Antibiotic resistance, molecular characterizations, and clinical manifestations of Campylobacteriosis at a military medical center in Hawaii from 2012–2016: a retrospective analysis

Hawaii has one of the highest incidences of Campylobacteriosis in the United States, but there remains little published data on circulating strains or antimicrobial resistance. We characterized 110 clinical Campylobacter isolates (106 C. jejuni, 4 C. coli) processed at Tripler Army Medical Center in Honolulu, HI from 2012–2016. Twenty-five percent of C. jejuni isolates exhibited fluoroquinolone (FQ) resistance, compared with 16% for tetracycline (TET), and 0% for macrolides. Two of the four C. coli isolates were resistant to FQ, TET, and macrolides. C. jejuni isolates further underwent multilocus sequence typing, pulsed-field gel electrophoresis, and molecular capsular typing. Nineteen capsule types were observed, with two capsule types (HS2 and HS9) being associated with FQ resistance (p < 0.001 and p = 0.006, respectively). HS2 FQ-resistant isolates associated with clonal complex 21, possibly indicating clonal spread in FQ resistance. Macrolides should be considered for treatment of suspect cases due to lack of observed resistance.


Study Population and Clinical Chart Reviews. This protocol was approved by the Tripler Army Medical
Center (TAMC) Institutional Review Board (Protocol #15R28). TAMC is a tertiary care referral hospital in Honolulu, HI serving DoD and VA beneficiaries in the Asia-Pacific. Informed consent was waived by the TAMC Institutional Review Board, and all research was conducted in accordance with appropriate guidelines and regulations. Stored Campylobacter isolates collected from stool samples submitted for evaluation of symptomatic diarrheal disease were included in this study. The beneficiary population in Hawaii is well-integrated into the local economy, participating in similar activities and consuming from the same food sources as the local population.
Retrospective chart reviews were performed on the laboratory-confirmed cases. Epidemiological data included age, sex, recent antibiotic use (within 3 months), occupation, potential food exposures, travel history, and pets. Pediatric patients were divided into two groups: 0-8 years and 9-18 years due to side effect concerns with specific antibiotic classes (tetracyclines and fluoroquinolones). Adults were subgrouped by decade until 40 years of age since these groups reflect different populations seen at our facility: young active duty military and deployment populations, longer-serving military populations, and retiree and veteran populations. We reviewed clinical data, including temperature, vomiting, diarrhea, bloody stool, fecal leukocytes, white blood cell count, serum chemical parameters (liver-associated enzymes, serum creatinine), treatment offered, and response to therapy.

Clinical Samples, Microbiology, and Antimicrobial Sensitivity Testing. One hundred ten
Campylobacter isolates obtained from January 2012 -February 2016 were shipped to the Armed Forces Research Institute of Medical Sciences (AFRIMS) in Bangkok, Thailand using Microbank TM cryobeads (Prolab Diagnostic, Round Rock, TX). Each sample was inoculated into Preston selective enrichment broth and incubated under microaerobic conditions (37 °C, 10% CO 2 , 5% O 2 and 85% N 2 ) for 24 hours. Subsequently, the isolates were sub-cultured on Brucella agar plate with 5% sheep blood and incubated at 37 °C under microaerobic conditions. After 48-72 hours, isolate identification was confirmed using motility examination and phenotypic testing including oxidase, catalase, indoxyl acetate hydrolysis, rapid hippurate hydrolysis, nitrate reduction, growth temperature and oxygen tolerance tests 23 . As part of an investigation into emerging culture-independent microbial identification methods, a subset of isolates (n = 49) were also analyzed using polymerase chain reaction/electrospray ionization mass spectrometry. This method found that all tested isolates could be identified to the genus level and were consistent with the conventional identification and molecular testing performed at AFRIMS (see Supplementary Information, Supplementary Table S1).
