Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Molecular epidemiology and antimicrobial susceptibility of Clostridium difficile isolates from two Korean hospitals

  • Asiimwe Nicholas ,

    Contributed equally to this work with: Asiimwe Nicholas, Yu Kyung Kim

    Affiliation Department of Microbiology, Kyungpook National University School of Medicine, Daegu, Republic of Korea

  • Yu Kyung Kim ,

    Contributed equally to this work with: Asiimwe Nicholas, Yu Kyung Kim

    Affiliation Department of Laboratory Medicine, Yeungnam University College of Medicine, Daegu, Republic of Korea

  • Won-Kil Lee,

    Affiliation Department of Laboratory Medicine, Kyungpook National University School of Medicine, Daegu, Republic of Korea

  • Gati Noble Selasi,

    Affiliation Department of Microbiology, Kyungpook National University School of Medicine, Daegu, Republic of Korea

  • Seok Hyeon Na,

    Affiliation Department of Microbiology, Kyungpook National University School of Medicine, Daegu, Republic of Korea

  • Hyo Il Kwon,

    Affiliation Department of Microbiology, Kyungpook National University School of Medicine, Daegu, Republic of Korea

  • Yoo Jeong Kim,

    Affiliation Department of Microbiology, Kyungpook National University School of Medicine, Daegu, Republic of Korea

  • Hae Sook Lee,

    Affiliation Department of Laboratory Medicine, Kyungpook National University School of Medicine, Daegu, Republic of Korea

  • Kyung Eun Song,

    Affiliation Department of Laboratory Medicine, Kyungpook National University School of Medicine, Daegu, Republic of Korea

  • Jeong Hwan Shin,

    Affiliation Department of Laboratory Medicine, Inje University College of Medicine, Busan, Republic of Korea

  • Je Chul Lee

    Affiliation Department of Microbiology, Kyungpook National University School of Medicine, Daegu, Republic of Korea

Molecular epidemiology and antimicrobial susceptibility of Clostridium difficile isolates from two Korean hospitals

  • Asiimwe Nicholas, 
  • Yu Kyung Kim, 
  • Won-Kil Lee, 
  • Gati Noble Selasi, 
  • Seok Hyeon Na, 
  • Hyo Il Kwon, 
  • Yoo Jeong Kim, 
  • Hae Sook Lee, 
  • Kyung Eun Song, 
  • Jeong Hwan Shin


Clostridium difficile is one of the main etiological agents causing antibiotic-associated diarrhea. This study investigated the genetic diversity of 70 toxigenic C. difficile isolates from two Korean hospitals by employing toxinotyping, ribotyping, multilocus sequence typing (MLST), and pulsed-field gel electrophoresis (PFGE). Toxin gene amplification resulted in 68 A⁺B⁺ and two A-B+ isolates. Most isolates (95.7–100%) were susceptible to daptomycin, metronidazole, and vancomycin. Seventy C. difficile isolates were classified into five toxinotypes, 19 ribotypes, 16 sequence types (STs), and 33 arbitrary pulsotypes. All C. difficile isolates of ribotype 018 (n = 38) were classified into ST17, which was the most prevalent ST in both hospitals. However, C. difficile isolates of ST17 (ribotype 018) exhibited pulsotypes that differed by hospital. ST2 (ribotype 014/020), 8 (ribotypes 002), 17 (ribotype 018), and 35 (ribotypes 015) were detected in both hospitals, whereas other STs were unique to each hospital. Statistical comparison of the different typing methods revealed that ribotyping and PFGE were highly predictive of STs. In conclusion, our epidemiological study indicates that C. difficile infections in both hospitals are associated with the persistence of endemic clones coupled with the emergence of many unique clones. A combination of MLST with PFGE or ribotyping could be useful for monitoring epidemic C. difficile strains and the emergence of new clones in hospitals.


Clostridium difficile, a gram-positive obligate anaerobic spore-forming bacillus, is a leading etiological agent of antibiotic-associated clinical manifestations that range from mild diarrhea to serious pseudomembranous colitis. C. difficile infection (CDI) is of great concern because of its prevalence and high morbidity and mortality rates in both nosocomial and community-acquired infections worldwide [1, 2]. The toxins, enterotoxin A (TcdA) and cytotoxin B (TcdB), produced by toxigenic C. difficile strains play a crucial role in the pathogenesis of CDI. However, there are increasing reports of TcdA-negative/TcdB-positive (A-B+) variants in Korea and other countries [3, 4]. The prevalence of A-B+ or A+B+ C. difficile strains varies considerably across hospitals and countries [5, 6]. In addition, some C. difficile strains, either toxigenic or non-toxigenic, produce a multi-domain, actin ADP-ribosylating binary toxin (CDT), which is thought to modify actin in a manner that facilitates bacterial adhesion [7]. Outbreaks or epidemics have been attributed to specific types of C. difficile, such as restriction endonuclease analysis group BI, North America pulsed-field gel electrophoresis (PFGE) type 1, and polymerase chain reaction (PCR) ribotype 027 (NAP1/BI/027) and PCR ribotype 078 [8, 9]. PCR ribotypes 017, 014/020, and 018 have also caused nosocomial outbreaks worldwide [1012].

