Antimicrobial Susceptibilities and Molecular Epidemiology of Clinical Isolates of Clostridium difficile in Taiwan

Abstract

The antimicrobial susceptibility and virulence factors of Clostridium difficile clinical isolates in Taiwan have not previously been reported. One hundred and thirteen isolates were collected from two major teaching hospitals in Taiwan from 2001 to 2009. Molecular typing was performed by an automated repetitive extragenic palindromic sequence-based PCR (rep-PCR) method (DiversiLab; Bacterial Barcodes, Inc., Athens, GA) and PCR ribotyping. Detection of tcdA, tcdB, cdtA, and cdtB genes was performed using a multiplex PCR assay, and gyrA and gyrB genes of moxifloxacin-nonsusceptible isolates were sequenced. All isolates were susceptible to vancomycin and metronidazole. Ninety-five (84%) isolates were susceptible to moxifloxacin, and the MIC₉₀ for nemonoxacin was 4 μg/ml. Tigecycline showed favorable antibacterial activity (MIC₉₀ of 0.06 μg/ml). Thirteen rep-PCR types were identified as a predominant rep-PCR type (type A; non-North American pulsed-field gel electrophoresis type 1 [NAP1], -NAP7, or -NAP8) accounting for 52.2% (59 isolates). Nine of 18 moxifloxacin-nonsusceptible isolates belonged to the rep-PCR type A. The rep-PCR type A and C isolates were distinct from NAP1 (ribotype 027) and NAP8 (ribotype 078) as determined by PCR ribotyping. Seventy-four (65%) isolates harbored tcdA and tcdB, and 15 (13%) harbored cdtAB encoding binary toxin. Eleven isolates had a gene deletion in tcdC, including a 39-bp deletion (9 isolates) and an 18-bp deletion (2). In conclusion, dissemination of a predominant C. difficile clone in southern and northern Taiwan was noted. However, no NAP1 (ribotype 027) isolate could be discovered in this study.

Clostridium difficile-associated diarrhea is a global health problem, and a highly virulent strain (restriction endonuclease analysis group type BI/North American pulsed-field gel electrophoresis [PFGE] type 1/PCR ribotype 027 [BI/NAP1/027]) has disseminated throughout North America and Europe (15, 25). Cases of fluoroquinolone-resistant C. difficile PCR ribotype 027 have also been reported in three Asia-Pacific countries (5). Exposure to nearly all classes of antimicrobials has been associated with C. difficile infection (CDI) (26). Some antimicrobial-resistant C. difficile strains have contributed to a higher virulence and transmission possibility (5, 11, 20, 22). Clindamycin- and fluoroquinolone-resistant isolates of C. difficile have been associated with several large outbreaks (6, 22, 28). A predominant BI/NAP1/027 strain of C. difficile with fluoroquinolone resistance was associated with high morbidity and mortality rates of CDI (20). All virulent strains of C. difficile carry a pathogenicity locus (PaLoc), which contains toxin genes. Two major C. difficile toxins are enterotoxin (TcdA) and cytotoxin (TcdB), encoded by tcdA and tcdB, respectively. The tcdC gene is a negative regulator of these two toxins. Although Warny et al. demonstrated that deletions in the tcdC gene (a putative negative regulator for toxins A and B) were associated with the increased production of toxins A and B (34), subsequent studies could not confirm this finding (5, 15). The role of the binary toxin encoded by cdtA and cdtB remains unclear (5, 15, 25).

Three characteristics of epidemic BI/NAP1/027 strains include the presence of fluoroquinolone resistance, binary toxin, and increased production of toxins A and B and a deletion on the tcdC gene (20). BI/NAP1/027 strains exhibit high-level resistance to gatifloxacin and moxifloxacin (29).

The incidence of CDI in Taiwan has recently been reported as 42.6 cases per 100,000 patient days, or 3.4 cases per 1,000 discharges, and the incidence was higher in intensive care units (4, 14). The antimicrobial susceptibility and microbiological typing of C. difficile isolates were rarely reported in Taiwan. Thus, the purposes of this study in Taiwan were to analyze the profiles of C. difficile susceptibility to 16 antimicrobial agents, including nemonoxacin and tigecycline, and to investigate the molecular epidemiology of these isolates, particularly those with resistance to fluoroquinolones.

MATERIALS AND METHODS

Bacterial isolates.

