Epidemiology, E. coli bacteraemia, genomic analysis, whole genome sequencing, antimicrobial resistance, multi-drug resistance, phylogenetic analysis, surveillance
Background:
Surveillance of E. coli from blood cultures shows a steady increase in the rate of E. coli bacteraemia (ECB) in Wales from 46 in 2003 to 74 per 100,000 population in 2013.
Despite the significant morbidity and mortality associated with ECB, the strains responsible for this increase are poorly characterised, with limited phenotypic information available.
Aim:
Public Health Wales are collaborating with the Cardiff School of Biosciences to identify temporal, geographic and genomic risk factors by whole genome sequencing (WGS). In combination with a case series analysis, the sequence data and linked risk factor data will identify ECB risk profiles.
Methods:
Population-based prospective WGS study. Microbiology laboratories in Wales submitted all non-duplicate isolates of E. coli from blood cultures to the Specialist Antimicrobial Chemotherapy Unit (SACU) between April 2013 and March 2014 for WGS by Cardiff University.
The dataset was linked to routine microbiological surveillance data to obtain isolate antimicrobial resistance (AMR) profiles. Novel AMR profiles and profiles with phylogeography were characterised by polymerase chain reaction. Selected samples were WGS based on their determined phylogenetic groups. Additionally, all isolates from the case series analysis (Apr – June 2014) will be sequenced.
Results:
The year-long study generated 720 bacterial genomes for genomic analysis. The population structure is dominated by four key lineages, which are responsible for a disproportionate number of ECB episodes; MLST Sequence Types 131, 69, 95, and 73. These sequence types are all distinct in terms of their virulence and antimicrobial resistance profiles, and have previously been identified as the principal causes of ECB in other parts of the UK. Additionally there is limited evidence of phylogeography within the dataset. Collectively, the study depicts antimicrobial resistance as the predominant characteristic of ECB.
1. Sequence Type distribution
To characterise the population structure of E. coli bacteraemia in Wales, we identified the distribution of MLST sequence types found within our cohort. This analysis reveals that four key lineages that have historically been associated with bacteraemia (STs 131, 73, 95 and 69) account for 62% of samples analysed.
We also observe a number of other lineages that have consistently caused disease at a low frequency during the study. 9% of cases were due to individual strains from lineages that occurred once in the dataset.
2. AMR gene frequency
Examining the presence/absence of genes associated with antimicrobial resistance across our dataset reveals that there are two distinct patterns of resistance amongst the four lineages. Isolates of ST131 and ST69 show an increased carriage of genes associated with resistance, in comparison to ST73 and to ST95 which show generally lower levels of resistance-associated gene carriage.
3. Phylogenetic analysis:
For each of the lineages we are working to identify any evidence of phylogeography. Genomics provides us with a high-resolution view, to identify the likely origin of samples within the study dataset, and reveal if there is evidence of local transmission. The figure below shows the phylogenetic relationships of the 59 ST131 isolates sequenced from 4 hospitals over the course of a year, alongside a global collection of ST131 isolates from sequence databases. Combining the sequence data generated with data from public databases reveals a lack of phylogeography within our key lineages; a finding that implies that the increase in E. coli bacteraemia within Wales is largely due to a general, population-level increase, rather than a specific, local, increase of one particular clone.
Conclusion:
This high-resolution data, together with other root cause analysis results, will provide a clear target for developing and deploying interventions to halt the rise of ECB in Wales.
Limitations:
The resistance values are predicted based upon the presence/absence of genes found within a resistance database currently maintained locally. This database includes 1693 genes, but it is likely there are resistance genes in existence that are not encompassed in this library. This analysis also assumes that the presence of a gene indicates resistance; this also may not be true and would require phenotypic testing to validate.