Several epidemiological tools such as PCR ribotyping, toxinotyping, PFGE, and multilocus sequence typing (MLST) have been used to study genotypic and phenotypic variation in C. difficile. PCR ribotyping is the most commonly used typing technique, and this technique has the power to differentiate between C. difficile isolates originating from different hospitals and countries [5]. However, band-based typing tools such as PCR ribotyping and PFGE can be difficult to interpret and use in phylogenetic analyses of C. difficile isolates. In recent years, sequence-based typing tools with online support, such as MLST (, have been established and widely used for global epidemiological study of C. difficile isolates.

Several epidemiological studies of CDI in Korea have been reported [1214]; however, these studies focused on the identification of hypervirulent epidemic strains and examination of genetic diversity among C. difficile isolates using PCR ribotyping. Ribotypes 001, 017, and 014/020 of C. difficile were the most prevalent in Korea in the last decade, but their prevalence has changed through time, with ribotype 001 decreasing and ribotypes 017, 014/020, and 018 increasing [1214]. A hypervirulent strain of ribotype 027 was also reported in a Korean hospital; however, few CDIs have been caused by this ribotype [14]. However, sequence types (STs) of C. difficile determined by MLST and their correlation with PCR ribotypes, pulsotypes established by PFGE, and toxinotypes have not yet been studied in Korea. This study investigated the clonal distribution of C. difficile isolates from two Korean hospitals using toxinotyping, ribotyping, PFGE, and MLST. In addition, the congruence of different epidemiological typing methods was assessed.

Materials and methods

Bacterial isolates and culture conditions

A total of non-duplicate toxigenic 70 C. difficile isolates from the diarrheal stools of patients who had visited or were admitted to two Korean hospitals, Kyungpook National University Hospital (KNUH) in Daegu (n = 31) and Busan Paik Hospital (BPH) in Busan (n = 39), between 2013 and 2015 were used in this study. Two hospitals were university-affiliated tertiary hospitals, and KNUH and BPH had a total of 904 and 895 beds, respectively. All C. difficile isolates were obtained from Kyungpook National University Hospital Culture Collection for Pathogens (KNUH-CCP). C. difficile isolates collected in the KNUH-CCP were from representative cases of CDIs based on prevalence of CDIs in each hospital, phenotypes of the isolates, and clinical information of the patients during the study periods. Species were identified by 16S ribosomal RNA gene sequence analysis. Two reference strains, ATCC 43255 from the American Type Culture Collection (Manassas, VA, USA) and KCTC 5009 from the Korea Collection for Type Cultures (Osong, Korea), were used. Bacterial culturing, management, and storage were performed based on protocols described by Edwards et al [15]. C. difficile isolates were anaerobically cultured at 37°C on Brain-Heart Infusion (BHI) medium (BD Biosciences, San Jose, CA, USA) supplemented with sodium taurocholate (10% w/v, Sigma-Aldrich, St. Louis, MO, USA), l-cysteine (10% w/v, Sigma-Aldrich), hemin in 1 M sodium hydroxide (10% v/v, Sigma-Aldrich), and vitamin K1 in 95% ethanol (0.02% v/v, Sigma-Aldrich).

Genomic DNA extraction

Genomic DNA was isolated from two or three colonies of C. difficile grown on BHI medium overnight in anaerobic conditions. DNA was extracted using a DNA Extraction Kit (Biofact Co., Korea) according to the manufacturer’s instructions. The extracted DNA was used for MLST, toxinotyping, and PCR analysis of toxin genes.

Amplification of toxin genes

PCR was performed to detect toxin genes of C. difficile. The primers sets used were NK11 and NK9 for the repetitive domain of the toxin A gene (tcdA), NK104 and NK105 for the toxin B gene (tcdB), cdtApos and cdtArev for the binary toxin A gene (cdtA), and cdtBpos and cdtBrev for the binary toxin B gene (cdtB) [1618]. The absence of the pathogenicity locus (PaLoc) was confirmed using the primers Lok1 (5′-AAA ATA TAC TGC ACA TCT GTA TAC-3′) and Lok3 (5′-TTT ACC AGA AAA AGT AGC TTT AA-3′) as described by Rupnik ( PCR products were electrophoresed on 1% agarose gel and bands were visualized using a ChemiDoc MP System (Bio-Rad, Hercules, CA, USA).


