| dc.description.abstract | Enterococcus faecium is a commensal bacterium of the human gastrointestinal tract that has emerged as an important opportunistic pathogen, particularly in immunocompromised patients. Multi-drug resistant strains, notably vancomycin resistant E. faecium (VREfm), constitute a substantial burden of disease globally. For over a decade, the Republic of Ireland reported some of the highest rates of resistant invasive enterococcal infection in Europe, ranging from 28.4-45.8%. Despite this, little is known about the epidemiology, population structure or transmission dynamics of E. faecium in Irish hospitals. Conventional bacterial typing methods, such as pulsed field gel electrophoresis (PFGE), often lack the discriminatory powers necessary to resolve complex transmission networks. However, in recent years advances in high-throughput whole-genome sequencing (WGS) technology have revolutionised investigations into the evolution, genetic diversity, and transmission of VREfm in healthcare settings. The work described in this thesis utilised high-resolution short- and long-read WGS to comprehensively investigate the VREfm population structure in seven Irish hospitals and the association of endemic, clonal lineages in the context of a protracted, retrospectively identified, silent outbreak.
The first part of this study (Chapter 3) investigated the population biology of E. faecium from seven Irish hospitals (H1-H7) in various geographical locations. Between June 2017 and August 2022, a total of 1,028 E. faecium isolates (894 screening, 102 bloodstream infection (BSI), 12 urine and 20 ‘other’ clinical specimens) from individual patients were investigated, with the majority of isolates (73%, 753/1028) from a large Dublin acute hospital (H1). Short-read WGS and core-genome multi-locus sequence typing (cgMLST) based on 1,423 core loci using the Ridom SeqShere+ software suite (version 8.5.1, Ridom GmbH, Germany) were used to assess epidemiology. A previously published threshold of ≤20 allelic differences was used to infer genetic relationships and to deem isolates as closely related. The majority (979/1028; 95.2%) were vancomycin resistant, of which 963/978 (98.3%) harboured the vanA transposon (vancomycin MIC > 4 mg/L). A small proportion (1.5%, 15/979) of isolates harboured both a plasmid-mediated vanA and a chromosomal vanB. All isolates belonged to hospital-adapted Clade A1, and the majority (732/1028; 72.3%) belonged to multi locus sequence type (MLST) sequence type (ST) 80. The population within major STs was highly polyclonal, with cgMLST further segregating isolates into 65 related isolate clonal complex (CT) groups and 60 singletons with an average cluster size of 15 (median: 5, range: 2-145). The most frequent CTs identified were CT2933 (N=145), CT1916 (N=116), CT1598 (N=64), CT1599 (N=58) and CT1511 (N=58), accounting for 43% (439/1028) of all isolates, with the three major CTs belonging to ST80. Inter and intra-hospital dissemination of highly-related clones was evident over extended time periods (up to 47 months). Additionally, the recovery of closely related (≤20 allelic differences) vancomycin susceptible E. faecium (VSEfm) and VREfm demonstrated the potential for the emergence of novel VREfm clones by the loss/gain of the vanA transposon.
Comparison of the 1,028 Irish isolates with 1,325 Clade A1 international isolates from 30 countries across six continents between 1986-2016 using cgMLST revealed very little overlap. The majority of Irish isolates (98.5%) clustered independently on distinct network branches on a cgMLST-based minimum spanning tree (MST) indicating that local evolution plays a significant role in the epidemiology of VREfm. The extrachromosomal vanA-regions of seven Irish VREfm representing a variety of ST and CTs (SJ10 [ST789, CT1601], SJ11 [ST1478, CT24], BSI_SJ40 [ST80, CT1598], SJ82 [ST203, CT20], SJ245 [ST117, CT2929], RC-19-039 [ST17, CT2934] and RC_19_023 [ST80, CT1916]) were fully resolved by hybrid assembly of short-read and long-read WGS sequences. The vanA regions varied in size and differed from the prototype Tn1546 vanA operon from VREfm BM4147. Differences included multiple orientations of the vanA operon genes, a cadmium efflux accessory protein gene insertion and additional non-coding DNA sequences including repeat regions. The SJ11vanA region was used as a reference for comparison against the corresponding vanA regions of all Irish VREfm (N=979) isolates; 86.3% (845/979) of isolates harboured a vanA region with >90% sequence identity to the SJ11vanA. Six of the vanA regions examined harboured deletions in the tnpA transposase gene, mainly in the 3′ end, the extent of which varied. The seven closed VREfm harboured vanA regions located on a diverse array of circular and linear plasmids. Linear plasmids are a novel finding in E. faecium, and have been previously identified in VREfm from Japan, Switzerland, and India. Linear topology of plasmids was confirmed using S1 nuclease treatment followed by PFGE. Both circular and linear plasmids harbouring vanA were shown to be conjugative by filter mating with a plasmid-free VSEfm recipient strain.
