Introduction

Clostridium difficile, a Gram-positive, anaerobic, sporogenic bacterium, is considered to be the leading cause of hospital-acquired and antibiotic-associated diarrhea [1, 2]. It colonizes the intestinal tract in up to 3% of healthy adults and in up to 80% of healthy newborns [3–5]. Clinical symptoms range from watery diarrhea to pseudomembranous colitis and toxic megacolon [1]. C. difficile infection (CDI) occurs mainly sporadically; however, outbreaks are not uncommon [6, 7]. The incidence and severity of CDI has been increasing since 2002 [8–12]. C. difficile strains of polymerase chain reaction (PCR) ribotypes 027 and 078 are known to cause more severe disease and are associated with higher morbidity, higher mortality, and an increased risk of relapse [12–16]. Recent studies also indicate an increase in the severe outcomes of community-acquired CDI [17–20].

van den Berg et al. [21] postulated that the coexistence of multiple PCR ribotypes of C. difficile in fecal samples limits the value of PCR ribotyping for epidemiologic studies. As patients often develop recurrence of CDI, using PCR ribotyping to differentiate between relapse versus re-infection is only feasible if CDI within an individual is usually monoclonal. It was the aim of this study to elucidate whether CDI is of monoclonal or polyclonal genesis.

Materials and methods

From December 2009 until June 2010, 11 patients who developed nosocomial CDI >48 h after admission to the 1,252-bed, tertiary care hospital in Salzburg, Austria, were chosen arbitrarily. During that time period, this institution was experiencing its usual baseline rate of CDI without any signs of clustering. Demographic data and information on clinical presentation and outcome were gained by clinical visits, chart reviews, and telephone calls. The study was presented to the local ethics committee; in view of the legal obligation for the surveillance of CDI, no formal approval was required.

Stool samples had been screened for C. difficile toxins A and B using enzyme-linked immunoassay (RIDASCREEN Clostridium difficile Toxin A/B ELISA, R-Biopharm AG, Darmstadt, Germany). Toxin-positive stool samples were spread onto cycloserine/cefoxitin agar plates (bioMérieux, Marcy l’Etoile, France) and incubated for 48 h at 35°C in an anaerobic atmosphere produced by AnaeroGen (Oxoid, Hampshire, England, GB). Five individual colonies from different areas of each primary agar plate (n = 11) were picked and subcultured. In pilot experiments, we often experienced difficulties to find more than five clearly distinct colonies per primary culture plate.

The 55 subcultured isolates were sent to the National Reference Centre for Clostridium difficile of the Austrian Agency for Health and Food Safety (AGES) in Vienna, Austria, using BBLPort-A-Cul-Tubes (BD Diagnostic Systems, Sparks, CA, USA). A total number of 47 isolates originating from 11 CDI patients were finally available (eight isolates failed attempts to reculture) (refer to Table 2).

Capillary gel electrophoresis-based PCR ribotyping and testing for toxin genes was performed in a blinded manner with primers as described elsewhere [22]. In brief, for PCR ribotyping, the 16S primer was labeled at the 5′ end with tetrachlorofluorescein. Twenty-five μl HotStar Taq Master Mix (Qiagen, Hilden, Germany) was used with 0.3 μl (10 pmol/μl) of each primer, 20.7 μl water, and 1.5 μl DNA. Amplification was done with a PCR thermocycler running a 15-min 95°C initial enzyme activation, 22 cycles of 1 min at 95°C for denaturation, 1 min at 57°C for annealing, 1 min at 72°C for elongation, and a 30-min 72°C final elongation step. PCR fragments were analyzed using an ABI 310 (Applied Biosystems, Foster City, CA, USA) genetic analyzer, with a 41-cm capillary loaded with a POP4 gel (Applied Biosystems, Foster City, CA, USA). A 50–625-bp TAMRA ladder (CHIMERx, Milwaukee, WI, USA) was used as an internal marker for each sample. Injection was done with 5 kV over 5 s, with a total running time of 28 min at the 15-kV run voltage. The size of each peak was determined by Peakscanner software 1.0 (Applied Biosystems, Foster City, CA, USA). To confirm the production of toxins A and B by the 47 toxigenic isolates, the Vidas C. difficile Toxin A & B (CDAB) assay (bioMérieux, Marcy l’Etoile, France) was utilized.

Results

Cultures for C. difficile were grown from the fecal samples of 11 CDI patients. The 11 patients had a median age of 67 years (range 25–90; mean age 68.7 years) and 4 (36.4%) were male. Symptoms of nosocomial CDI ranged from mild diarrhea to fatal pseudomembranous colitis. The patients’ salient clinical features are summarized in Table 1.