PFGE was performed according to previously described protocol with some modifications [19]. Briefly, bacteria were cultured in BHI broth until the OD600 reached 1.8–2.0. The bacterial cells were lysed with 100 μl of 2× lysis buffer (2 M NaCl2, 200 mM EDTA (pH 8.0), 1% Brij-58, 0.4% sodium deoxycholate, 1% sodium lauryl sarcosine, lysozyme 4 mg/ml, RNase 100 mg/ml, and mutanolysin 40 U/ml). Plugs containing the lysed bacterial cells were prepared using 2% PFGE-grade low melting point agarose. The plugs were then incubated in 1× lysis buffer over night at 37°C, followed by incubation with TE buffer (10 mM Tris-HCl, 1 mM EDTA) containing 10 mg/ml proteinase K, for 24 h at 50°C, with buffer changes. The plugs were then washed with TE buffer (10 mM Tris-HCl, 1 mM EDTA) and digested with 50 U of SmaI (Thermo Fisher Scientific, Waltham, MA, USA) for 6 h at 25°C. The digested products were electrophoresed on a 1% PFGE-grade agarose gel using a CHEF Mapper system (Bio-Rad). The band patterns were analyzed with GelCompar II software (Applied Maths, Kortijk, Belgium) to produce a dendrogram. A dendrogram was constructed using the unweighted pair-group method with arithmetic mean clustering. Pulsotypes were arbitrarily classified based on a similarity value of 0.85.

PCR ribotyping

PCR ribotyping of C. difficile isolates was performed as previously described [20]. PCR amplification was performed using the primers 5′-CTG GGG TGA AGT CGT AAC AAG G-3′ and 5′-GCG CCC TTT GTA GCT TGA CC-3′. The PCR conditions comprised an initial denaturation step at 95°C for 3 min, 30 cycles of denaturation at 95°C for 30 S, annealing at 55°C for 40 S, and extension at 72°C for 70 S, and a final elongation step at 72°C for 5 min. Amplification products were electrophoresed on 2% agarose gels. The electrophoretic patterns were compared with those available in a published library [21]. Nomenclature of new ribotypes that did not match with any UK ribotypes was based on the Kim’s ribotyping scheme [22].


The 19-kb PaLoc region comprises toxin genes (tcdA and tcdB) and additional open reading frames (tcdC, tcdD, and tcdE) [23]. Toxinotyping is a restriction fragment length polymorphism typing method used to differentiate C. difficile isolates based on sequence distinctions in the PaLoc region [24]. Toxinotyping was conducted on both toxigenic and non-toxigenic C. difficile isolates as described by Rupnik ( PCR products and their restriction fragments were analyzed on 1% agarose gels. Toxinotypes of C. difficile isolates were determined by comparing their band patterns with that of the standard strain ATCC 43255. Patterns that visually differed from the pattern of this strain were compared with other known toxinotypes.


C. difficile isolates were subjected to MLST to identify allelic polymorphisms at seven loci (adk, atpA, dxr, glyA, recA, sodA, and tpi) as previously described by Griffiths et al. [25]. Isolates were assigned to STs and clades in the C. difficile MLST database ( Clonal relationships among STs, based on allelic polymorphisms, were determined using the EBURST program (eBURST version 3 (, as previously described [26].

Antimicrobial susceptibility testing

The minimum inhibitory concentrations (MICs) of ciprofloxacin (CIP), daptomycin (DAP), linezolid (LNZ), metronidazole (MTZ), moxifloxacin (MXF), rifampicin (RIF), and vancomycin (VAN) were determined by E-test according to the manufacturer’s instructions (bioMérieux, Marcy l'Etoile, France). Bacteroides fragilis ATCC 25285 was used for quality control. Interpretation of antimicrobial susceptibility was based on the guidelines of the Clinical Laboratory Standards Institute (CLSI) [27] and European Committee on Antimicrobial Susceptibility Testing (EUCAST) [28].

Comparison of the typing methods

The Simpson’s index of diversity (SID), which measures the probability that two subjects randomly selected from a sample population belong to dissimilar types, was used as a discriminatory index. The adjusted Wallace coefficient (WC) was used to evaluate the bidirectional agreement between typing methods [29]. The SID and adjusted Wallace coefficient were calculated using a web-based tool (


Isolation of C. difficile

Seventy C. difficile isolates were obtained from patients who diagnosed CDIs. Patients were admitted to the internal medicine (n = 37), emergency room (n = 12), neurosurgery (n = 6), general surgery (n = 3), neurology (n = 4), outpatient (n = 2), pediatric (n = 2), rehabilitation (n = 2), orthopedic surgery (n = 1), and urology (n = 1) wards. Fifty isolates were from patients over 60 years of age, 14 were from patients aged 30–59 years, and six were from patients aged 10–30 years.

Toxin gene profiles

Sixty-eight isolates were A+B+ and two were A-B+ (Fig 1A). The cdt gene encoding the binary toxin was identified in one isolate (CDN7725), and toxin profile of this isolate was A+B+CDT+. Two A-B+-type C. difficile isolates were detected in KNUH.

Fig 1.

Distribution of toxin gene profiles (A), toxinotypes (B), PCR ribotypes (C), and sequence types (D) among 70 C. difficile isolates from two Korean hospitals. In Fig 1C, UD means that the isolate was not assigned to any known ribotype. KNUH, Kyungpook National University Hospital; BPH, Busan Paik Hospital.


Sixty-six isolates exhibited band patterns identical to that of the standard strain ATCC 43255; they were assigned to toxinotype 0 (Fig 1B). Four isolates were found to carry a whole or part of PaLoc and were classified into toxinotypes I, III, VIII, and XII. C. difficile isolates of toxinotypes I, III, and VIII were from the internal medicine ward in KNUH, whereas the single toxinotype XII isolate was from the emergency room in BPH.