A total of 113 isolates of C. difficile, including 83 isolates collected from National Taiwan University Hospital (NTUH) and 25 from National Cheng Kung University Hospital (NCKUH) from August 2007 to June 2009, were obtained for analysis. Another five isolates identified from 2001 to 2004 from patients at NTUH were also included in the analysis. These isolates were recovered from stool specimens of patients with unexplained fever or concurrent gastrointestinal symptoms, such as diarrhea, abdominal discomfort, or ileus. Not all these patients underwent C. difficile toxin tests of their stool specimens.

Antimicrobial susceptibility testing.

The MICs of 15 antimicrobial agents, including penicillin, ampicillin-sulbactam, cefmetazole, ertapenem, meropenem, imipenem, doripenem, clindamycin, vancomycin, metronidazole, fusidic acid, moxifloxacin, gemifloxacin, tigecycline, and nemonoxacin, were determined using the agar dilution method recommended by the Clinical and Laboratory Standards Institute (CLSI) (7). An inoculum of 10⁵ CFU bacteria was applied to each plate with a Steers replicator on supplemented Brucella blood agar (BBL Microbiology Systems, Cockeysville, MD) (7). The plates were incubated in an anaerobic chamber for 48 h at 35°C. For testing daptomycin, the Brucella agar contained calcium (50 μg/ml), as recommended previously (10). The antimicrobial agents used for susceptibility testing were obtained from their corresponding manufacturers.

The MIC was defined as the lowest concentration of each antimicrobial agent that inhibited the growth of the tested isolate. Bacteroides fragilis ATCC 25285, Bacteroides thetaiotaomicron ATCC 29741, Eubacterium lentum ATCC 43055, and C. difficile ATCC 700057 were used for quality control for each run of susceptibility testing.

The MIC interpretive breakpoints for some antimicrobial agents tested followed the guidelines recommended by the CLSI (Table 1) (7) but were not established by the CLSI for vancomycin, doripenem, fusidic acid, daptomycin, tigecycline, and nemonoxacin for C. difficile at the time of this study.

TABLE 1.

Antimicrobial susceptibilities of 113 Clostridium difficile isolates to 16 antimicrobial agents^b

Agent	MIC (μg/ml)			No. (%) of isolates			MIC interpretive breakpointsa (S/I/R)
	Range	50%	90%	S	I	R
Penicillin	0.12-16	2	4	1 (1)	18 (16)	94 (83)	≤0.5/1/≥2
Ampicillin-sulbactam	0.25-8	2	4	113 (100)	0 (0)	0 (0)	≤8/16/≥32
Cefmetazole	0.06-128	16	64	61 (54)	39 (35)	13 (12)	≤16/32/≥64
Ertapenem	0.03-32	8	16	15 (13)	71 (63)	27 (24)	≤4/8/≥16
Imipenem	2-32	8	16	18 (16)	51 (45)	44 (39)	≤4/8/≥16
Meropenem	0.03-8	2	4	104 (92)	9 (8)	0 (0)	≤4/8/≥16
Doripenem	0.25-8	4	4	NA	NA	NA	NA
Clindamycin	0.06->256	4	>256	37 (33)	24 (21)	52 (46)	≤2/4/≥8
Vancomycin	0.25-4	0.5	1	NA	NA	NA	NA
Metronidazole	0.03-4	0.5	1	113 (100)	0 (0)	0 (0)	≤8/16/≥32
Fusidic acid	0.5-32	1	2	NA	NA	NA	NA
Daptomycin	0.12-2	0.5	1	NA	NA	NA	NA
Moxifloxacin	0.06-16	1	16	94 (83)	1 (1)	18 (16)	≤2/4/≥8
Gemifloxacin	0.25->32	2	32	NA	NA	NA	NA
Nemonoxacin	0.25-16	0.5	4	NA	NA	NA	NA
Tigecycline	0.03-0.25	0.06	0.06	NA	NA	NA	NA

^aMIC breakpoints applied were those recommended for anaerobes by the Clinical and Laboratory Standards Institute (CLSI) (4).

^bNA, MIC interpretive breakpoints were not available from the CLSI (4); S, susceptible; I, intermediate; R, resistant.

Sequencing analysis of gyrA and gyrB genes for moxifloxacin-nonsusceptible isolates.

