Antimicrobial Susceptibility and Molecular Characterization Using Whole-Genome Sequencing of Clostridioides difficile Collected in 82 Hospitals in Japan between 2014 and 2016

We studied the antimicrobial susceptibility and molecular characteristics, using draft whole-genome sequencing, of Clostridioides (Clostridium) difficile strains before and after treatment in adults with C. difficile infection (CDI) enrolled in a phase III, randomized, nationwide study of fidaxomicin versus vancomycin in Japan (ClinicalTrials.gov identifier NCT02179658). C. difficile strains were cultured from stool samples collected before and after standard treatment with either fidaxomicin or vancomycin.

one strain isolated from a patient cured of CDI had an elevated FDX MIC of 16 mg/liter at the time of recurrence in two phase III randomized, double-blind trials at sites in North America and seven European countries (12). This strain's mechanism of reduced susceptibility to FDX was not analyzed.
Few studies have evaluated clinically relevant strains of C. difficile in Japan with a view to examine their antibiotic susceptibilities and resistance mechanisms. Such information may help tailor the choice of treatment for CDI. In this study, we performed molecular characterization using whole-genome sequencing of C. difficile strains isolated from 188 samples from patients with CDI enrolled in a phase III study in Japan.

RESULTS
Participants and C. difficile strains. In all, 215 participants were randomized in the phase III study, with 106 participants to FDX and 109 participants to VCM (Fig. 1). A total of 285 C. difficile strains were recovered from the stool samples of 188 patients (87/106 [82.1%] who received FDX and 101/109 [92.7%] who received VCM). Of these, 188 were nonduplicate C. difficile strains recovered from patients FIG 1 Patient flow in the phase III, randomized, nationwide study of fidaxomicin versus vancomycin in Japan. a, patients diagnosed with Clostridioides difficile infection (CDI) were positive for C. difficile toxin A, B, or both in a stool specimen obtained within 96 h before randomization. b, days 1 to 10. c, if a stool specimen was obtained from the patient within 24 h of completing the treatment (days 10 and 11), C. difficile culture was performed. d, if diarrhea occurred within 28 days (Ϯ 3 days) after treatment, C. difficile culture was performed. e, one secondary C. difficile strain was isolated unexpectedly during fidaxomicin administration. f, two secondary C. difficile strains were isolated unexpectedly during vancomycin administration.
Whole-genome sequence analysis of C. difficile strains. We performed wholegenome sequencing of all 285 C. difficile isolates, with a mean (standard deviation) map read depth of 76.7 (30.0) (see Data Set S1 in the supplemental material). The assembled genomes had an average of 218.7 (128.6) contigs and an N 50 value of 53,033 bp (39,128 bp).
Antibiotic susceptibilities of C. difficile strains isolated from CDI episodes at baseline. No strains having reduced susceptibilities to FDX (MIC, Ն1 mg/liter) or with resistance to VCM and metronidazole (MNZ) were isolated from episodes of CDI at baseline. All strains belonging to ST17, ST81, and ST183 showed resistance to moxifloxacin (MFLX) and clindamycin (CLDM) ( Table 3). The numbers of strains having resistance to MFLX and CLDM among ST8 were 18 (69.2%) and 21 (80.8%) of 26 strains, respectively, and among ST2, the respective numbers of strains were 3 (14.3%) and 18 (85.7%) of 21 strains. One ST1 strain was resistant to MFLX (Table 3).
The FDX, MNZ, and MFLX MICs had increased by more than 4-fold compared with the MICs before treatment in six cases, three cases, and one case, respectively (Table 4).   C. difficile Characteristics Pre-and Posttreatment Antimicrobial Agents and Chemotherapy C. difficile strains with substantially reduced susceptibility to FDX were obtained only in the patient group that had received FDX treatment (six isolates with 30-to 2,000-fold reduced susceptibility) and not from the VCM group. No isolate from the VCM group showed reduced VCM susceptibility (Table 4). There was no substantial increase in VCM and CLDM MICs. Of the three cases from whom the isolated C. difficile strains had 4-fold reduced susceptibility to MNZ after VCM treatment, only one case (patient no. 8) had received MNZ prior to VCM treatment. One C. difficile isolate with 4-fold reduced susceptibility to MFLX was isolated from patient no. 157, who had not received any fluoroquinolone prior to VCM treatment. Six of 10 paired strains had reduced susceptibility to FDX after FDX treatment (MICs range, 0.25 to 64 mg/liter); all six patients who harbored these strains were reported to be cured after treatment. Identified single-nucleotide polymorphisms (SNPs) with amino acid substitutions were from valine to leucine, glycine, or aspartate at position 1143 in RpoB (Val1143Leu/ Gly/Asp) and from arginine to glycine at position 89 (Arg89Gly) in RpoC, which are previously described mutations associated with decreased FDX sensitivity (Table 4). Amino acid substitutions from glutamine to proline at position 1149 in RpoB (Glu1149Pro) and from arginine to cysteine at position 326 in RpoC (Arg326Cys) were putative mutations (Table 4). No other SNPs were detected in the C. difficile strains with reduced susceptibility to FDX. SNPs were not detected in paired strains with reduced susceptibility to MNZ or MFLX (Table 4).
