PFGE, Lior serotype, and antimicrobial resistance patterns among Campylobacter jejuni isolated from travelers and US military personnel with acute diarrhea in Thailand, 1998-2003

From a total of 312 C. jejuni isolates obtained in this study, 266 isolates were from U.S. soldiers and 46 were from foreign travelers seen at Bumrungrad Hospital, Bangkok. The prevalence of C. jejuni from diarrheal cases was detected as 11.0% (46/417) in the Bumrungrad Hospital study and 30.4% (266/875) in samples from U.S. military personnel in the Cobra Gold exercises of 1998-2003 (Table 1). Overall, a total of 16 Lior serotypes were detected; the three most common serotypes were HL: 36, HL: 11, and HL: 5; these serotypes accounted for 25.0% (78/312), 13.8% (43/312), and 7.4% (23/312) of all of the isolates, respectively. Untypable strains composed 17.6% (55/312) of the isolates. The distribution of the serotypes and the incidence of C. jejuni isolates among foreign travelers and U.S. military personnel are shown in Table 2. The two most common serotypes by study location and year are shown in Table 1.

The incidence of the serotypes described herein were included from the first to tenth ranks among the global isolates [25] but were slightly different from those reported previously in Thai children [24]. For example, serotype HL: 5 was detected infrequently among Thai children [24]. In this study, serotype HL: 5 was more common, and serotypes HL: 9 and HL: 2 were less common. The majority of C. jejuni isolates from serotype HL: 5 (11/23) were isolated in 2001, and most C. jejuni HL: 19 isolates (15/17) were obtained in 1999. The untypable isolates comprised 54.3% (25/46) of the isolates from foreign travelers in 2001-2002 compared to 11.3% (30/266) of the isolates from U.S. military personnel in this study. The adult travelers from Bumrungrad Hospital spanned several nationalities, including Japanese, European, American, and Australian. The high percentage of untypable isolates might reflect the diverse foods consumed by these travelers, while the low percentage of untypable isolates in the military personnel might reflect the limited diet consumed by U.S. military personnel on deployment.

At an 80% similarity level, a dendrogram combining data for all 312 C. jejuni isolates from Thailand was clustered into 62 genotypes (Figure 1). Four major genotypes composed 49.7% (155/312) of the C. jejuni isolates in this study, and 30 of the 62 genotypes included only a single C. jejuni isolate. Of the 55 isolates that were untypable by Lior serotyping, 42 could be grouped into a genotype with other known serotypes.

In 1998-2003, 94.9% (296/312) and 93.0% (289/311) of the isolates were resistant to quinolones (NAL and CIP), but 99.0% (306/309) were susceptible to macrolides (ERY and AZM). The high prevalence of quinolone (NAL and CIP) resistance and macrolide (ERY and AZM) sensitivity is consistent with previously reported results from Thailand [21, 24, 26]. For the 138 C. jejuni isolates from 2001-2003, the percentages of resistant isolates detected were as follows (Table 3): CF, 100% (138/138); TE, 81.9% (113/138); SXT, 58.0% (80/138); AMP, 30.4% (42/138); KM, 6.5% (9/138); SU, 3.6% (5/138); NM, 2.2% (3/138); and SM, 0.7% (1/138). None of these 138 isolates were resistant to GM, CM or CL. Notably, tetracycline resistance was also detected in over 80% of isolates that were similar to C. jejuni isolates from Taiwan (95%) [27], Korea (87%) [6, 28], Canada, and the U.S. (50%) [29, 30].

The two most common resistance patterns observed in these 138 isolates were multiple resistance to four antimicrobials (NAL, CIP, CF, and TE), observed in 79.0% (109/138) of the isolates, and multiple resistance to five antimicrobials (NAL, CIP, CF, TE, and SXT), detected in 47.8% (66/138) of the isolates. Another resistance pattern (NAL, CIP, CF, TE, SXT, and AMP) was found in 15.9% (22/138) of the isolates, and a fourth pattern of resistance (NAL, CIP, CF, TE, and KM) was detected in 5.8% (8/138) of the isolates. The first common pattern was similar to a previous report of antimicrobial resistance to NAL, CIP, and TE in 53% of clinical C. jejuni isolates in Thailand [31]. These results confirm widespread quinolone and tetracycline resistance among C. jejuni isolates from traveler's diarrhea in Thailand.