Table 1 Summarized data on patients and isolates from the primary stool culture plate available for polymerase chain reaction (PCR) ribotyping
Full size table

The 47 isolates yielded eight different PCR ribotypes: Ribotype 053 (12 isolates, 25.5%) from three patients; Ribotype 001 (10 isolates, 1.3%) from two patients; Ribotype 018 (7 isolates,14.9%) from two patients; Ribotype 538 (5 isolates, 10.6%), Ribotype 002/2 (5 isolates,10.6%), Ribotype 438 (4 isolates,8.5%), Ribotype 014 (3 isolates, 6.4%), and Ribotype 232 (1 isolate, 2.1%) were found in only one patient each (Table 2).

Table 2 Summarized results on the PCR ribotypes and toxin types
Full size table

Of the 11 patients, only one (9.1%) had more than one different PCR ribotype on her primary stool culture plate: four subcultures of patient 4 contained isolates that were PCR ribotype 438 and one subculture was PCR ribotype 232.

Four isolates (PCR ribotype 438) tested positive for toxin A, toxin B, and binary toxin. The other 43 isolates (92.5%) tested positive for toxin A plus toxin B only.

Discussion

Indra et al. [22] suggested PCR ribotyping to be a valuable tool for recognizing related cases of CDI in healthcare facilities. Also, McDonald et al. [9] postulated that the clustering of certain ribotypes in a healthcare setting should prompt an in-depth epidemiological investigation of a possible nosocomial outbreak and emphasized the need to develop a surveillance system for CDI. However, the molecular typing of C. difficile can support the detection and spread of clusters of CDI only if the pathogenesis is monoclonal. PCR ribotyping is also helpful in differentiating recurrence from re-infection in the individual patient only if polyclonal genesis is a relatively rare event.

There was controversy about the occurrence of the simultaneous existence of different strains in one fecal sample [21, 23]. While van den Berg et al. [21] described 23 patients with a first episode of CDI, of which two harbored two different PCR ribotypes within one fecal sample, another study group postulated monoclonal rather than polyclonal origin of CDI. To investigate the possibility of multiple-strain carriage, O’Neill et al. [24] assessed ten patients by taking ten colonies from a primary culture plate for restriction enzyme analysis (REA) typing. All isolates from each patient were indistinguishable from each other, which indicates that the carriage of multiple C. difficile strains is a rather rare event. Devlin et al. [23] also used REA for analyzing multiple isolates of C. difficile in a single stool sample of various patients and found that the REA patterns to be indistinguishable from each other for the individual patients.

Recurrences of CDI can either be a relapse caused by persisting C. difficile spores that are resistant to antibiotic therapy or a re-infection with a new strain. Without the subtyping of isolates, this differentiation is impossible to achieve. Recurrences occur in approximately 25% of CDI patients [25]. The persistence of C. difficile spores after the elimination of vegetative cells by standard antibiotic treatment regimens, i.e., genuine relapses, are supposed to be the dominant mechanism [17, 25]. Recurrence can also be due to the exogenous acquisition of a new strain (re-infection), which seems to be the reason in up to 50% of relapsed CDI cases [26]. Using REA, O’Neill et al. typed multiple isolates from ten patients with recurrent CDI and showed that less than half of those patients relapsed due to the same organism. Our patient 4, a 67-year-old female cancer patient, suffered from recurring diarrhea even before the fatal episode of colitis under study. The fact that this patient was the only one that harbored two different strains and was also the only one with proof of recurrent diarrhea is of special interest. Testing was only performed during the recurrent episode. We hypothesize that one of the two strains was present during the initial untested episode; the fatal episode could then be explained by a superinfection with a new strain. The other ten patients had a documented first manifestation of CDI. This phenomenon of double infection seems similar to what is known for, e.g., tuberculosis [27, 28]. We conclude that the possibility of double infections should be considered also in patients with CDI.

Dale Gerding recently patented the use of non-toxigenic C. difficile strains as a vaccine capable of preventing colonization and illness with toxigenic C. difficile (European Patent EP0952773), wherein the non-toxigenic strain “is administered after initiation of antibiotic treatment, in an amount sufficient to establish colonization of the gastrointestinal tract of a subject to prevent C. difficile associated disease” (Colonization is unlikely to occur if antibiotics have not been given, because the normal bacterial flora of the gastrointestinal tract can prevent colonization). This concept of protective colonization is highly dependent on a predominantly monoclonal mechanism of CDI within individual patients.

PCR ribotyping is presently the dominant C. difficile typing method used in Europe. Multilocus variable number tandem repeat analysis (MLVA) has been shown to be more discriminatory than other typing methods [29, 30]. However, this method is clearly prone to over-discrimination: Tanner et al. investigated the presence of several subtypes of 027 isolates in the same fecal sample and found that 5 of 39 samples yielded at least one strain with a different MLVA pattern [31]. The fact that, in our study, C. difficile infections of all but one patient were caused by a monoclonal organism underscores the considerable potential of PCR ribotyping to support epidemiological outbreak investigation and to differentiate between relapse and re-infection.