Antimicrobial susceptibility testing to azithromycin (AZM), erythromycin (ERY), nalidixic acid (NAL), ciprofloxacin (CIP), tetracycline (TET) and ceftriaxone (CRO) was performed on confirmed isolates using commercially available E-tests (Liofilchem, Roseto degli Abruzzi TE, Italy) to determine the minimal inhibitory concentration (MIC). The latter was used as an internal control since C. jejuni is resistant to cefoperazone. Susceptibility results were interpreted following Clinical and Laboratory Standards Institute (CLSI) guidelines and National Antimicrobial Resistance Monitoring System (NARMS) using Campylobacter jejuni ATCC 33560 as the control strain 24 . The CLSI guidelines do not offer susceptibility recommendations for Campylobacter to CRO, so were interpreted following the guidelines of Enterobacteriaceae 25 . Capsular Typing. Capsule typing was performed on genomic DNA extracts of C. jejuni isolates with four multiplex primer sets using 36 specific primers targeting capsule genes as developed and PCR reactions developed by Poly et al. 26,27 . Two µL of each C. jejuni isolate DNA was subjected to each multiplex PCR in a 25 µL reaction mixture containing 1X PCR buffer (10 mM Tris-HCl, pH 8.3, 50 mM KCl), 2.0 mM MgCl2, 300 µM concentration of each dNTPs (deoxynucleotide triphosphate), 0.4 µM of each primers sets (Alpha, Beta, Gamma and Delta) and 2.5 U of AmpliTaq Gold DNA polymerase. Amplification steps were as follows: 94 °C for 5 min; 28 cycles of 94 °C for 1 min, 52 °C for 1 min and 72 °C for 1 min and a final extension step at 72 °C for 10 min. Amplicons were visualized after gel electrophoresis at 120 V for 1 hour on a 2.0% Agarose-1000 gel (Invitrogen, SCIENTIFIC REPORTS | (2018) 8:11736 | DOI:10.1038/s41598-018-29461-z USA) for Beta and Gamma set and 2.5% Agarose-1000 gel for Alpha and Delta set and staining with ethidium bromide. DNA of C. jejuni of known capsule types and 2-log DNA ladder (New England BioLabs, USA) were used as positive controls and a size marker, respectively.
Multilocus Sequence Typing (MLST). MLST was performed according to developed protocol on seven housekeeping genes (protein product shown in parentheses): aspA (aspartase A), glnA (glutamine synthase), gltA (citrate synthase), glyA (serinehydroxymethyl transferase), pgm (phosphoglucomutase), tkt (transketolase), and uncA (ATP synthase-α) 28 . Sequences-based identification of MLST profiles used Bionumerics Version 7.5 with the MLST plugin (Applied Maths NV, Belgium). Isolates were characterized by their sequence type (ST) and as members of clonal complexes (CC). The MLST profiles identified in this study were submitted to the PubMLST database http://pubmlst.org/campylobacter 29 . Pulsed Field Gel Electrophoresis (PFGE). PFGE of C. jejuni using SmaI was performed according to standard protocol 30 . The agarose DNA plug was digested with 40 U SmaI (Roche, Germany) according to the manufacturer's instructions. PFGE was performed in a CHEF Mapper system (Bio-Rad, USA) at 14 °C in 0.5x TBE (Tris/borate/EDTA). Run times and pulsed times were 6.76-35.38 s for 18 h with linear ramping. Gels were stained with ethidium bromide (0.5 µg/ml), and band patterns were visualized by Gel Documentation System (Syngene, United Kingdom). Gel images were analyzed with BioNumerics version 7.5 (Applied Maths, Belgium) to obtain a phylogenetic tree. Cluster analysis of the Dice similarity indices based on UPGMA was done to generate a dendrogram describing the relationship among each C. jejuni isolate.

Statistical analysis.
Chi-square tests and Fisher's exact tests were done to assess categorical associations between variables. All statistical analyses were conducted using SAS version 9.2 (SAS Institute, Cary, NC). Data availability. The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.
Disclaimer. The views expressed in this article by the authors do not represent those of the United States Army, the Department of Defense, or the United States Government.

Results
Clinical Data. A retrospective chart review of 110 patients with positive cultures was performed to evaluate the clinical presentation and potential risk factors for Campylobacteriosis in Hawaii. Species was identified as C. jejuni in 106 isolates and C. coli in 4 isolates. The demographic breakdown of cases can be seen in Table 1. Seasonality of infection was observed: 61% of C. jejuni isolates were collected between January and June compared with 39% from July through December; over half (53%) were collected between May and August, suggesting a relative summer peak. A travel history was obtained in 69 patients (65 C. jejuni and 4 C. coli), and 29% (n = 20) reported international travel within 3 months prior to presentation. All four C. coli cases reported recent international travel, compared with only 23% of C. jejuni cases who reported a travel history (p = 0.005). Comparisons regarding common clinical signs and symptoms can be seen in Table 2. Patients ≤8 years were more likely to present with hematochezia (90% vs 48%, p < 0.001), but were less likely to present with abdominal pain or cramping (82% vs. 99%, p = 0.014). Additional laboratory data such as serum chemistries were collected in 56% of patients (n = 62). No patients had liver-associated enzyme elevations greater than twice the upper limit of normal, and only 8% (n = 5) presented with elevated serum creatinine >1.25 mg/dL. Fecal leukocytes were present in 37% of the 48 patients tested. Twenty-two patients (22%) had documented antimicrobial use in the three months preceding infection: beta-lactams (n = 7), fluoroquinolones (n = 3), tetracyclines (n = 4), clindamycin (n = 2), nitrofurantoin (n = 1), trimethoprim-sulfamethoxazole (n = 2), anti-fungals (n = 1). TET resistant isolates were detected in only two patients with a recent history of antibiotic use.