PCR ribotypes

Sixty-nine C. difficile isolates were categorized into 18 different PCR ribotypes (Fig 1C). One C. difficile isolate from KNUH could not be assigned to any known ribotype. PCR ribotype 018 was predominant, accounting for 38 isolates (25 from BPH and 13 from KNUH), followed by ribotypes 014/020 (n = 8), 015 (n = 4), 002 (n = 4), and AB9 (n = 2). The remaining 14 C. difficile isolates represented 14 different ribotypes. PCR ribotypes 018, 014/020, 015, and 002 were detected in both hospitals, whereas other ribotypes were unique to each hospital.


The 70 C. difficile isolates were classified into 16 STs (Fig 1D). ST17 was the most prevalent (n = 39), followed by ST8 (n = 6), ST35 (n = 6), ST2 (n = 6), and ST110 (n = 2). The remaining 11 STs were represented by single isolates. Twelve and eight STs were identified in KNUH and BPH, respectively. Sixty-eight A+B+ were classified into 18 ribotypes and 15 STs. Fourteen isolates from emergency room and outpatients belonged to five ribotypes and STs, respectively. With the exception of ribotype AB68, all C. difficile isolates of ribotype 018 were classified into ST17. Based on allelic polymorphisms, STs were allocated to clonal complexes using the eBURST program ( Of the 16 STs, eight STs were placed into three different groups; group 1 [ST2 (founder), ST102 (single locus variant [slv] of ST2 in sodA), and ST110 (slv of ST2 in recA)], group 2 [ST4 (founder), ST17 (slv of ST4 in tpi), and ST53 (slv of ST4 in atpA)], and group 3 [ST99 (slv of ST55 in adk) and ST55]. Group 3 had no founding genotype. The remaining eight STs were singletons.

PFGE analysis

All 70 C. difficile isolates were classified into 33 arbitrary pulsotypes based on a similarity value of 0.85 (Fig 2). Arbitrary pulsotype 1 (CD01) was the most prevalent, accounting for 20 isolates, followed by pulsotype CD05 (n = 6). C. difficile isolates showing two prevalent pulsotypes, CD01 and CD05, belonged to ribotype 018 and ST17. The 39 C. difficile isolates from BPH yielded 16 pulsotypes, with pulsotype CD01 predominant (n = 18), whereas the 31 C. difficile isolates from KNUH produced 20 pulsotypes, with pulsotype CD05, belonging to ribotype 018, the most prevalent (n = 6). The binary toxin gene-positive isolate CDN7725 belonged to ribotype C13, ST47, toxinotype III, and pulsotype CD29.

Fig 2. Dendrogram showing 33 pulsotypes obtained by pulsed-field gel electrophoresis of 70 C. difficile isolates.

Genomic DNA was digested with SmaI. The unweighted pair group method with arithmetic mean clustering method was used with the Dice coefficient with position tolerance and optimization of 1.2%. Pulsotypes were arbitrarily assigned based on a similarity value of 0.85. Undefined, isolates that were not assigned to any known ribotype; AB and C, ribotype nomenclature adopted by Kim’s ribotyping scheme [3, 22].

Antimicrobial resistance

The distribution of MICs of antimicrobial agents against C. difficile isolates is shown in Table 1. Antimicrobial resistance rates of the 70 C. difficile isolates were as follows: MXF, 55.7%; RIF, 27.1%; DAP, 4.3%; and MTZ, 2.9%. No isolates were resistant to VAN. Sixty-seven isolates showed very high CIP MICs (≥ 32 μg/ml). One C. difficile isolate (CDN10470) with ST17 and ribotype 018 was resistant to CIP, DAP, MTZ, MXF, and RIF. Two ST17 isolates were resistant to MTZ. Of the 39 C. difficile isolates resistant to MXF, 32 belonged to ST17. There was no significant difference in antimicrobial resistance rates of C. difficile isolates between hospitals.

Table 1. Antimicrobial susceptibility of 70 C. difficile isolates from two Korean hospitals.

Statistical comparison of the performance of the epidemiological typing tools

SID was used to assess the ability of each epidemiological tool to differentiate between two test samples picked randomly from the sample population. Of the typing methods used in this study, PFGE, which yielded 33 arbitrary pulsotypes, exhibited the highest SID of 0.907, whereas toxinotyping, which defined five different types, showed the lowest SID of 0.112 (Table 2). The WC was used to analyze the degree of concordance between typing schemes (Table 3). Ribotyping’s congruence with MLST was greatest, with a WC of 0.976 (95% CI, 0.957–0.995), and PFGE showed agreement with MLST, with WC of 0.961 (95% CI, 0.938–0.985). However, low congruence was observed between toxinotyping and PFGE (WC = 0.010; 95% CI, 0.000–0.079) and toxinotyping and ribotyping (WC = 0.031; 95% CI, 0.000–0.211).

Table 2. Comparison of four different epidemiological typing methods using Simpson's index of diversity.