Regarding C. difficile ST17 strains isolated in this study, there were nine pairs of strains that had a genetic distance of Ϲ2 SNPs, and while the possibility of a small number of transmission events could not be ruled out, there was no evidence of infection outbreaks (Fig. 2A). The transmission cutoff value based on the nucleotide substitution rate of C. difficile ST17 estimated in this study was evaluated as acceptable. The threshold of Ϲ2 SNPs was also adopted as the direct transmission cutoff value for STs other than ST17 (RT037) in previous reports (28,29). For ST2, ST8, ST81, and ST183, when the transmission threshold of Ϲ2 SNPs was applied, no outbreak was reported involving any of these STs.
After antibiotic susceptibility testing of isolated C. difficile strains, SNP analysis of whole-genome sequences from the strains with reduced susceptibilities isolated before and after treatment with FDX or VCM suggested that there was reduced susceptibility to FDX in six strains after FDX treatment. Mutations at positions 1143 of RpoB (Val1143Leu/Gly/Asp) and 89 of RpoC (Arg89Gly) have been reported to be determinants of reduced susceptibility to FDX (9). RpoB Gln1149Pro and RpoC Arg326Cys mutations had not been reported in previous studies. In a previous in vitro study, the frequency of detection of C. difficile strains with reduced susceptibility to FDX at 4-fold the MIC ranged from 1.28 ϫ 10 Ϫ8 to Ͻ1.41 ϫ 10 Ϫ9 (30). Also, the Val1143Gly/Asp mutation in RpoB appears to be associated with a fitness cost in vitro and reduced virulence in vivo (10). In this study, the RpoB and RpoC mutants that appeared at low frequency may have been selected for by FDX, because these mutants were recovered only in FDX-treated patients. However, it is not clear what the clinical implications are of using FDX to treat CDI caused by strains having reduced susceptibility to FDX. In practice, six cases of CDI from whom C. difficile strains with reduced susceptibilities to FDX were isolated were deemed to be clinically cured at the end of treatment with FDX (Table 4). Despite infection with strains having reduced susceptibilities to FDX, clinical response appears to reflect the high concentrations of FDX in stool (mean, 1,225 mg/ kg) achieved with a standard regimen of FDX (6).
A potential limitation of this study was that C. difficile strains recovered from patients after antibiotic treatment were isolated at different times, where culture of stool was performed after 10 days of antibiotic treatment, but isolates were also included from specimens collected during the follow-up period, i.e., between days 11 and 28. In a previous report, the same C. difficile ribotypes isolated at baseline were also isolated over a long period of time (14 to 56 days) in recurrent CDI cases (31). Therefore, the differences in the timing of sampling for analysis of C. difficile isolates in the present study appear to be acceptable. Another potential limitation was that the mutations not previously reported, RpoB Gln1149Pro and RpoC Arg326Cys, in C. difficile strains with reduced FDX susceptibility were not analyzed by allelic exchange methods, which are often used to examine the fitness cost of such mutations (11). Because RpoB Gln1149Pro and RpoC Arg326Cys mutations were only detected by comparing the core genomes of FDX-susceptible strains with those from strains having reduced susceptibilities to FDX, which were isolated from the same patient, we suggest that the mutations could contribute to the reduced susceptibility of C. difficile to FDX.
In conclusion, our findings showed that no C. difficile strain with reduced FDX susceptibility was isolated from patients with CDI in hospitals nationwide in Japan before FDX administration. However, mutant C. difficile strains with reduced FDX susceptibilities may have been selected for in the gut of patients treated with FDX. Future studies should assess the potential emergence of CDI caused by C. difficile strains having reduced susceptibility to FDX after the widespread introduction of FDX as a treatment for CDI in Japan.