Partial sequencing of the gyrA and gyrB genes was performed for all isolates which were not susceptible to moxifloxacin (MICs of ≥4 μg/ml). The gyrA and gyrB genes of the isolates were amplified using the primer couple gyrA1 (5′-AAT GAG TGT TAT AGC TGG ACG-3′) and gyrA2 (5′-TCT TTT AAC GAC TCA TCA AAG TT-3′), amplifying 390 bp of the gyrA gene, and the primer couple gyrB1 (5′-AGT TGA TGA ACT GGG GTC TT-3′) and gyrB2 (5′-TCA AAA TCT TCT CCA ATA CCA-3′), amplifying 390 bp of the gyrB gene, as described previously (8). PCR amplification consisted of 30 cycles of denaturation at 94°C for 30 s, annealing at 54°C (gyrB)/58°C (gyrA) for 30 s, and extension at 72°C for 30 s. PCR products were purified using the Illustra GFX PCR DNA and gel band purification kit (GE Healthcare) and sequenced by the use of the BigDye Terminator version 3.1 cycle sequencing kit (Applied Biosystems) and an Applied Biosystems 3730 DNA analyzer.

Detection of tcdA, tcdB, cdtA, and cdtB.

A multiplex PCR assay was used for the detection of tcdA, tcdB, cdtA and cdtB, with 16S rRNA genes used as an internal PCR control (31). The PCRs were done with reaction mixture volumes of 25 μl containing PCR buffer [50 mM Tris-HCl, 10 mM KCl, 5 mM (NH4)₂SO₄, pH 8.3], 2.6 mM MgCl₂, 260 μM (each) dATP, dCTP, dGTP, and dTTP, and 1.25 U of FastStart Taq polymerase (Takara Bio Inc.), and the primers were used at concentrations described previously (31). Thermocycler conditions used were 10 min at 94°C, followed by 35 cycles of 50 s at 94°C, 40 s at 54°C and 50 s at 72°C, and a final extension of 3 min at 72°C. Isolates with binary toxin genes (cdtA and cdtB) were subjected to tcdC gene sequencing (31).

Determination of strain genetic relatedness.

An automated repetitive extragenic palindromic sequence-based PCR (rep-PCR) typing method (DiversiLab; Bacterial Barcodes, Inc., Athens, GA) for all 113 isolates was used to determine their DNA fingerprinting in accordance with the manufacturer's instructions (27). Similarity of the isolates (rep-PCR types) was determined and interpreted as previously described (19, 27). PCR ribotypes for isolates of the main rep-PCR types were determined in accordance with previous descriptions (27, 33).

RESULTS

Antimicrobial susceptibilities.

The MICs of the antimicrobial agents tested (except agents for which MIC ranges were not provided by the CLSI) for isolates of the four control strains for each run were all within the recommended ranges (7). The MIC ranges, MIC₅₀s, MIC₉₀s, and percentages of the susceptibility of 113 C. difficile isolates to 14 antimicrobial agents are summarized in Table 1.

All isolates were susceptible to ampicillin-sulbactam and metronidazole. More than 90% of isolates were inhibited by vancomycin and daptomycin at 1 μg/ml. Ninety-five (84%) isolates tested were susceptible to moxifloxacin (MIC of ≤2 μg/ml). The MIC₉₀s for gemifloxacin and nemonoxacin were 32 and 4 μg/ml, respectively. Ninety-two percent of isolates were susceptible to meropenem; however, the rates of susceptibility to ertapenem and imipenem were 13% and 16%, respectively. Doripenem exhibited in vitro activity similar to that of meropenem (both with MIC₉₀s of 4 μg/ml). Tigecycline showed excellent antibacterial activity, with a MIC₉₀ of 0.06 μg/ml.

Sequencing of gyrA and gyrB genes for moxifloxacin-nonsusceptible isolates.

Partial sequencing of gyrA and gyrB genes for 18 moxifloxacin-nonsusceptible isolates revealed amino acid changes from Thr82 to Ile (12 isolates), Asp81 to Asn (one isolate), and Asp71 to Gly (one isolate) in GyrA and a change from Asp426 to Val (one isolate) in GyrB (Table 2). Three isolates had changes in both GyrA and GyrB: Thr82 to Ile in GyrA plus Ser416 to Ala in GyrB (2 isolates) and Thr82 to Ile in GyrA plus Asp426 to Val in GyrB (1 isolate).