Antimicrobial resistance by study location and the results of the chi-square test are shown in Table 3. Interestingly, KM-resistant isolates were detected in 6.5% (9/138) of the isolates, but there was a significant difference (from 2.2% to 21.8%) in the frequency of resistance in isolates from different locations (p < 0.001). The percentage of AMP-resistant isolates also varied by location (p < 0.05). The resistance of isolates to SXT varied greatly from 90.7% at Phitsanulok to 58.7% at Bumrungrad Hospital, 26.7% at Sakaew, and to undetectable levels in isolates from Pranburi (p < 0.001). The finding of differences in AMP, KM, and SXT resistance among C. jejuni isolates from selected sites should not be considered as indicative of significant changes over time in Thailand.

In this study, the antimicrobial susceptibility tests were performed by disk diffusion assay. Other methods, including agar dilution and broth micro dilution methods, and epsilometer test (E-tests) have been used by different laboratories to measure antimicrobial susceptibilities for Campylobacter spp. [32‐35]. However, good agreement of antimicrobial susceptibility test between disk diffusion and agar dilution tests has been observed in several classes of antimicrobials especially quinolone/fluoroquinolones and aminoglycosides suggesting that disk diffusion test could be used as qualitative assay but not quantitative assay for antimicrobial susceptibility among Campylobacter spp. [36]. Although another study suggested that interpretation of erythromycin by disk diffusion assay was unreliable and should be confirmed by MIC-based methods [37], but our previous data showed high correlation of antimicrobial susceptibility by disk diffusion and agar dilution tests in erythromycin and azithromycin-resistant C. jejuni and C. coli isolates [38].

Dendrograms A thru D in Figs. 2, 3, 4, and 5 were generated for four study locations from 2001-2003. Dendrogram A includes 46 isolates obtained from Bumrungrad Hospital, Bangkok, over the two-year period from 2001-2002 (Figure 2). Although the common resistance pattern (NAL, CIP, CF, and TE) was observed in 76.1% (35/46) of the isolates, there was substantial heterogeneity in the PFGE patterns (Figure 2). The PFGE analysis suggests diverse genotypes. No specific multi-resistant antimicrobial patterns were noted in the remainder of these isolates. The dendrogram and the diversity of serotypes and antimicrobial susceptibility patterns suggest that these isolates are heterogeneous and clonally diverse.

Dendrogram B includes the 54 isolates collected from U.S. soldiers in Phitsanulok in 2001 (Figure 3). As above, 87% (47/54) of these isolates were generally multi-resistant to NAL, CIP, CF, and TE; in addition, 87% (47/54) were resistant to NAL, CIP, CF, and SXT. The genotype B3 included 24 isolates exhibiting 86.1% similarity that belonged to three serotypes: HL: 5 (11), HL: 102 (3), and HL: untypable (10). A subset of 12 isolates (B3a) within the genotype B3 had 99.7% similarity and resistance to AMP, suggesting clonality (Figure 3). Similarly, genotype B1 consisted of five isolates of serotype HL: 36 with 100% similarity and an identical antimicrobial resistance pattern (NAL, CIP, CF, TE, and SXT). C. jejuni isolates in genotypes B1 and B3a were isolated over a 5-day period and 8-day period, respectively.

Dendrogram C (Figure 4) shows 15 isolates collected from U.S. soldiers in Sakaew over a 1-month period in 2002. Genotype C1 includes 5 isolates of serotype HL: 36 with 99.2% similarity and an identical antimicrobial resistance pattern (NAL, CIP, CF, AMP, and TE), which also suggests clonality. The other 10 isolates had <67.5% similarity and included five serotypes and seven antimicrobial resistance patterns, suggesting more diverse sources (Figure 4). All C. jejuni isolates in genotype C1 were isolated over a period of 6 days.

Dendrogram D (Figure 5) shows the 23 isolates collected from U.S. soldiers in Pranburi during a 1-month exercise in 2003 and genotype D1 includes a unique cluster of five serotype HL: 4 isolates with an unusual multi-resistant antimicrobial pattern (NAL, CIP, CF, KM, and TE), suggesting clonality for these isolates. Similar to dendrograms B and C, genotype D1 included C. jejuni isolates collected during a 10-day period.

