INTRODUCTION
Clostridium difficile infection (CDI) is the leading cause of hospital-acquired diarrhoea and is associated with a considerable health and economic burden [Reference Dubberke and Olsen1–Reference Miller4]. Between 12% and 18% of patients progress to severe disease [Reference Henrich5–Reference Lungulescu7], and about 20% develop ⩾1 recurrent episodes (poor outcome) [Reference McFarland8]. Although several host-related and pathogen-related factors have been proposed as risk factors for poor outcome of CDI, the reasons why some patients experience severe disease and poor outcome remain unclear [Reference Voth and Ballard9].
The virulence of C. difficile is caused by two major toxins (toxin A and toxin B). Some C. difficile ribotypes, including ribotype 027 strains, also produce a binary toxin (BT) [Reference Goncalves10, Reference Barbut11]. In the past decade, C. difficile ribotype 027/NAP1/BI and its BT have been associated with poor outcome [Reference Warny12].
However, the role of BT in the outcome of CDI episodes caused by ribotypes other than 027/NAP1/BI has rarely been explored.
Our objective was to investigate the role of BT isolates in the severity and outcome of CDI in a non-027 ribotype setting.
MATERIAL AND METHODS
Design and study population
From January 2013 to June 2013, we prospectively included all patients diagnosed with CDI in our institution. Children aged <2 years and patients with recurrence of an episode from before the study period were not included. Patients were followed-up throughout the study period and for at least 2 months after their last CDI episode or recurrence.
In our institution, the presence or absence of BT is not routinely reported to the patient's physician nor was it reported during the study period.
Definitions
A CDI episode was defined as a positive test result for toxigenic C. difficile and the presence of diarrhoea (⩾3 unformed stools in 24 h) or colonoscopic findings demonstrating pseudomembranous colitis.
The type of CDI episode according to the potential site of acquisition was defined according to the criteria of the European Society of Clinical Microbiology and Infectious Diseases (ESCMID) Study Group for Clostridium difficile [Reference Kuijper13].
Severity of CDI was defined according the guidelines of the Society of Healthcare Epidemiology of America (SHEA) and the Infectious Diseases Society of America (IDSA) for mild to moderate and severe CDI [Reference Cohen14]. The definition of severe complicated CDI was slightly modified, as follows: CDI in patients with septic shock or colectomy or megacolon or CDI-related admission to the intensive care unit (ICU) within 7 days of the positive sample or CDI-attributable mortality.
Recurrent CDI (R-CDI) was defined as the return of symptoms after recovery from a previous episode (at least 3 days without diarrhoea and clinical improvement) and a positive stool sample separated from the previous sample by 15–60 days. Episodes occurring >60 days after the previous episode were not considered recurrences but new episodes.
Treatment was considered to have failed when the patient did not recover from a CDI episode (⩾3 loose bowel movements or stool consistency did not improve and persistence of fever, abdominal pain or analytical parameters of disease severity did not improve) and a stool sample separated from the previous sample by <15 days was positive.
Poor outcome was defined as R-CDI, treatment failure, or progression to severe complicated CDI.
Mortality was considered to be associated with CDI when death was not clearly attributable to other unrelated causes, when it occurred within 10 days of the CDI diagnosis, or when it was due to well-known complications of CDI.
Data collection
The demographic data collected included age, sex, hospital department or outpatient clinic at the time of diagnosis, and history of hospital admissions up to 3 months prior to infection (to determine the source of CDI). Data regarding the underlying condition were recorded using the McCabe & Jackson classification, and comorbidity was assessed according to the Charlson index [Reference Charlson15, Reference McCabe and Jackson16]. Risk factors for CDI during the month prior to the diagnosis of CDI were recorded.
The data recorded for the CDI episode were as follows: days of diarrhoea, presence of abdominal pain, abdominal distension, fever, hypotension, toxic megacolon, pseudomembranous colitis, and severity of CDI. Antibiotic treatment for the CDI episode and outcome (need for ICU admission, need for surgery for the CDI episode, recurrence, mortality, and CDI-associated mortality) were also recorded.
Laboratory procedure
All samples underwent toxigenic culture in Clostridium selective agar medium (bioMérieux, France); plates were incubated under anaerobic conditions at 35–37 °C for 48 h. Following incubation, colony morphotypes compatible with C. difficile were selected with the help of a binocular magnifying glass. Colonies suspected of being toxigenic C. difficile were finally identified using the immunochromatographic system (C Diff Quik-Chek Complete assay, TechLab, USA) and the MRC-5 cell line cytotoxicity test.