TABLE 2.

Genotypic characteristics of 18 moxifloxacin-nonsusceptible Clostridium difficile isolates (MICs of ≥4 μg/ml)

No. of isolates	Substitution in:		Presence of:			MIC (μg/ml)
	GyrA	GyrB	tcdA	tcdB	cdtAB	Moxifloxacin	Gemifloxacin	Nemonoxacin
4	Thr82 to Ile		+	+	−	16	≥32	8
3	Thr82 to Ile		−	−	−	16	≥32	8
2	Thr82 to Ile		+	+	−	16	32	4
2	Thr82 to Ile		−	−	−	16	32	4
1	Thr82 to Ile		−	−	−	16	2	4
1	Thr82 to Ile	Ser416 to Ala	+	+	+	16	≥32	4
1	Thr82 to Ile	Ser416 to Ala	+	+	+	8	32	4
1	Thr82 to Ile	Asp426 to Val	+	+	−	8	4	8
1		Asp426 to Val	+	+	−	8	8	16
1	Asp71 to Gly		+	+	−	4	2	4
1	Asp81 to Asn		−	−	−	8	8	4

tcdA, tcdB, cdtA, and cdtB.

Seventy-four (65%) isolates were tcdA and tcdB positive, and of them, 15 curried binary toxin genes (cdtAB). Among the latter, 11 isolates had a gene deletion in tcdC, a 39-bp (9 isolates) or 18-bp (2 isolates) deletion. Only those isolates with a 39-bp deletion had a premature stop codon (C→T transition) at position 184. Two isolates with a 39-bp deletion had a Thr 82-Ile change in GyrA and Ser 416-Ala in GyrB. Other isolates with gene deletions in tcdC did not show amino acid substitutions in GyrA or GyrB (Table 3).

TABLE 3.

Genotypic characteristics of 11 Clostridium difficile isolates with gene deletions in tcdC

No. of isolates	Presence of:			tcdC deletion (bp)	tcdC stop codon at position 184 (C184T)	Substitution in:
	tcdA	tcdB	cdtAB			GyrA	GyrB
7	+	+	+	39	+
2	+	+	+	39	+	Thr82 to Ile	Ser416 to Ala
2	+	+	+	18	−

Rep-PCR typing and PCR ribotyping.

Among the study isolates, 13 different rep-PCR types, types A to M, were identified. A major rep-PCR type, type A, comprised 52.2% (59 isolates) of all isolates. The rep-PCR type A was distributed in both centers and could be traced back to 2001. Rep-PCR type C isolates (n = 14) were first identified in 2001 and detected only at NTUH. Of note, of 18 moxifloxacin-nonsusceptible isolates, half belonged to the rep-PCR type A. All isolates, including rep-PCR type A and C isolates, were distinct from NAP1/027 and NAP8 (ribotype 078) as determined by PCR ribotyping, indicating that the hypervirulent ribotype BI/NAP1/027 strain was not present in this study. The rep-PCR type C isolates belonged to NAP7. Other isolates were different from NAP1 to NAP6 and NAP8 compared with the results of rep-PCR with known ribotypes (NAP1 to NAP8) (27).

DISCUSSION

The prevalence of binary toxin genes in this study was lower than that of Danish strains (23%) but was higher than those in other studies (31). Some PCR ribotypes, such as type 027 or 078, had these virulence factors. Both of these ribotypes had tcdA and tcdB binary toxin genes and a mutation in the tcdC gene. However, their positions of mutation were different. Type 027 had an 18-bp deletion and a deletion at position 117 in the tcdC gene. Type 078 had a 39-bp deletion and a point mutation at position 184 (C184T). Recent isolates of these ribotypes in severe outbreaks were found to be resistant to fluoroquinolones (11). Except for two isolates with gene deletions in tcdC, none of the isolates in this study exhibited fluoroquinolone resistance. Only two isolates with a 39-bp deletion had amino acid substitutions in GyrA and GyrB. The MIC results for quinolones for these two isolates were different for gemifloxacin (32 μg/ml versus >32 μg/ml) and moxifloxacin (8 μg/ml versus 16 μg/ml) but the same for nemonoxacin (4 μg/ml).

