Investigating the increasing incidence of Gram-negative bloodstream infections in England using linked epidemiology, genomic and clinical datasets

The incidence of bloodstream infections (BSIs) caused by Gram-negative bacteria has increased over the past decade with substantial associated morbidity and mortality. Furthermore the association of these bacteria with plasmids and other mobile genetic elements which facilitate rapid dissemination of genes causing antibiotic resistant phenotypes is a cause for global concern. Whilst infection control interventions and antimicrobial stewardship initiatives have had some success in controlling the incidence of other key pathogens causing bloodstream infections, to date there has been little progress to reverse the trends observed for E. coli and Klebsiella spp in particular. In this thesis I combine large-scale whole genome sequencing and electronic healthcare record datasets to try to better understand factors driving the increasing incidence of Gram-negative bloodstream infections with a particular focus on E. coli and Klebsiella spp. In E. coli, I observe a clonal population structure with four dominant sequence types (ST 131/95/73 and 69) comprising 48% of all isolates. In contrast, most Klebsiella spp. isolates belonged to sporadic STs and the most common ST (ST490) only comprised 5% of isolates. I find evidence of a relative increase in the carriage of antimicrobial resistance genes in E. coli whereas the converse was true for Klebsiella spp.. I find that 10-valent O-antigen targeted vaccines would be predicted to provide high levels of coverage against both E. coli (2347/3278, 72%) and K. pneumoniae (541/549, 99%) BSIs. I conduct a genome wide association study to explore genomic factors associated with patient mortality discovering significant associations for three genes (papGII OR 0.94, 95%CI 0.91-0.96, p=5.22x10-7, hin OR 1.23, 95%CI 1.15-1.33, p=3.17x10-6 and a hypothetical protein OR1.34, 95%CI 1.23-1.46, p=5.22x10-7). Finally I utilise recent advances in long read sequencing to characterise the plasmidome associated with these pathogens and the key antimicrobial resistance/virulence genes which they carry. The key finding from this is the similarity between plasmids carrying antimicrobial resistance genes and those with no such genes leading me to introduce the concept of “high risk” plasmid groups. A common recurring theme is the importance and challenge of appropriate study design in molecular epidemiology. Whilst high costs have previously necessitated isolate selection by e.g. antimicrobial phenotype, the work I have conducted in this thesis demonstrates the limitations and dangers of extrapolating from such an approach. I conclude that the linkage of large scale sequencing and clinical datasets affords opportunities to better understand the epidemiology of pathogens, test hypothesis about their transmission and ecology, investigate the likely efficacy of possible intervention strategies and discover new insights into their biology. Global collaboration to develop secure and harmonised methodologies for sharing granular metadata in addition to the vast quantities of sequencing data already available would significantly accelerate future such work, to the benefit of patients across the world.