Overall, 62% of patients received antibiotic treatment (53% of ages 0-18, 65% aged 19+) of patients. Patients ≤18 years were twice as likely to be treated with azithromycin compared with those 19 years or older (44% vs. 22%, p = 0.034). Few complications were seen in our patient population: Four developed colitis, one patient had appendicitis, and one patient had biliary dyskinesia. There was one case of post-infectious irritable bowel syndrome. No cases of GBS were observed. Ancillary historical data including occupation, pet/animal exposure, or specific food exposures were too seldom documented to provide reasonable interpretation.
Antimicrobial Susceptibility Testing. All isolates underwent AST using an E-test method (Table 3).
In total, 26% (n = 29) of our Campylobacter isolates were FQ-resistant, and 17% (n = 19) of isolates were TET-resistant (8% resistant to both). All isolates had minimal inhibitory concentrations ≥6 µg/ml to CRO (Supplementary Table S2). When broken down by species, 25% (n = 27) of C. jejuni isolates were FQ-resistant, but only 16% (n = 17) isolates were resistant to TET. Eight (7.5%) of C. jejuni isolates were resistant to both CIP and TET. No C. jejuni isolates were resistant to AZM or ERY. Two (50%) C. coli isolates from patients with documented travel to Southeast Asia (Cambodia and Philippines) were resistant to the FQs, tetracyclines and macrolides. Overall, international travel was associated with NAL and CIP resistance (53% vs. 20% for those with and without travel history, p = 0.016) and TET resistance (58% vs. 10%, p < 0.001, respectively).
Genotypic and Capsular Typing. C. jejuni capsule typing using a multiplex-PCR method, PFGE and MLST were performed to better describe circulating C. jejuni strains and determine genetic relatedness. Nineteen C. jejuni capsule types or complexes were identified, and only five isolates were nontypable using the multiplex PCR method (Fig. 1). The most common capsule types were HS2, HS4-A, and HS4-AB, accounting for half the C. jejuni isolates. Only two capsule types, HS2 and HS9, were associated with increased antibiotic resistance. The HS2 isolates were strongly associated with resistance to NAL and CIP, with 80% (16/20) demonstrating resistance, compared to 13% of other capsule types (11/86) (p < 0.001). HS2 capsular type was not associated with TET-resistance. The HS9 isolates demonstrated a 75% (3/4) resistance rate for CIP and TET. These rates were significantly higher when compared with isolates other than HS2 for CIP (10%, p = 0.006) and with all other isolates for TET (14%, p = 0.013). All four patients with HS9 isolates had documented travel to East Asia or Southeast Asia within 3 months of their infection. Of the available historical data, other capsule types were also seen in returning travelers: HS15 (n = 3, Indonesia, Philippines, Japan), HS5/31,HS15 (n = 1, Korea). Only 1 HS2 isolate (#5) with a documented travel history (n = 15) was associated with recent international travel (Philippines).  A smaI dendogram using all 106 C. jejuni isolates was developed based on their PFGE genotypes (Fig. 2). The FQ-resistant HS2 isolates showed a high-degree of genetic similarity, with 11/16 of the isolates demonstrating identical genotypes. The only CIP-resistant HS2 isolates that were not members of CC-21 were singletons (#5 and #76). In comparison, the three HS9 isolates that were also FQ-resistant did not belong to any CC and

Discussion
The clinical presentations seen in this study are consistent with previously published studies on Campylobacter infection in pediatric and adult populations 3,8,31 . HS1/44 and HS4-AB capsular types have recently been associated with the development of GBS, however, no patients in our study developed GBS or polyneuropathy 32 . Our sample size is likely too small to detect the very small incidence of GBS (0.07%) 5 . We noted a higher rate (20%) of recent antibiotic use within the period three months prior to the incident infection compared to 9% in the 28 days prior reported by Effler and colleagues 10 . Although we evaluated recent antibiotic use over a longer preceding time period, our study supports the association between an increased risk of Campylobacter infection with recent antibiotic use.
