented with 5% sheep blood. Isolates were tested for susceptibility to the following antibiotics: nalidixic acid (NAL, 30 μg), ciprofloxacin (CIP, 5 μg), erythromycin (ERY, 15 μg), and azithromycin (AZM, 15 μg). It should be noted that all C. jejuni and C. coli isolated during 1991–1995 were tested for susceptibility to NAL, but not all were tested against CIP, ERY and AZM during these 5 years. Disks were placed on the surface of inoculated Mueller–Hinton II agar plates and incubated at 37°C for 24 h in a microaerobic environment. As no standardized interpretive criteria exists for Campylobacter spp., the inhibition zone diameters were compared against NCCLS standard guidelines for aerobic Gram-negative bacilli to interpret results as susceptible, intermediate or resistant.
Exclusion, inclusion criteria and data analysis
The available data for microbiological characteristics and antimicrobial susceptibility were combined. Repeat samples and multiple isolates from the same child were included only if different serotypes of C. jejuni or C. coli were isolated. The statistical analyses below were performed in statxact version 7 (Cytel Inc., Cambridge, MA, USA). The trend of the percent of isolates of each species in each setting resistant to each antibiotic was tested by the Cochran–Armitage trend exact P value procedure. The common odds ratio between resistance and setting with 95% confidence intervals was computed for each species for each antibiotic and the test for homogeneity of odds ratio was performed using the homogeneous association for 2×2×k tables procedure.
RESULTS
Distribution of C. jejuni and C. coli and their serotypes
The relative fraction of C. jejuni and C. coli isolates in Thai children from 1991 to 2000 is shown in Table 1. Both C. jejuni and C. coli were isolated more frequently from diarrhoea cases than from asymptomatic controls (1109 diarrhoea cases, 159 controls). C. jejuni represented approximately 75–85% of total numbers of Campylobacter spp. isolates.
Origin of C. jejuni and C. coli isolates from Thai children during 1991–2000
* From 1109 cases and 159 controls.
Resistance trends among C. jejuni and C. coli isolates
The temporal data for fluoroquinolone and macrolide resistance for C. jejuni and C. coli from Bangkok and rural settings is shown in Figures 1 and 2 respectively. There were significant increasing trends from 1991–2000 for NAL resistance among C. jejuni and C. coli isolates in both Bangkok and rural settings. In Bangkok (Fig. 1a), C. jejuni isolates showed a marked increase in NAL resistance from 5% in 1992 to 86–100% in 1999–2000 (P<0·0001). In comparison, in rural areas, NAL resistance among C. jejuni isolates increased from 0% to 1% in 1991–1992 to only 47% in 1999–2000 (P<0·0001). NAL resistance among C. coli isolates in Bangkok (Fig. 1b) was 28% in 1992 and increased to 80–100% in 1999–2000 (P<0·0001), while among C. coli isolates from rural areas, NAL resistance increased from 0% in 1991–1993 to 50–60% in 1999–2000 (P<0·0001).
Trends of nalidixic acid and ciprofloxacin resistance among C. jejuni and C. coli isolates from Bangkok and rural settings during 1991–2000; percent resistance by year and confidence interval (95% CI). * Cochran–Armitage trend exact P value (two-sided) <0·0001. ** Cochran–Armitage trend exact P value (two-sided) <0·01. † Cochran–Armitage trend exact P value (two-sided)=non-significant. NAL, Nalidixic acid; CIP, ciprofloxacin; BKK, Bangkok; CJ, C. jejuni; CC, C. coli.
Although resistance data for CIP, ERY, and AZM were only available for 1996–2000, the percent of C. jejuni isolates in Bangkok resistant to CIP (Fig. 1c) increased significantly from 76–80% in 1996–1997 to 100% in 2000 (P<0·01). The number of C. jejuni isolates from Bangkok and rural settings resistant to ERY and AZM remained low from 1996 to 2000. However, resistance to these agents in C. coli isolates from Bangkok fluctuated from 14% to 36% over the same period but these were absent from rural areas (Fig. 2).
Trends of erythromycin and azithromycin resistance among C. jejuni and C. coli isolates from Bangkok and rural settings during 1996–2000; percent resistance by year and confidence interval (95% CI). † Cochran–Armitage trend exact P value (two-sided)=non-significant. ERY, Erythromycin; AZM, azithromycin; BKK, Bangkok; CJ, C. jejuni; CC, C. coli.
The common odds ratio comparing the resistance by setting was computed for C. jejuni and C. coli for each antibiotic (Tables 2 and 3). In general, the odds ratio for resistance to the fluoroquinolones was significantly greater in Bangkok than in rural settings among both C. jejuni and C. coli isolates. Despite the limited data, the odds of resistance to the macrolides, ERY and AZM, were greatest among C. coli isolates in Bangkok.
Percentage of fluoroquinolone resistance and statistical analysis between C. jejuni and C. coli isolates in Bangkok and rural settings in Thailand during 1991–2000
* Data not available.
Percentage of macrolide resistance and statistical analysis between C. jejuni and C. coli isolates in Bangkok and rural settings in Thailand during 1996–2000
* Data not available.
Serotype distribution
The serotype distribution of C. jejuni and C. coli isolated from Bangkok and rural settings is shown in Table 4. The most common C. jejuni serotype in both settings was HL 36. C. jejuni HL 2, 4, 7, 9 and 11 each accounted for 4–11% of isolates in both settings but C. jejuni HL 94 was not recovered from cases in Bangkok. For C. coli the most common serotypes in Bangkok were HL 8, 29, and 55 (9–16%) while HL 8, and 44 were predominant in rural areas (>10%). Twenty to 40% of all C. jejuni and C. coli isolates were not typable with the available antisera.