Table 3. Concordance values of four different typing methods estimated using the adjusted Wallace coefficient.


This study reports, for the first time, the distribution of STs among C. difficile isolates from Korean hospitals. In addition, toxin gene profiling, toxinotyping, PFGE, and ribotyping were performed to explore the epidemiological characteristics of C. difficile isolates from two Korean hospitals. Not only were international epidemic C. difficile clones, including ribotypes 018 (ST17), 014/020 (ST2/ST110), and 015 (ST26/ST35) found in both study hospitals, but diverse sporadic clones unique to each hospital were also detected.

Toxigenic C. difficile strains secrete one or both TcdA and TcdB toxins. A-B+ C. difficile variants cannot secrete TcdA owing to deletions in the repetitive sequence of the tcdA gene. This toxin variant strain was first described in the early 1990s [30, 31]. A-B+ variants were found to constitute less than 10% of the toxigenic isolates from European countries and Japan, but were detected in up to 40% of isolates in Korea and Thailand [3, 1012]. In this study, two A-B+ C. difficile isolates were identified only in KNUH. Earlier studies demonstrated a high correlation between toxin types and STs; A-B+ variants from China and Thailand were found to belong to ST37 and ST45 (ribotype 017), respectively. Moreover, A-B+ variants from Korea belonged to ribotype 017 [22]. However, two A-B+ variants tested in this study represented ribotypes 014/020 (ST110) and 017 (ST37), respectively.

PCR ribotyping of C. difficile isolates revealed that ribotype 018 (54.3%) was the most prevalent in both hospitals. Our results were consistent with those of a previous study conducted in a single hospital in Korea, which also reported a high prevalence of ribotype 018 [14]. Another report showed that a ribotype from Japan, smz, which corresponds to ribotype 018, was a major clone occurring in both outbreak and non-outbreak CDIs [32]. Ribotype 018 is reported to be one of the most prevalent strains causing complicated infections in France and Italy [3335]. The prevalence and clinical relevance of ribotype 018 can be attributed to the fact that strains belonging to this genotype are relatively resistant to fluoroquinolones and have a unique variant of the splA gene, which is speculated to confer a high degree of adhesiveness that contributes to the success of this genotype [36]. In this study, the resistance rate of ribotype 018 C. difficile isolates to MXF was 84.2% (32 of 38 isolates). Moreover, all C. difficile isolates of ribotype 018, except for one isolate, exhibited the highest CIP MICs (> 32 μg/ml). Ribotypes 017, 014/020, 015, 002, 001, and 012 are increasingly becoming potential epidemic strains. These ribotypes were prevalent among C. difficile isolates obtained over the past decade in Asian as well as European countries [5, 14, 32, 37], suggesting that they are globally circulating strains. Ribotypes 027, 014/020, UM11, 053/163, 002, 001, 078–126, 017, and 014 were the chief agents in outbreaks and epidemics in North American countries [3841]. In an Australian survey of the hospital-acquired and community-acquired CDIs, ribotypes 014/020, 002, 054, 056, and 070 were the most prevalent in 2010, and C. difficile isolates of ribotypes 014/020, 056, 002, and 018 were the prevalent between 2012 and 2014 [42, 43]. Regarding Asian countries, the most common ribotypes in China have been 017, 012, and 046 [44], whereas ribotypes 018, 001, 017, and 014/020 are prevalent in Japan and Korea [11, 14]. We also found that most of the ribotypes were unique to a given hospital, with four ribotypes (AB7, AB15, AB32, and AB35) found in KNUH and seven ribotypes (AB9, AB18, AB24, AB25, AB42, AB46, and AB68) found in BPH. Fourteen C. difficile isolates from community-acquired CDIs, two from outpatients and 12 from emergency room, were analyzed. Two isolates from outpatients were A+B+ and ribotype 018. Twelve A+B+ isolates from emergency room showed five different ribotypes, 018 (n = 6), 014/020 (n = 3), AB15 (n = 1), AB35 (n = 1), and AB46 (n = 1). These results suggest that there is no difference in major clones of C. difficile between community-acquired and hospital-acquired CDIs. Our ribotyping results suggest that not only are epidemic C. difficile clones such as 018, 014/020, and 015 circulating in the study hospitals, but also that many sporadic clones are being introduced.

MLST of the 70 C. difficile isolates generated 16 different STs. We determined that ST17 (ribotype 018) was predominant, accounting for 38 isolates. One isolate of ribotype AB68 and pulsotype CD33 was found to belong to ST17. The other frequent types were ST2 (n = 6), ST8 (n = 6), and ST35 (n = 6). Eleven STs were represented by single isolates. MLST yielded a diverse pool of allelic polymorphisms, ranging from 4–10 per housekeeping gene locus. The adk gene, with four distinctive alleles, was the most conserved among the seven loci, whereas the recA gene had the highest diversity, exhibiting a total of 10 alleles. There was a high correlation between STs and ribotypes. For example, ST2, ST17, ST8, and ST35 correlated with the ribotypes 014/020 (100%), 018 (97.4%), 002 (66.7%), and 015 (66.7%), respectively. On the other hand, ST2 and ST110, which differed in the recA gene, belonged to the same ribotype, 014/020.