MATERIALS AND METHODS
Ethics. Procedures completed at LSI Medience Corporation were conducted according to Astellas Research Ethics Committee (AREC) standards established at Astellas Pharma, Inc. Procedures completed at Toho University were deemed to be beyond the scope of examination by the AREC. The study was conducted with approval from the institutional review board of the Toho University Omori Medical Centre (no. 2810-CL-3002). Written informed consent was obtained from patients prior to the start of any study-related procedures using the written information for patients and informed consent form that was approved by the institutional review board of each study site.
Summary of the clinical trial. A phase III, VCM-controlled, double-blind, parallel-group study of FDX was conducted in 82 hospitals in Japan between June 2014 and September 2016 (32). The study was registered at ClinicalTrials.gov with the identifier NCT02179658. Briefly, participants were Ն20 years of age with a diagnosis of CDI, defined by the presence of diarrhea (with more than four unformed bowel movements in the 24-h period before randomization) and C. difficile toxin A, B, or both in a stool specimen obtained within 96 h before randomization. Patients could have received up to four doses of MNZ or VCM before randomization but no other potentially effective concurrent treatment for CDI. Enrolled patients were randomized to receive either oral FDX (200 mg every 12 h with intervening placebo given 6 h after FDX) or oral VCM at the clinically recommended dose (33) (125 mg every 6 h with intervening placebo given 6 h after VCM) for 10 days. Patients were assessed every day during the 10-day treatment period, for 2 days afterwards, and at least weekly during the 28-day follow-up. Patients were assessed at an end-of-treatment visit for clinical cure and at an end-of-study visit when recurrence had not been reported.
Bacterial isolation and species identification. Stool samples collected within 96 h before study randomization, within 24 h after completing treatment with FDX or VCM (days 10 and 11), and within 28 days (Ϯ3 days) after completing treatment with FDX or VCM (days 11 to 31) were sent to a central laboratory (LSI Medience Corporation) and plated directly onto chromID C. difficile agar (bioMérieux, France). Cultures were incubated in an anaerobic chamber (5% CO 2 , 10% H 2 , and 85% N 2 ) at 35 Ϯ 2°C for 24 h. Identification of bacterial species was performed using Rapid ID 32A API system (bioMérieux). Frozen, stored C. difficile isolates were sent to the Department of Microbiology and Infectious Diseases at the Toho University School of Medicine.
Core-genome single-nucleotide polymorphism analysis. Core-genome SNP-based phylogenetic analysis was performed with whole-genome sequencing data. MiSeq sequencing data were aligned to the genomic sequence of the reference isolate, C. difficile 630, using the Burrows-Wheeler Aligner with 'SW' algorithm (37). We aligned the core-genome sequences using the Sequence Alignment/Map software (SAMtools mpileup, version 1.1) (38), which were read using VarScan (version 2.3.7) mpileup2cns (39), and a maximum likelihood phylogenetic tree was constructed using PhyML (40). Using this as the starting tree, we inferred homologous recombination events that imported DNA fragments from beyond the phylogenetic clade and constructed a clonal phylogeny with corrected branch lengths using ClonalFrameML (41). The core genome, excluding homologous recombination sequences estimated using ClonalFrameML, was subjected to SNP detection.
Antibiotic susceptibility testing. Antibiotic susceptibility testing was performed at LSI Medience Corporation using the agar dilution method according to Clinical and Laboratory Standards Institute (CLSI) M11-A8 guidelines (42). The antibiotic agents used were FDX, MNZ, VCM, MFLX, and CLDM. The resistance breakpoints for FDX have not been established, while the breakpoints for the remaining drugs were based on CLSI M100-ED28 (MNZ, Ն32 mg/liter; MFLX, Ն8 mg/liter; CLDM, Ն8 mg/liter) (43) and European Committee on Antimicrobial Susceptibility Testing clinical breakpoints version 8.1 (VCM, Ն4 mg/liter) (44). C. difficile ATCC 700057 was used for a susceptibility testing quality control.
Data availability. The BioProject no. of this study is PRJDB7714. Draft genome sequences were deposited at the DNA Data Bank of Japan (https://www.ddbj.nig.ac.jp/ddbj/index-e.html) and with GenBank accession numbers BIMY01000000 to BIXW01000000 (see Data Set S1).

SUPPLEMENTAL MATERIAL
Supplemental material for this article may be found at https://doi.org/10.1128/AAC .01259-19.