Serotype distribution (%) of 968 C. jejuni and 200 C. coli isolated from Thai children in Bangkok and rural settings (1991–2000)
* Numbers within parentheses indicate first (1), second (2) or third (3) rank in each setting.
DISCUSSION
This study has shown that C. jejuni and C. coli isolated from Thai children with diarrhoea exhibited significant trends of increasing resistance to NAL over 10 years. Moreover the incidence of resistance to both fluoroquinolones tested was significantly greater in isolates from Bangkok than rural settings (common OR >1). In contrast, macrolide resistance among both species in Bangkok was not significantly different from the rural setting.
Despite limited susceptibility data for CIP in isolates from 1991–1995, our results indicate a significant trend for increased resistance to CIP for C. jejuni isolates from Bangkok and moreover, most NAL-resistant isolates from 1996 to 2000 also showed resistance to CIP. A possible explanation for the higher fluoroquinolone resistance among city isolates is easier access to, and more frequent usage of these antimicrobial agents to treat human infections. Furthermore, these agents are widely used in veterinary medicine for treatment and prophylaxis or as growth promoters in animal and aquaculture farming system to increase production yields. Such usage may facilitate the development of antimicrobial resistance and accumulation of antimicrobials throughout the food chain. In Bangkok, meat and poultry products come from commercial farms, whereas in rural areas people tend to obtain these products from their own farms where few or no antimicrobial agents are used to raise such animals. A similar increase in fluoroquinolone resistance has been observed in both human and animal Campylobacter spp. from Thailand as well as in a number of other countries [Reference Engberg9, Reference Hanson12, Reference Li22–Reference Murphy24].
In this study, the number of human C. jejuni isolates resistant to ERY was low. This contrasts with a previous report in which more than half of C. jejuni isolates from poultry in Thailand during 2002–2003 were resistant to ERY [Reference Boonmar25]. We were not able to collect data on food contamination associated with diarrhoea but it is unlikely that poultry was the only source of the human C. jejuni infections. An explanation of the more frequent macrolide resistance found in C. coli, compared with C. jejuni, might be related to both microorganisms and host. Tylosin (a macrolide derivative) is used for growth promotion in pig farms and C. coli are more commonly present as normal flora in pigs than in poultry [Reference Aarestrup26]. This might indicate that pork was a major source of C. coli infection in this study. However, well-designed epidemiological follow-up studies are needed to access the degree of association between pork consumption and C. coli infections.
Relatively few (2·5%) of 1168 of isolates including 11 C. jejuni and 18 C. coli isolates were resistant to both fluoroquinolones and macrolides (data not shown) which is similar to our previous study of co-resistance patterns between NAL and AZM in Campylobacter spp. and Salmonella spp. isolated in Thailand [Reference Isenbarger4]. No single serotype was associated with these co-resistance isolates. Subsequent antimicrobial susceptibility testing of 413 C. jejuni and 85 C. coli isolates during 2001–2005 from three studies in Thailand showed that the rate of this co-resistance among campylobacter isolates remained unchanged (L. Bodhidatta, unpublished observations). However, the finding of co-resistance to fluoroquinolones and macrolides in campylobacters is alarming as these two drug classes are the major antimicrobials currently used to treat campylobacteriosis and other enteric bacterial diarrhoea. Close monitoring of the development of fluoroquinolone, macrolide resistances and their co-resistance in Campylobacter spp. should be undertaken in Thailand and elsewhere.
Data from the three unpublished studies cited above (L. Bodhidatta, unpublished observations) demonstrated high fluoroquinolone resistance (50–90% of isolates) but continued low macrolide resistance (0–14%) in Thailand. According to current Thai Ministry of Public Health (MOPH) guideline, quinolones are the first-line drugs of choice for the treatment of diarrhoea unless Campylobacter spp., is the cause of the diarrhoea, where ERY is recommended for treatment [27]. However, only a few laboratories in Thailand have the capability for isolation and identification of Campylobacter spp. including antimicrobial susceptibility testing. This high incidence of fluoroquinolone-resistant campylobacter isolates will inevitably result in prolongation of the illness or treatment failure in a number of cases.
The most common serotypes identified here, C. jejuni HL 36 and C. coli HL 8, were the five most common serotypes associated with human campylobacteriosis in a previous study of global isolates [Reference Woodward and Rodgers15]. In our C. coli isolates, HL 29 and 44 were the second leading serotype from urban and rural settings, but they were uncommon elsewhere [Reference Woodward and Rodgers15]. There was little difference in the distribution of serotypes between Bangkok and rural areas and the frequency of serologically non-typable isolates was unacceptably high. Thus serotyping alone is insufficient for studies of the epidemiology of C. jejuni and C. coli and a combination of more discriminant molecular techniques is necessary to expand our understanding of the molecular epidemiology of Campylobacter spp. infections in Thailand and elsewhere.
ACKNOWLEDGEMENTS
All study projects described here were financially supported by the Military Infectious Diseases Research Program, Walter Reed Army Institute of Research, Washington, DC, USA. We are grateful to staff and nurses from the Department of Enteric Diseases at the Armed Forces Research Institute of Medical Sciences (AFRIMS) for their excellent work during the period of studies, Duangjai Lamsan for her great effort in collecting and entering information into databases, and especially, Caroline Fukuda for proofreading, and contributing constructive input to the discussion.
DECLARATION OF INTEREST
None.