PFGE analysis of C. difficile isolates showed the highest discriminatory power of the epidemiological tools tested, with an SID of 0.907. Of the 33 arbitrary pulsotypes, CD01 (n = 20) and CD05 (n = 6) were the most prevalent; these pulsotypes correlated with ribotype 018 and ST17. Interestingly, 18 of the 20 CD01 isolates and all six CD05 isolates were from BPH and KNUH, respectively. ST2 was highly correlated with respective pulsotypes. However, the remaining 13 ST17 isolates were divided into 10 arbitrary pulsotypes. Moreover, ST8 and ST35 isolates showed diverse PFGE patterns and ribotypes. These results suggest that PFGE analysis is more valuable for intra-hospital epidemiological studies than other typing methods.

A statistical comparison of the different typing methods using the WC revealed that PCR ribotyping and PFGE results were highly predictive of MLST results; in other words, most isolates belonging to one ribotype or pulsotype grouped together in a single ST. However, ribotyping and MLST did not always yield similar outcomes. Occasionally, multiple STs were observed in a given ribotype [25, 45]. Conversely, multiple ribotypes clustered in a single ST. This observation suggests that no single epidemiological typing scheme can perfectly predict the outcome of the other typing methods. PFGE analysis yielded more groupings than any of the other three methods tested. However, a low correlation was found between the results of PFGE and these other methods. These results suggest that a combination of band-based typing (PFGE or PCR ribotyping) and sequence-based typing (MLST) could be a powerful tool for epidemiological study of C. difficile isolates.

This study has several limitations. First, we tested limited number of isolates from two hospitals. C. difficile isolates collected in our culture collection system for pathogens were from representative cases of CDIs, but it was possible that clonal diversity of C. difficile isolates is underestimated. However, in this study, 70 C. difficile isolates from two hospitals were classified into 19 ribotypes, 16 STs, and 33 arbitrary pulsotypes. Second, we did not analyze incidence of CDIs and presence of outbreaks in the study hospitals and clinical characteristics of the patients with CDIs regarding risk factors, severity of CDIs, and association with antibiotic exposure. These clinical parameters may influence the clonal distribution of C. difficile in the hospitals. However, despite these limitations, our study provides important epidemiological information of C. difficile isolates from two Korean hospitals concerning the persistence of specific international clones and the emergence of sporadic clones.

This study found that C. difficile strains of ST17/ ribotype 018 were the most prevalent in two Korean hospitals. No hypervirulent 027 strain was detected in this study. Interestingly, C. difficile strains of the epidemic ST17/ribotype 018 exhibited different pulsotypes in each hospital, with arbitrary pulsotype CD01 in BPH and CD05 in KNUH. With the persistence of epidemic clones and the emergence of sporadic clones, CDIs are becoming one of the important nosocomial infections in Korean hospitals. A combination of MLST with PFGE or PCR ribotyping could be useful for monitoring epidemic C. difficile strains and the emergence of new clones in hospitals.

Author Contributions

  1. Conceptualization: AN YKK WKL JCL.
  2. Investigation: GNS SHN HIK YJK.
  3. Methodology: YKK WKL.
  4. Resources: HSL KES JHS.
  5. Writing – original draft: AN YKK.
  6. Writing – review & editing: WKL JCL.


