In Vivo Selection of Rifamycin-Resistant Clostridium difficile during Rifaximin Therapy

TEXT

A 65-year-old male outpatient with an 18-month history of four earlier episodes of Clostridium difficile diarrhea, each treated with vancomycin, was diagnosed with a fifth episode. His symptoms began just 4 days after his last vancomycin treatment stopped. Rifaximin (200 mg three times a day) was begun 4 days later, on day 1 of this study. Neither fecal samples nor isolates from the earlier episodes were available for study.

This research was approved by TechLab, Inc., Institutional Review Board (IRB) number 1 (registration number IRB00003505, IRB organization number IORG0002905). On day 1, after giving his informed consent, the patient provided frequent fecal samples for research only. The results of the research tests we report were not used to support the patient's clinical diagnosis or guide treatment decisions. C. difficile spores were cultured on cycloserine-cefoxitin-fructose agar supplemented with sodium taurocholate. Four or five isolates from each sample were PCR ribotyped (9) and tested with Etest strips on brucella blood agar (7) for the MIC of rifampin (RIF). RIF and rifaximin are both members of the rifamycin class of antibiotics. Resistance to one indicates resistance to the other, and they may be used interchangeably when assaying MICs (8). rpoB_1418-1812, the rifamycin resistance-determining region (RRDR) of each of the four or five isolates per sample, was sequenced (1, 2). All isolates were of ribotype 056. All produced toxins A and B. None produced binary toxin. All were susceptible to vancomycin and moxifloxacin.

On day 1, only sensitive (RIF MIC, 0.002 μg/ml) isolates with a wild-type RRDR were recovered. Exposure to rifaximin led rapidly to the emergence of resistance, and within 32 h, only resistant (RIF MIC, ≥32 μg/ml) isolates were found. The isolates had a C1506→T substitution in rpoB, a single nucleotide polymorphism (SNP) resulting in the His502Tyr change seen in resistant C. difficile (8). As treatment continued, only resistant strains of ribotype 056 were seen, the original strain and a second strain with the C1506→T substitution plus an additional C1488→T substitution that encodes Pro496Ser. This second SNP within rpoB of C. difficile has not been reported before (7, 8). Its effect on resistance is unknown. The patient was asymptomatic for about 2 weeks after taking rifaximin, although throughout that time, both strains of resistant ribotype 056 isolates persisted. Ribotype 056 isolates are not frequent in our studies (<2% of almost 2,000 isolates), nor does ribotype 056 RIF resistance appear widespread. These are the only RIF-resistant members of ribotype 056 in our collection of 41 isolates.

On day 49, when the two resistant strains were still present, the diarrhea returned and 1 week later, a 10-day course of fidaxomicin was given and the patient's diarrhea resolved. There was no other antibiotic exposure at that time. All of the research assays for C. difficile (culture, cytotoxicity, toxin, and glutamate dehydrogenase immunoassays, and reverse transcription-PCR assays for tcdB and for tpi, which encodes triphosphate isomerase, a metabolic enzyme [5]) were negative for nearly a week, but then the patient had >30 days of asymptomatic colonization by susceptible ribotype 056 isolates (RIF MIC, 0.002 μg/ml). Since he was diarrhea free, however, antimicrobial therapy was not implemented. The reappearance of RIF-sensitive ribotype 056 isolates probably occurred because the patient, susceptible to postantibiotic colonization, picked up spores shed into his home environment during much earlier episodes of diarrhea or, despite our negative findings (see above), because of its emergence from a persistent but inapparent level within the colon to one of abundance. The reversion of both resistant strains to the wild type seems less probable, particularly in the strain with two rpoB SNPs.

Despite the long absence of diarrhea, the patient again experienced discomfort and soft stools on day 232, although he did not consider himself to have diarrhea. Culture yielded a RIF-sensitive C. difficile isolate of ribotype 005, a toxigenic ribotype, new to the patient since we began monitoring him. Antibiotics were not considered. He remains, as of day 324, asymptomatically colonized with ribotype 005 bacteria.

RIF-resistant ribotype 056 isolates were recovered within 24 h of treatment. The rapid emergence of C. difficile with SNP-mediated fluoroquinolone resistance in a patient receiving both ciprofloxacin and levofloxacin is a similar example of the selection of resistant C. difficile during antimicrobial exposure (6). Rifaximin therapy for C. difficile diarrhea, especially for recurring cases, has been endorsed (3) and may become more common. If so, we might see the emergence of resistant isolates of ribotype 056 (or of any other toxigenic ribotype acquiring resistance to RIF) along the lines of the recent surge in the incidence of moxifloxacin-resistant ribotype 027 isolates. In fact, instances of RIF-resistant C. difficile associated with rifaximin therapy have been seen (4), while the incidence of resistance, largely confined to the equivalent of ribotype 027 isolates, has doubled at one institute (from <40 to >80%) in 4 years (3). While a RIF MIC of ≥32 μg/ml may not mean a lack of efficacy, since rifaximin reaches very high concentrations in the stool (up to 8,000 μg/g), antibiotic resistance plays a continuing and significant role in the biology of C. difficile.