< Back to the previous page ADD TO BASKET

Whole genome sequencing: from molecular characterization to typing

Relatedness of European C. difficile strains from humans, food and animals by whole genome sequencing, ribotyping and toxinotyping; results from COMBACTE-CDI

April 18 • P1130

K. Davies1, V. Viprey1, V. Tkalec2, N. Devos3, E. Santiago4, D. Ewin1, W. Spittal1, J. Vernon1, W. Fawley5, J. Dombrecht4, A. Benson1, G. Davis1, B. Sente3, M. Rupnik2, P. Cleuziat4, M. Wilcox1, on behalf of the COMBACTE-CDI consortium

1) Healthcare Associated Infections Research group, University of Leeds, Leeds, UK
2) National laboratory for health, environment and food, NLZOH, Maribor, Slovenia
3) GSK, Rixensart, Belgium
4) bioMérieux, Chemin de l’Orme, Marcy l’Étoile, France
5) National infection service, Public Health England, Leeds, UK

Background: Previous data from human-derived Clostridium difficile (CD) strains indicates country-specific clustering for some but not all ribotypes. COMBACTE-CDI provides the opportunity to explore this further with the addition of food/animal strains.

Methods: On two days, all diarrhoeal samples (regardless of tests requested, n=3163) from 119 sites in 12 European countries, were CD cultured at the coordinating laboratory (positive=287). Contemporaneous CD isolates were collected from animals (n=126) and food (n=67) in the same countries. All isolates underwent PCR-ribotyping, toxinotyping, and whole genome sequencing (WGS); dendrograms were constructed in bioMérieux EPISEQ® CS using wgMLST with 95% cut-off and virulome compared between strains.

Using WGS, isolates clustered by ribotype and toxinotype, although matches were not exact. Some ribotypes had within-country clustering (027, 181) while others did not (002, 015, 014, 020). Similarity threshold was increased to 99% to analyse highly-related ribotype 078 and 126 strains; demonstrating five distinct clusters with no within-country clustering for human isolates; animal and food isolates did cluster by country (Figure 1). Analysis highlighted one UK 027 isolate, clustered within 027 isolates from Poland; one 181 isolate from UK clustered within 181 isolates from Romania; one 018 isolate from an Italian community patient clustered next to an isolate from an Italian hospital patient. Virulome analysis on EPISEQ® CS aligned with toxinotyping, with two discrepancies (PaLoc not detected by EPISEQ® CS). Both parts of the binary toxin genes were present in all isolates from ribotypes 027, 181, 078 and 126; however several ribotypes only had cdtB (015, 001 and 005). All isolates from ribotypes 078 and 126 were deficient in fliA, which was ubiquitous in almost all other ribotypes.

Discussion: There was not an exact match between WGS and ribotype, probably because these are two distinct typing schemes examining different parts of the genome. Previous data on strains that show within-country clustering was confirmed, while in contrast to previous data there was some country-specific clustering of 078 and 126 strains; however this was for food or animal source isolates only. EPISEQ® CS analysis highlighted potential transmission events for further investigation, and differences in virulence factors between ribotypes.

Figure 1. Dendrogram of ribotype 078 and 126 strains, showing; country of origin, source (human, animal or food), location (hospital or community), toxinotype and derived MLST. The clustering scheme was produced in bioMérieux EPISEQ® CS, with colour coding superimposed over the original metadata list using Excel.


Combacte CDI

Paper poster