  1. 1. Gursoy S, Guven K, Arikan T, Yurci A, Torun E, Baskol M, et al. (2007) Clostridium difficile infection frequency in patients with nosocomial infections or using antibiotics. Hepatogastroenterology 54: 1720–1724. pmid:18019703
  2. 2. Gupta A, Khanna S. (2014) Community-acquired Clostridium difficile infection: an increasing public health threat. Infect Drug Resist 7: 63–72. pmid:24669194
  3. 3. Kim H, Jeong SH, Roh KH, Hong SG, Kim JW, Shin MG, et al. (2010) Investigation of toxin gene diversity, molecular epidemiology, and antimicrobial resistance of Clostridium difficile isolated from 12 hospitals in South Korea. Korean J Lab Med 30: 491–497. pmid:20890081
  4. 4. Collins DA, Hawkey PM, Riley TV. (2013) Epidemiology of Clostridium difficile infection in Asia. Antimicrob Resist Infect Control 2: 21. pmid:23816346
  5. 5. Bauer MP, Notermans DW, van Benthem BH, Brazier JS, Wilcox MH, Rupnik M, et al. (2011) Clostridium difficile infection in Europe: a hospital-based survey. Lancet 377: 63–73. pmid:21084111
  6. 6. Huang H, Fang H, Weintraub A, Nord CE. (2009) Distinct ribotypes and rates of antimicrobial drug resistance in Clostridium difficile from Shanghai and Stockholm. Clin Microbiol Infect 15: 1170–1173. pmid:19624517
  7. 7. Gerding DN, Johnson S, Rupnik M, Aktories K. (2014) Clostridium difficile binary toxin CDT: mechanism, epidemiology, and potential clinical importance. Gut Microbes 5: 15–27. pmid:24253566
  8. 8. O'Connor JR, Johnson S, Gerding DN. (2009) Clostridium difficile infection caused by the epidemic BI/NAP1/027 strain. Gastroenterology 136: 1913–1924. pmid:19457419
  9. 9. Goorhuis A, Bakker D, Corver J, Debast SB, Harmanus C, Notermans DW, et al. (2008) Emergence of Clostridium difficile infection due to a new hypervirulent strain, polymerase chain reaction ribotype 078. Clin Infect Dis 47: 1162–1170. pmid:18808358
  10. 10. Ngamskulrungroj P, Sanmee S, Putsathit P, Piewngam P, Elliott B, Riley TV, et al. (2015) Molecular epidemiology of Clostridium difficile infection in a large teaching hospital in Thailand. PLoS One 10: e0127026. pmid:26000789
  11. 11. Kuwata Y, Tanimoto S, Sawabe E, Shima M, Takahashi Y, Ushizawa H, et al. (2015) Molecular epidemiology and antimicrobial susceptibility of Clostridium difficile isolated from a university teaching hospital in Japan. Eur J Clin Microbiol Infect Dis 34: 763–772. pmid:25471195
  12. 12. Lee JH, Lee Y, Lee K, Riley TV, Kim H. (2014) The changes of PCR ribotype and antimicrobial resistance of Clostridium difficile in a tertiary care hospital over 10 years. J Med Microbiol 63: 819–823. pmid:24696516
  13. 13. Kim H, Lee Y, Moon HW, Lim CS, Lee K, Chong Y. (2011) Emergence of Clostridium difficile ribotype 027 in Korea. Korean J Lab Med 31: 191–196. pmid:21779194
  14. 14. Kim J, Kang JO, Kim H, Seo MR, Choi TY, Pai H, et al. (2013) Epidemiology of Clostridium difficile infections in a tertiary-care hospital in Korea. Clin Microbiol Infect 19: 521–527. pmid:22712697
  15. 15. Edwards AN, Suarez JM, McBride SM. (2013) Culturing and maintaining Clostridium difficile in an anaerobic environment. J Vis Exp 79: e50787.
  16. 16. Kato N, Ou CY, Kato H, Bartley SL, Brown VK, Dowell VR Jr, et al. (1991) Identification of toxigenic Clostridium difficile by the polymerase chain reaction. J Clin Microbiol 29: 33–37. pmid:1993763
  17. 17. Kato H, Kato N, Watanabe K, Iwai N, Nakamura H, Yamamoto T, et al. (1998) Identification of toxin A-negative, toxin B-positive Clostridium difficile by PCR. J Clin Microbiol 36: 2178–2182. pmid:9665986
  18. 18. Stubbs S, Rupnik M, Gibert M, Brazier J, Duerden B, Popoff M. (2000) Production of actin-specific ADP-ribosyltransferase (binary toxin) by strains of Clostridium difficile. FEMS Microbiol Lett 186: 307–312. pmid:10802189
  19. 19. Gal M, Northey G, Brazier JS. (2005) A modified pulsed-field gel electrophoresis (PFGE) protocol for subtyping previously non-PFGE typeable isolates of Clostridium difficile polymerase chain reaction ribotype 001. J Hosp Infect 61: 231–236. pmid:16002184
  20. 20. O'Neill . (1996) Modification of a PCR ribotyping method for application as a routine typing scheme for Clostridium difficile. Anaerobe 2: 205–209.
  21. 21. Kuijper EJ, van den Berg RJ, Brazier JS. (2009) Comparison of molecular typing methods applied to Clostridium difficile. Methods Mol Biol 551: 159–171. pmid:19521874
  22. 22. Han SH, Kim H, Lee K, Jeong SJ, Park KH, Song JY, et al. (2014) Epidemiology and clinical features of toxigenic culture-confirmed hospital-onset Clostridium difficile infection: a multicentre prospective study in tertiary hospitals of South Korea. J Med Microbiol 63: 1542–1551.32. pmid:25187603
  23. 23. Braun V, Hundsberger T, Leukel P, Sauerborn M, von Eichel-Streiber C. (1996) Definition of the single integration site of the pathogenicity locus in Clostridium difficile. Gene 181: 29–38. pmid:8973304
