Introduction
Clostridium difficile infection (CDI) represents a major healthcare burden in the developed world [
18]. Metronidazole and vancomycin have been the mainstays of CDI treatment in recent decades [
24]; however, high recurrence rates and reports of reduced susceptibility to metronidazole among epidemic
C. difficile PCR ribotypes (RTs) have highlighted the need for new agents [
1,
24]. Fidaxomicin is a macrocyclic antibiotic with low MICs against
C. difficile, approved by the EMA in 2011 for the treatment of CDI in adults [
7]. In two phase 3, double-blind, randomized, parallel-group trials, it demonstrated non-inferiority in initial cure of CDI and lower rates of recurrence, compared with oral vancomycin [
5,
15]. Fidaxomicin is also associated with greater preservation of the intestinal microbiota compared with vancomycin [
14].
The aims of the 5-year ClosER study (2011–2016) were to identify and monitor the longitudinal antimicrobial susceptibility of C. difficile clinical isolates, including those previously implicated in selection pressure, to establish a comprehensive susceptibility database baseline for ongoing surveillance and to provide data on the geographical distribution of clinical C. difficile strain types across Europe.
Methods
ClosER was a Pan-European, multicentre, in vitro surveillance study, planned to provide data for 1 year prior to the introduction of fidaxomicin (July 2011–June 2012) and 2 years post-introduction (2012–2014). It was subsequently extended for a further 2 years (2014–2016). Participating centres were mostly national or regional C. difficile referral laboratories selected using the European Study Group on Clostridium difficile (ESGCD) network and with ESGCD approval. The number of sites approached per country was based on population (1 site per 15 million population) or reported incidence of CDI (≥ 2 sites for countries with > 20 cases per 10,000 patient days per hospital). Fifty-one sites from 28 European countries were asked to participate. Criteria for site inclusion were active sampling and testing for CDI, sufficient numbers of clinical CDI cases to reach a target of 25 de-duplicated cases during the 6-month collection period and a willingness to submit the required number of samples over the 3 years. The 40 sites that contributed samples during years 1–3 were contacted to request their participation in years 4 and 5 of the study. Of these, 28 sites agreed to participate in the extended study, three sites formally ended their participation after year 3, and nine sites were unresponsive.
Isolates or faecal samples were submitted to a central laboratory (Leeds, UK) for PCR ribotyping, determination of toxin status and assessment of susceptibility to metronidazole, vancomycin, rifampicin, fidaxomicin, moxifloxacin, clindamycin, imipenem, chloramphenicol and tigecycline, using methods described previously [
9,
10] (Online Resource
1). Participating sites were asked to provide the following demographic data to accompany each sample: age, gender, history of CDI in the previous 6 months, healthcare or community CDI episode, and antimicrobial administration 1 month prior to the episode.
RT diversity and Cumulative Resistance Scores were calculated each year for individual countries (Online Resource
1).
Discussion
To date, this is the largest Pan-European study of
C. difficile RT prevalence and antimicrobial resistance. Almost 3500 isolates were received, yielding 264 distinct RTs. Over the first 4 years of the study, the prevalence of the 10 most common RTs remained stable and the changes observed in year 5 should be interpreted with caution due to a substantially reduced number of submissions and participating countries. The most prevalent RTs found in our study correspond closely to those previously reported in 2011 [
3] and 2016 [
6], indicating overall stability in RT prevalence over time. Some fluctuation in relative RT prevalence between years and between countries was observed, as expected due to endemic and epidemic spread of
C. difficile [
8]. Previously described epidemic or highly prevalent RTs, such as 014, 027, 001 and 078, remained highly prevalent in this study. RT005, RT087 and RT356 were more prevalent in year 1 than previously observed [
3], but of these, only RT005 remained highly prevalent throughout the study.