NARMS data from 2014 noted US mainland CIP-resistance rates of 26.7% for C. jejuni and 35.6% for C. coli in clinical isolates 14 . The NAL and CIP-resistance rate of 25% seen in our study is comparable to that of the US mainland (p = 1.00). In contrast, the rate of TET resistance in C. jejuni isolates was significantly lower than that reported in NARMS (16% vs. 48.6%, p < 0.001). The lower TET-resistance observed in our study is somewhat surprising considering the widespread use of tetracyclines as livestock growth promoters, and the high resistance rates reported in isolates from the both the US mainland and Asia. Hawaii currently follows the Food and Drug Administration guidance on the use of antibiotics in livestock, permitting the use of tetracyclines. The C. coli isolates in our study had a higher proportion that were antibiotic-resistant when compared with C. jejuni isolates, which has been demonstrated in human and agricultural samples, and particularly in relation to macrolides 14,16,33,34 . The resistance rates seen in our population appear to more closely resemble the US mainland than the higher-reported rates seen in Asia. Unfortunately, there is no data for comparison on antimicrobial resistance in Campylobacter species in the broader Pacific islands.
Globally, the two most common reported capsular types in the literature are HS4/HS4-complexes and HS2. In their systematic review of the global distribution of C. jejuni, Pike et al. reported that of the available data, HS4-complexes comprised 23.5% of isolates in North America, while HS2 accounted for 10.7%. In contrast, HS2 capsular types were more common in Asia (11%) than HS4-complexes (8.9%) 22 . In our study, HS2 capsule types comprised 18.2% of our C. jejuni isolates and were strongly associated with FQ-resistance suggesting a potential for the development of increasing FQ-resistance in C. jejuni strains in Hawaii. The majority of HS2 isolates (ST-8098) were grouped under CC-21, which has been isolated from multiple different sources including humans, chickens, cattle, and the environment. In two recent studies, Kovac and colleagues demonstrated clonal spread of CC-21 C. jejuni isolates from multiple sources in Europe that harbored FQ resistance 35,36 . Unfortunately, C. jejuni capsule typing was not performed as part of their investigations. The demonstrated clonal spread arising from multiple sources and the high rate of FQ resistance makes HS2/CC-21 isolates of particular concern in Hawaii and emphasize the necessity of continued surveillance.
The polysaccharide capsule is a major determinant in C. jejuni immunogenicity and pathogens, and has been proposed as a target for vaccine development. Capsule-type predominance differs geographically; however, the HS4-complexes, HS2, and HS1/44 are most commonly identified. Capsular-typing methods utilizing the Penner agglutination assay are expensive and complex thus limiting its widespread use. Use of the newer and relatively easier multiplex-PCR assay allowed us to identify a higher proportion of HS2 isolates, HS4 isolates, and HS 8/17 capsule types than would have been anticipated based on previously reported data 22 . The leading C. jejuni candidate vaccine platform in development undergoing early clinical trials covers the HS23/36 and HS4-complex serotypes 37,38 . The high rate of FQ resistance in HS2 and HS9 capsule types seen in our study suggests these capsular types may need to be integrated into vaccine portfolios.
Although this is the most in-depth assessment of Campylobacteriosis in Hawaii to date, this study has several limitations. Firstly, this was a retrospective study of pre-identified samples positive for Campylobacter infection so no comparator control group was available. Secondly, the clinical information obtained on chart review was highly variable and limited by provider documentation in the medical records and the work up performed. Thirdly, the sample size was relatively small and was taken from a defined population of DoD and VA beneficiaries who commonly are living and working on the island for varying periods of time (weeks to years). While our study did not assess the local population, our patients were generally integrated into the community, engaging in the similar recreational activities and consuming from the same local food source.
While the rate of FQ-resistance rate in Campylobacter species in Hawaii is the same as the US mainland, we identified a clonal HS2/CC-21 strain with a markedly higher resistance rate than other circulating strains. With antimicrobial selection pressure, it is possible this strain will continue to propagate in the future, resulting in increased FQ resistance in Hawaii. Our study highlights the need for continued surveillance of the epidemiology and antimicrobial sensitivities of Campylobacter species in Hawaii to guide clinical treatment and to inform future vaccine candidate platforms. Due to low resistance rates, macrolides should be considered for empiric treatment of suspected Campylobacteriosis cases in Hawaii.