The genes of toxin A (tcdA conserved fragment, tcdA deleted fragment), toxin B (tcdB), and BT (cdtA and cdtB) were detected using multiplex polymerase chain reaction (PCR) with an internal amplification control following a method adapted from other studies [Reference Persson17, Reference Lemee18].
All the toxigenic strains (presence of any of the genes of the toxins) were characterized using PCR ribotyping according to the procedure described by Stubbs et al. [Reference Stubbs19]. Phylogenetic analysis of ribotyping profiles was performed using the unweighted pair-group method with arithmetic mean (UPGMA) and Dice coefficients (Bionumerics v. 5.0, Applied Maths, Belgium). The profiles of our isolates were compared with international ribotyping profile libraries. Ribotypes were named using the international designation. When the correspondence with international ribotypes was unknown, the letter R followed by a number was used.
Data analysis
All analyses were performed using SPSS v. 18.0 (SPSS Inc., USA). Qualitative variables appear with their frequency distribution. Quantitative variables are expressed as the median and interquartile range (IQR). Groups were compared using Fisher's exact test for categorical variables and the t test (normal distribution) or Mann–Whitney test (non-normal distribution) for continuous variables. In order to determine the risk factors for poor outcome, we performed a bivariate analysis between the variables collected and poor outcome. All variables with P < 0·05 and the BT variable were included in a multivariate logistic regression model to assess their correlation with poor outcome; we also performed separate analyses for R-CDI, mortality, and CDI-related mortality. The odds ratio (OR) and 95% confidence interval (CI) were calculated. A P value <0·05 was considered significant.
Ethical issues
This study was approved by the Ethics Committee of Hospital General Universitario Gregorio Marañón and the Spanish Agency for Medicines and Health Care Products.
RESULTS
During the study period, 326 CDI episodes met our inclusion criteria. Of these, 319 episodes were available for molecular analysis of the strains. We detected 54 (16·9%) episodes caused by C. difficile strains with a BT gene. Six of the 54 episodes involved a co-existing strain which lacked the BT gene.
Out of the 54 episodes caused by strains with BT, 49 (90·7%) were ribotype 078/126, in three (5·6%) episodes there was no correspondence with international ribotypes and two (3·7%) were ribotype 023.
The demographic and clinical characteristics of CDI episodes are described in Table 1. Of the 319 CDI episodes, 82·4% were mild to moderate and 17·6% were considered severe to severe complicated. Outcome was poor in 63 (19·7%) episodes; of these, 48 (15·0%) were R-CDI, six (1·9%) were considered treatment failures, and nine (2·8%) progressed to severe complicated disease. Pseudomembranous colitis was observed in four patients. Overall mortality was 9·4%, and mortality attributable to CDI was 1·6%.
Epidemiological and baseline clinical characteristics of CDI patients with and without BT
BT, Binary toxin; CDI, Clostridium difficile infection; IQR, interquartile range.
Epidemiological and baseline characteristics of patients with CDI caused by BT-positive or BT-negative strains (Table 1)
We found no differences between patients with CDI caused by a BT-positive strain and those with a BT-negative strain regarding age, sex, type of underlying condition or Charlson comorbidity index score. We did not find any significant differences in risk factors between the groups.
Regarding the CDI episode, no differences were found in the potential site of acquisition (community or healthcare setting). In both groups, most of the CDI episodes were mild to moderate (75·9% BT positive vs. 83·8% BT negative, P = 0·292), there were no significant differences in CDI severity between BT-positive and BT-negative episodes.
Comparison of outcome between patients with CDI caused by BT-positive or BT-negative strains (Table 2)
Patients with BT-positive strains presented a median of 1 more day of diarrhoea (BT-positive median 4·0 days of diarrhoea vs. BT-negative median 3·0 days, P = 0·198); however, it was not statistically significant. Patients with BT-positive strains had a shorter overall hospital stay than patients with BT-negative strains (BT-positive median 14·0 overall days stay vs. BT-negative median 19·0 days, P = 0·0·012); however, there were no significant differences in days of hospital stay after the diagnosis of the CDI episode (P = 0·384).
Outcome of CDI patients with and without BT
BT, Binary toxin; CDI, Clostridium difficile infection; IQR, interquartile range.