Fidaxomicin MICs remained consistently low throughout the 5 years of the study, and there was no evidence of a reduction in susceptibility following its introduction in 2011. This is consistent with an earlier antimicrobial susceptibility survey of isolates from two phase III studies of 1164 patients that reported fidaxomicin MIC90s of 0.25 mg/L [
11]. The same survey identified a single strain of
C. difficile from a patient with recurrence with a fidaxomicin MIC of 16 mg/L; however, the relatedness of the pre- and post-treatment strains was not determined [
11], and the association of resistance with drug exposure cannot be made definitively [
21]. Schwanbeck et al. described fidaxomicin resistance (MIC > 64 mg/L), associated with a V1143D mutation in rpoB, in a single clinical
C. difficile isolate of 50 isolates tested [
19]; however, the fitness burden imposed by the mutation was higher when generated in vitro [
13,
19] than observed in the clinical isolate [
19]. In the light of high fidaxomicin gut concentrations, the significance of this is unclear but highlights the need for further monitoring. We report a single fidaxomicin-resistant isolate of RT344 (MIC ≥ 4 mg/L), isolated in year 5 of the study; this isolate was also resistant to moxifloxacin, clindamycin and imipenem, but sensitive to all other antimicrobials tested. All other RT344 isolates submitted were susceptible to fidaxomicin.
RT027, the most prevalent RT in the
ClosER study, has previously been associated with multiple antimicrobial resistance [
17] and reduced susceptibility to fidaxomicin compared with other PCR ribotypes (MIC90s 0.5 mg/L versus ≤ 0.25 mg/L, respectively) [
11]. However, no such association was found in the
ClosER study. The geometric mean fidaxomicin MIC for the RT027 isolates submitted to this study (0.04–0.08 mg/L) was below the susceptibility breakpoint. Likewise, RT017, 012, 018 and 356 showed resistance to multiple antimicrobials in this study but were not associated with higher fidaxomicin MICs. Moreover, the clinical significance of RT-specific variations in fidaxomicin susceptibility is questionable, particularly given the high concentrations of fidaxomicin (> 1000 μg/g) attained in the gut [
20].
While susceptibility to fidaxomicin remained stable over years 1–5, susceptibility to metronidazole and vancomycin increased. This could be attributed to a reduction in metronidazole and vancomycin use and/or greater strain diversity. The present study confirmed previously reported associations between prevalent RTs, such as RT027 and RT001, and resistance to moxifloxacin, clindamycin and chloramphenicol [
2,
16,
22]. However, there were examples of these RTs from many countries showing almost full susceptibility to all agents. Imipenem resistance is not well-documented for
C. difficile, but we found evidence of both intermediate and full resistance in all years.
Fluctuations in antimicrobial susceptibility between countries and between years are reflective of the varying prevalence of RTs. Multiple antimicrobial resistance was most evident in certain epidemic RTs, such as RT027 and 001, but was also notable in RT017, RT012, emerging RT198 (exclusive to Hungary) and RT356 (exclusive to Italy). We found a consistent inverse correlation between RT diversity and mean CRS for a given country, possibly due to the introduction of mandatory reporting programmes with a subsequent increase in awareness, antimicrobial stewardship and infection control interventions reducing rates of endemic RTs.
The selection criteria for submissions stipulated 25 de-duplicated toxin-positive faecal samples or C. difficile isolates, with no further requirements. There may, therefore, be selection bias in the samples submitted from any location. Participating centres were mainly national or regional C. difficile reference facilities and, therefore, some submissions likely included outbreak strains, possibly influencing the data. The majority of submissions were isolates rather than faecal samples. The recovery rate was generally poorer from faecal samples (mean 86%; median 84%) than from isolates (mean 97%; median 100%). In years 1 and 2, three sites submitted faecal samples, with recovery rates between 64 and 100%. In years 3 and 4, two of these sites submitted isolates: for one of these sites, recovery rates increased to 100% in both years, while for the other site, recovery rates increased to 84% and 92% in years 3 and 4, respectively.