Fluoroquinolone-resistant C. difficile strains have been found to have different amino acid substitutions in GyrA and GyrB. The most common substitution is the amino acid change from Thr82 to Ile in GyrA. Changes from Arg447 to Lys or Leu and Asp426 to Asn or Val have been found in GyrB (32). Moxifloxacin resistance was the most frequently identified quinolone resistance in this study, and most of these isolates had the change from Thr82 to Ile in GyrA (88%). However, none of these isolates had an 18-bp deletion in tcdC. Among those isolates with quinolone resistance, gemifloxacin showed the highest MICs (72% of the isolates with a MIC of ≥32 μg/ml) and was followed by moxifloxacin. The change from Asp426 to Val in GyrB has previously been reported in an outbreak of toxin A-negative/toxin B-positive C. difficile strains (9). In this study, two isolates with Asp426-to-Val changes were toxin A positive/toxin B positive. Overall, the isolates of C. difficile in this study showed lower susceptibility to quinolones, but their toxinogenic characteristics were different from those of ribotype 027. Only two isolates had binary toxin and gene deletions in tcdC, but neither had an 18-bp deletion.

In vitro activity of nemonoxacin against C. difficile has not been reported (18). In the present study, nemonoxacin had lower MICs against C. difficile isolates than moxifloxacin and gemifloxacin. Possible explanations for this finding include that nemonoxacin requires mutations in three different bacterial genes, gyrA, gyrB, and parC, rather than the two genes needed for fluoroquinolones (16). Whether the lower MIC values of nemonoxacin can be translated into the potential therapeutic or preventive benefits of such a drug in dealing with C. difficile infections remains to be defined.

Our C. difficile isolates were universally susceptible to metronidazole and ampicillin-sulbactam. The distribution of antibiotics in the gut should also be a therapeutic concern. Metronidazole was still a favorable drug for CDI treatment but has low concentrations in the gut even in the presence of diarrhea (3). Clindamycin showed the highest MICs of all antimicrobial agents tested. The use of clindamycin is associated with a high risk of inducing CDI (26). In a previous study, daptomycin showed higher MICs than metronidazole and vancomycin (24). Meropenem and doripenem showed relatively low MIC levels, which could suggest a lower risk of CDI among carbapenems. Hecht et al. (12) also reported that meropenem and doripenem had low MIC₉₀ values (2 μg/ml) but did not compare them to those of other carbapenems.

Oral metronidazole or vancomycin is the standard therapy for C. difficile infection. Standard therapy becomes less effective with hypervirulent strains (30). The possible reasons include decreased susceptibility to standard antibiotics and hyperproduction of toxins (30). No effective agents have been recommended for cases with treatment failure. Patients with severe and refractory CDI have been treated successfully with intravenous tigecycline (13, 21). Tigecycline had the lowest MIC₉₀ values for C. difficile and was followed by daptomycin, metronidazole, and vancomycin (1 μg/ml). The fecal concentrations of tigecycline in formed stools are relatively high (23) but did not induce increasing amounts of C. difficile or cytotoxin production in C. difficile in a gut model (1). Previously reported tigecycline MIC₉₀ values for C. difficile were low, similar to those of the present study, ranging from 0.06 to 0.25 μg/ml (13, 24).

Although PCR ribotyping and PFGE analysis were not performed, rep-PCR typing has been documented to be well correlated with PCR ribotyping for the screening of hypervirulent C. difficile strains (17, 27). Pasanen et al. showed that all isolates belonging to PCR ribotypes 027 and 001 clustered in their own rep-PCR groups, indicating that this method was able to screen out the hypervirulent ribotype BI/NAP1/027 strain (27). Our results were also in accordance with their findings (27). All isolates belonging to the main rep-PCR types (types A and C) were not hypervirulent ribotype BI/NAP1/027 strains by PCR ribotyping. Furthermore, the rep-PCR typing method is easy to use, requires less hands-on time than PCR ribotyping or PFGE typing, and can be a reliable option as the first-line molecular typing in local clinical microbiology laboratories (27).

In summary, this study documented the dissemination of fluoroquinolone-resistant rep-PCR type isolates of C. difficile in Taiwan. The resistance of isolates of several rep-PCR types could be attributable to several point mutations of gyrA and gyrB genes. Although no highly virulent NAP1 (ribotype 027) isolates were detected in this study, some isolates exhibited similar toxin profiles.