We found no differences between BT-positive and BT-negative CDI episodes in terms of recurrence rate (13·0% vs. 15·5%, P = 0·835), overall mortality (11·1% vs. 9·1%, P = 0·612), or CDI-related mortality (0% vs. 1·9%, P = 0·553).
In order to determine the risk factors for poor outcome, we performed a multivariate logistic regression model which included BT as a variable to assess its correlation with poor outcome. There was no association between BT and poor outcome (OR 0·793, 95% CI 0·243-2·591, P = 0·701).
We also performed separate analyses for R-CDI, mortality, and CDI-related mortality. None of these analyses revealed any association with the presence of BT (P > 0·05).
DISCUSSION
In our institution, CDI episodes caused by strains harbouring BT genes accounted for 16·9% of the CDI episodes. None of the strains were ribotyope 027; most strains with BT were ribotype 078/126. We found that the presence of BT genes was not associated with poor outcome.
Since the first description of BT by Popoff et al. [Reference Popoff20] in 1988, the prevalence of BT-positive strains has been relatively low (6·4%) [Reference Stubbs21]. However, the outbreaks caused by C. difficile ribotype 027/NAP1/BI in Europe, Canada, and the United States revealed an increase in prevalence, which reached as high as 82% of all clinical isolates during the outbreaks [Reference Warny12, Reference McDonald22, Reference Loo23]. In our study, the prevalence of BT-positive strains was similar to that reported for settings with a low prevalence of ribotype 027 [Reference Barbut11, Reference Barbut24, Reference Alcala25]. None of the strains belonged to ribotype 027, this is line with low prevalence of ribotype 027 in our country (Spain), which has been reported to range between 0% and 2·9% in two prevalence national studies and a European multicentre study [Reference Alcala25–Reference Alcala27].
The predominant ribotype of BT-positive strains was ribotype 078/126, which has been reported to be a ‘hypervirulent strain’ [Reference Goorhuis28]. Nowadays, the association between outcome and specific ribotype is open to debate, and studies of the association with the so-called hypervirulent ribotypes 078 and 027 in non-epidemic settings revealed no association between specific ribotypes and disease severity [Reference Walk29, Reference Sirard30].
An association between disease severity and BT was described in two retrospective studies by Barbut et al. [Reference Barbut24, Reference Barbut31]; however, in both studies the number of patients included that were positive for BT was very low (14 and 26 patients). We found no significant association between disease severity and BT. Another recent prospective study also failed to demonstrate a statistically significant association between disease severity and BT [Reference Kim32].
The presence of the BT gene was linked to an increased risk of R-CDI in a recent study [Reference Stewart33], interestingly in this study, ribotype 027 was not associated to R-CDI while ribotype 078 was found to be associated with R-CDI [Reference Stewart33]. We found that the frequency of recurrence did not vary in BT-positive and BT-negative CDI patients. Other recent studies have also observed this lack of association between presence of BT and recurrence [Reference Kim32, Reference Goldenberg and French34].
In a retrospective registry-based study, higher case fatality was observed for patients with a BT-positive strain than for those with a BT-negative strain [Reference Bacci35]. However, in this study, the authors were not able to obtain information on the cause of death, therefore CDI-related mortality and its association with the BT could not be examined. In our study, we found no differences between overall mortality of BT-positive and BT-negative CDI episodes. In fact, all cases of CDI-related mortality were recorded in BT-negative CDI patients. This finding is in line with the results published by Kim et al. [Reference Kim32], in which they observed no association between overall mortality or CDI-associated mortality.
Our study is limited by the fact that it was performed in a single centre, therefore our data cannot necessarily be extrapolated to other centres; a further limitation is that most of the BT strains corresponded to ribotype 078. In addition, we were not able to measure the actual production of BT.
In conclusion, in a non-027 setting, the prevalence of BT is significant. We found no association between recurrence of infection caused by BT strains, progress of disease severity, or mortality (poor outcome). It is possible that the association between BT and poor outcome has been overestimated, since it is based on data from the 027/NAP1/BI outbreaks.
ACKNOWLEDGEMENTS
We thank Thomas O'Boyle for his help in the preparation of the manuscript. This study was partially financed by Astellas Pharma Inc., the Rafael del Pino Foundation, and Fondo de Investigaciones Sanitarias (FIS), Research Project number PI13/00687. Elena Reigadas holds a grant from the Río Hortega programme of the Carlos III Health Institute, Spanish Government.
DECLARATION OF INTEREST
None.