We advised sites on how to prepare isolates as spores for transport but did not obtain information on whether this advice had been followed. We experienced consistently poor C. difficile recovery from isolate submissions by one site during all 3 years of their participation (76%, 40% and 52%). Another site had a recovery rate of 100% in year 1, but this rate dropped to 88% and 82% for years 2 and 3, respectively, after changing to different transport conditions. It was notable that all of these were not submitted as advised in transport media, and it is possible that this contributed to the poor recovery rates. However, these sites were not the only participating locations from their country (1 of 3 and 1 of 4, respectively) and, therefore, the effect of low recovery rates was lessened somewhat. Despite these site-specific limitations, overall recovery rates were > 96% for 90%, 83%, 85% and 91% of sites in years 1, 2, 3 and 4, respectively.
There was a substantial decrease in the responsiveness of sites, and consequently the number of submissions, during years 4 and 5. Possible reasons for a lack of response included site staff resourcing issues, the extended duration of the study and the loss or retirement of named site contacts or national coordinators. Although sample transport was provided, courier transport was problematic in some countries and there was no financial incentive for sites to submit samples. In year 5, only the Czech Republic, France, Germany, Ireland, Spain and the UK submitted isolates. From the UK, only one site submitted samples in year 5, while all four sites consistently submitted samples in years 1–4. Countries with high RT027 prevalence were, therefore, not represented in year 5, skewing the data. Accompanying patient data, particularly information on antimicrobial treatment, was often missing.
Acknowledgements
We would like to thank the following ClosER study participants: Franz Allerberger, Sabine Pfeiffer, Anita Fiedler, Steliana Huhulescu and Markus Hell (Austria); Michel Delmée, Kate Soumillion, Johann van Broeck and Eléonore Ngyuvulu Mantu (Belgium); Kate Ivanova and Elina Dobreva (Bulgaria); Panayiota Maikanti-Charalampous (Cyprus); Otakar Nyc (Czech Republic); Jørgen Engberg (Denmark); Janne Aittoniemi (Finland); Frédéric Barbut, Catherine Eckert, Hélène Marchandin and Hélène Jean-Pierre (France); Mathias Herrmann, Barbara Gärtner, Fabian Berger, Lutz von Müller, Reinier Mutters, Sören Schubert and Jana Bader (Germany); Maria Orfanidou, Stavroula Smilakou and Ioannis Deliolanis, Eleni Malamou-Lada (Greece); Elisabeth Nagy, Edit Urban, Zsuzsanna Barna and Katalin Kristóf (Hungary); Mairead Skally, Fidelma Fitzpatrick, Lynda Fenelon, Frank Dennehy and Katharina Stein (Ireland); Paola Mastrantonio, Patrizia Spigaglia, Claudio Farina, Francesca Vailati, Marco Passera, Luca Masucci, Maurizzio Sanguinetti, Domenico Nagel, Gianluca Quaranta, Rosalia Graffeo, Teresa Zaccaria, Giovanni Gesu and Maria Chiara Sironi (Italy); Arta Balode (Latvia); Hanna Pituch (Poland); Mónica Oleastro (Portugal); Elena Nováková, Vladimíra Sadloňová and Jana Kompaníková (Slovakia); Maja Rupnik and Sandra Janežič (Slovenia); Emilio Bouza, Elena Reigadas, Luis Alcalá, Josefina Liñares, Fe Tubau and Jordi Niubo (Spain); Torbjörn Norén (Sweden); Andreas Widmer, Reno Frei, Adrian Egli, Martin Altwegg and Livia Berlinger (Switzerland); Ed Kuijper and Celine Harmanus (The Netherlands); Derek Fairley (UK-N. Ireland); James McKenna, Henry Mather, John Coia (UK-Scotland), Trefor Morris (UK-Wales), David Griffiths, Derrick Crook, Irene Monahan and Tim Planche (UK-England).
We are grateful to Warren Fawley, Peter Parnell, Emma Best and Paul Verity at CDRN Leeds for performing PCR ribotyping and assignments. We would like to thank Chris Longshaw for his substantial contributions to the study. We would like to thank the ESCMID Study Group for Clostridium difficile (ESGCD) for their professional support.
Medical writing support was provided by Iona Easthope of Cello Health MedErgy, funded by Astellas Pharma, Inc.
These data were presented in part as a poster at the 28th European Congress of Clinical Microbiology and Infectious Diseases (ECCMID), Madrid, Spain, 21–24 April 2018.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.