Understanding molecular epidemiology of Clostridium difficile in Scotland

Clostridium difficile infection (CDI) is a leading cause of healthcare-associated diarrhoea and since 2007 in Scotland marked changes to intervention strategies including infection control and antibiotic stewardship contributed to an overall 70% reduction in infection rates. In parallel, there has been a shift in the prevalence of particular ribotypes from those which were fluoroquinolone resistant (001, 027, 106) to other ribotypes such as 005, 014, 015, 078. The reason why these new PCR-ribotypes are becoming dominant is unclear at present although the increased rates of C. difficile being diagnosed within the community rather than a healthcare setting rather have a role in the increasing prevalence of these strains.

To study this hypothesis in more detail, whole genome sequencing is being used to examine the evolutionary relationships between these isolates. At present, DNA has been isolated from a collection of 299 isolates of C. difficile from patients whose infections can be defined as community or hospital associated. The collection includes strains isolated over a seven year period (2007-14) from a number of geographical regions.

Genomic samples have been sequenced on Illumina MiSeq at the University of Glasgow Polyomics facility to obtain 300bp long pair-end reads. These were quality trimmed and subsequently mapped against the fully annotated 630 genome. From the mapped alignments the single best alignment for each read was screened and reads discarded that aligned with low identity or to multiple (ambiguous) locations in the reference sequence. In addition, SNPs appearing in tight clusters suggesting recent recombination events were excluded as they could mask the true phylogenetic signal. Variants detected in multiple reads were then combined together to generate a unique sequence of SNPs and indels to which each sequence was compared using a fully automated SNP calling pipeline. Data generated as a consequence were then employed to generate phylogenetic trees. Using PERMANOVA, geographical origins of different strains were found to be significant (p<0.01) at both fine and mesoscopic scale indicating localized transmission. Furthermore, time of isolation also appeared significant suggesting discrimination of evolutionary events maybe possible. Further refinement of this pipeline is ongoing but the findings will inform the design of new infection control interventions for C. difficile in the future.