Under approved human use protocols, stool specimens were obtained from patients with diarrhea and from asymptomatic controls in an acute diarrhea study among foreign travelers from highly developed countries at Bumrungrad Hospital in Bangkok, Thailand during 2001-2002. Stool specimens were collected from U.S. soldiers with acute diarrhea and from asymptomatic controls; the soldiers were deployed for the Cobra Gold exercises lasting one for four weeks at different sites in Thailand during 1998-2003. Only C. jejuni isolates from acute diarrhea cases were included in this study. Table 1 describes the number of C. jejuni isolates from cases in each study location.

All stool specimens were cultured for Campylobacter spp. using a modified filtration method described previously [41]. Suspected colonies, growing on Brucella Agar (Difco, Detroit, MI, USA) with 5% sheep blood (BAP), were identified as Campylobacter spp. by colony characteristics, Gram staining, oxidase tests, and catalase tests, followed by phenotypic tests including hippurate hydrolysis, nitrate reduction, H₂S TSI, oxygen tolerance, and microaerobic growth at 25°C, 37°C, and 42°C. C. jejuni isolates were differentiated from C. coli by the hippurate hydrolysis test. All C. jejuni isolates were kept in glycerol medium at -70°C for further analysis.

Lior serotyping was performed by an agglutination assay with specific antiserum obtained from the National Laboratory for Enteric Pathogens (NLEP) in Winnipeg, Manitoba, Canada. These antisera were routinely used to serotype C. jejuni and C. coli isolates at AFRIMS. The antisera detect heat-labile antigens [9] and identify 33 common HL serotypes.

C. jejuni isolates were tested for susceptibility to antimicrobial drugs using a disk diffusion assay as described previously [42], with modifications. BAP subcultures of patient isolates at 18- to 48-h were suspended in Mueller Hinton broth (BD Diagnostic Systems, Sparks, MD, USA.) to obtain a turbidity equivalent to a 1.0 McFarland standard, and suspensions were inoculated onto Mueller Hinton II agar supplemented with 5% sheep blood. At the time of each study, all C. jejuni isolates were tested for susceptibility to the following antimicrobials (BD Diagnostic Systems): NAL (30 μg), CIP (5 μg), ERY (15 μg), and AZM (15 μg). The 138 C. jejuni isolates in 2001-2003 were further tested for susceptibility to 11 additional antimicrobials by disk diffusion assay. These antimicrobial disks included AMP (10 μg), CM (30 μg), KM (30 μg), GM (10 μg), SM (10 μg), TE (30 μg), SXT (1.25/23.75 μg), SU (250 μg), CL (10 μg), NM (30 μg), and CF (30 μg). Disks were placed on the surfaces of inoculated Mueller Hinton II agar plates. Inoculated plates were incubated at 37°C for 24 h in a microaerobic environment. The plates were re-incubated up to 48 h if insufficient growth of C. jejuni isolates on the Muller Hinton II agar plates was obtained at 24 h. Because no standardized interpretive criteria exist for Campylobacter spp., the inhibition zone diameters were measured and interpreted following the disk manufacturer's instructions and compared against the Clinical and Laboratory Standards Institute (formerly NCCLS) standard guidelines for aerobic gram-negative bacilli to interpret the results as susceptible, intermediate, or resistant [43]. Escherichia coli ATCC 25922 and Staphylococcus aureus ATCC 25923 were used as standard organisms for all disk diffusion assays.

Pearson's chi-square tests of independence for the antimicrobial susceptibility data (NAL, CIP, ERY, AZM, TE, SXT, AMP, and KM) between locations were performed using the Monte Carlo-exact (2-sided) method in SPSS version 12.0 (SPSS Inc., Chicago, IL, USA). A p-value < 0.05 was considered significant.

PFGE was performed according to the One-Day (24-28 h) Standardized Laboratory Protocol for Molecular Subtyping by the CDC [12] with the minor modifications described below. The cell density of each isolate was adjusted to an O.D. of 0.6 using a spectrophotometer (Spectramax 190; Molecular Devices, Sunnyvale, CA, USA) that was different from the spectrophotometers suggested by the CDC. The PFGE patterns were analyzed to generate dendrograms of the combined Sma I and Kpn I similarities using BioNumerics version 5.0 (Applied Maths, Sint-Martens-Latem, Belgium) by UPGMA type and Dice coefficient with 1.5% optimization and tolerance. Dendrograms were made for the composite data of all of the C. jejuni isolates and for the different locations.