  24. 24. Rupnik M. (1998) A novel toxinotyping scheme and correlation of toxinotypes with serogroups of Clostridium difficile isolates. J Clin Microbiol 36: 2240–2247. pmid:9665999
  25. 25. Griffiths D, Fawley W, Kachrimanidou M, Bowden R, Crook DW, Fung R, et al. (2010) Multilocus sequence typing of Clostridium difficile. J Clin Microbiol 48: 770–778. pmid:20042623
  26. 26. Feil EJ. (2003) eBURST: Inferring patterns of evolutionary descent among clusters of related bacterial genotypes from multilocus sequence typing data. J Bacteriol 186: 1518–1530.
  27. 27. Clinical and Laboratory Standards Institute. (2015) Performance standards for antimicrobial susceptibility testing: 23rd informational supplement. CLSI document M100-S25. CLSI, Wayne, PA.
  28. 28. EUCAST (2016) Breakpoint tables for interpretation of MICs and zone diameters (Version 6.0). The European Committee on Antimicrobial Susceptibility Testing.
  29. 29. Hunter PR, Gaston MA. (1998) Numerical index of the discriminatory ability of typing systems: an application of Simpson's index of diversity. J Clin Microbiol 26: 2465–2466.
  30. 30. Borriello SP, Wren BW, Hyde S, Seddon SV, Sibbons P, Krishna MM, et al. (1992) Molecular, immunological, and biological characterization of a toxin A-negative, toxin B-positive strain of Clostridium difficile. Infect Immun 60: 4192–4199. pmid:1398930
  31. 31. Lyerly DM, Barroso LA, Wilkins TD, Depitre C, Corthier G. (1992) Characterization of a toxin A-negative, toxin B-positive strain of Clostridium difficile. Infect Immun 60: 4633–4639. pmid:1398977
  32. 32. Senoh M, Kato H, Fukuda T, Niikawa A, Hori Y, Hagiya H, et al. (2015) Predominance of PCR-ribotypes, 018 (smz) and 369 (trf) of Clostridium difficile in Japan: a potential relationship with other global circulating strains? J Med Microbiol 64: 1226–1236. pmid:26238868
  33. 33. Poilane I, Fantinato C, Cruaud P, Collignon A. (2008) Epidemiological study of Clostridium difficile strains isolated in Jean-Verdier-Rene-Muret hospitals from 2001 to 2007. Pathol Biol 56: 412–416. pmid:18842360
  34. 34. Spigaglia P, Barbanti F, Dionisi AM, Mastrantonio P. (2010) Clostridium difficile isolates resistant to fluoroquinolones in Italy: emergence of PCR ribotype 018. J Clin Microbiol 48: 2892–2896. pmid:20554809
  35. 35. Baldan R, Trovato A, Bianchini V, Biancardi A, Cichero P, Mazzotti M, et al. (2015) Clostridium difficile PCR ribotype 018, a successful epidemic genotype. J Clin Microbiol 53: 2575–2580. pmid:26041894
  36. 36. Spigaglia P. (2012) Characterisation of Clostridium difficile 018: an epidemic PCR-ribotype recently emerged in Italy. 22nd European Congress of Clinical Microbiology and Infectious Diseases (ECCMID).
  37. 37. Barbut F, Mastrantonio P, Delmee M, Brazier J, Kuijper E, Poxton I, et al. (2007) Prospective study of Clostridium difficile infections in Europe with phenotypic and genotypic characterisation of the isolates. Clin Microbiol Infect 13: 1048–1057. pmid:17850341
  38. 38. Waslawski S, Lo ES, Ewing SA, Young VB, Aronoff DM, Sharp SE, et al. (2013) Clostridium difficile ribotype diversity at six health care institutions in the United States. J Clin Microbiol 51: 1938–1941. pmid:23554188
  39. 39. Rodriguez-Palacios A, Stampfli HR, Duffield T, Peregrine AS, Trotz-Williams LA, Arroyo LG, et al. (2006) Clostridium difficile PCR ribotypes in calves, Canada. Emerg Infect Dis 12: 1730–1736. pmid:17283624
  40. 40. Labbe AC, Poirier L, Maccannell D, Louie T, Savoie M, Beliveau C, et al. (2008) Clostridium difficile infections in a Canadian tertiary care hospital before and during a regional epidemic associated with the BI/NAP1/027 strain. Antimicrob Agents Chemother 52: 3180–3187. pmid:18573937
  41. 41. Tickler IA, Goering RV, Whitmore JD, Lynn AN, Persing DH, Tenover FC. (2014) Strain types and antimicrobial resistance patterns of Clostridium difficile isolates from the United States, 2011 to 2013. Antimicrob Agents Chemother 58: 4214–4218. pmid:24752264
  42. 42. Cheng AC, Collins DA, Elliott B, Ferguson JK, Thean S, Paterson DL, et al. (2016) Laboratory-based surveillance of Clostridium difficile circulating in Australia, September-November 2010. Pathology 48: 257–260. pmid:27020502
  43. 43. Furuya-Kanamori L, Riley TV, Paterson DL, Foster NF, Huber CA, Hong S, et al. (2017) Comparison of Clostridium difficile ribotypes circulating in Australian hospitals and communities. J Clin Microbiol 55: 216–225. pmid:27807147
  44. 44. Hawkey PM, Marriott C, Liu WE, Jian ZJ, Gao Q, Ling TK, et al. (2013) Molecular epidemiology of Clostridium difficile infection in a major chinese hospital: an underrecognized problem in Asia? J Clin Microbiol 51: 3308–3313. pmid:23903542
  45. 45. Dingle KE, Griffiths D, Didelot X, Evans J, Vaughan A, Kachrimanidou M, et al. (2011) Clinical Clostridium difficile: clonality and pathogenicity locus diversity. PLoS One 6: e19993. pmid:21625511