The second part of this study (Chapter 4) investigated the transmission and persistence of VREfm in two critical care wards (WA and WB) of H1 over 39-months using WGS. Rectal screening (N=330, September 2019-December 2022) and environmental (N=48, November 2022-December 2022) E. faecium were investigated using short-read WGS. Isolate relatedness was assessed by cgMLST and core-genome single nucleotide polymorphism (cgSNP) analysis. Likely transmission chains were identified using the SeqTrack algorithm (https://graphsnp.fordelab.com) based on cgSNP data and recovery location. Well-characterized E. faecium (N=908) from seven Irish hospitals including H1 (June 2017-July 2022) were also investigated. In total, 378 E. faecium isolates recovered over 39 months from wards WA and WB of H1 were analysed, including 330 rectal screening isolates (WA, 109 (33%); WB, 221 (67%)). Twenty-eight environmental isolates were recovered from 3/13 handwash basin drains (19 from two WA drains and nine from one WB drain). An additional 20 environmental isolates were recovered from surface sampling of near-patient high-touch ward sites. All patient isolates were vanA-encoding VREfm. The majority (43/48, 89%) of environmental isolates were VSEfm and five were vanA-encoding VREfm. cgMLST identified three predominant ST80 CTs (CT2933, CT2932 and CT1916) (55% of isolates) of related isolates (≤20 allelic differences). cgSNP analysis differentiated these CTs into multiple distinct closely related genomic clusters (≤10 cgSNPs). Parsimonious network construction identified 55 likely inter- and intra-ward transmissions with epidemiological support (≤30 days) between patients involving 73 isolates (≤10 cgSNPs) from seven genomic clusters. Numerous other likely transmissions over longer time periods without evident epidemiological links were identified, suggesting persistence and unidentified reservoirs contribute to dissemination. The three CTs predominated among E. faecium (N=1286) in seven hospitals, highlighting inter-hospital spread and cross-institutional dominance without known epidemiological links. In silico analysis using ABRicate v.0.9.8 (https://github.com/tseemann/abricate) of VSEfm short-reads identified four environmental isolates harbouring vanA operon remnants including the vanZ and vanY genes, but lacking the vanHAX and vanRS genes. One isolate (JHE024) was fully closed by hybrid assembly, which identified vanZ and vanY on an 88,110 bp circular plasmid. The organization of the van remnant and adjacent sequences were similar to vanA regions on plasmids in Irish VREfm, indicating the VSEfm may have arisen from a VREfm ancestor. These findings highlight that VSEfm carriage should be considered in Infection Prevention and Control (IPC) surveillance policies.
The final part of this study (Chapter 5) hypothesised that within-patient diversity and the presence of multiple VREfm and VSEfm co-colonising strains could complicate transmission analysis and explain the diverse array of Tn1546-like vanA transposon iterations found in Chapter 3. A total of 150 VREfm colonies were sequenced from rectal screens from 10 patients (15 isolates per patient, P1-P10) on seven H1 wards (W1 N=1, W2 N=1, W3 N=4, W9 N=1, W18 N=1, W20 N=1, WE N=1). The majority (70%, 7/10) of patients harboured a diverse colonising population, consisting of two (N=4), three (N=2) or four (N=1) unique VREfm CTs. A high level of genomic variation was observed between isolate pairs from different CTs within the same patient using split k-mer analysis suggesting distinct VREfm acquisition events. The cgMLST-based MST revealed three CTs (CT2933, CT6489 and CT6738) each consisting of closely related (<20 allelic differences) VREfm from multiple patients recovered between October 2022-Janurary 2023 (105 days) suggesting possible direct and/or indirect transmission. Further analysis of the VREfm (N=150) with an additional 622 isolates from Chapter 3 from individual patients and environmental sites (VREfm N=579, VSEfm N=43) from the same seven H1 wards (W1 N=107, W2 N=156, W3 N=327, W9 N=7, W18 N=22 and W20 N=3), provided further evidence for the persistence of highly-related minority subpopulations across multiple H1 wards. These findings highlight the limitations of single-colony sampling strategies used currently in outbreak investigations.
Plasmid sequences from a representative subset of 39 rectal screening VREfm from the 10 patients (P1 N=6, P2 N=2, P3 N=3, P4 N=4, P5 N=5, P6 N=4, P7 N=4, P8 N=7, P9 N=2 and P10 N=2) were fully resolved by hybrid assembly of short-read and long-read WGS sequences. A total of 193 plasmids were identified in the 39 VREfm. Pairwise plasmid genetic relatedness (N=193) was assessed using a reference-free approach based on Mash distances (https://github.com/marbl/Mash). Using a mash threshold of ≤0.0001, equivalent to ≥99.5% average nucleotide identity, near-identical plasmids were identified in genetically divergent E. faecium subtypes, both within the same patient and multiple patients. Among the 39 within-patient VREfm investigated in detail from patients P1-P10, 72% (28/39) harboured vanA on linear plasmids belonging to a discrete cluster, suggesting acquisition of vanA is an important factor driving the evolution of linear plasmids in Irish VREfm. Examination of the structural organisation of the van operon from the 39 VREfm revealed the presence of identical vanA regions (≥99.9% average nucleotide identity) in clonally divergent CTs within the same patient. There were also many examples of distinct iterations of the van transposon carried by multiple diverse CTs within the same patient. These findings serve to highlight the promiscuous nature of the Tn1546-like vanA transposon and its ability to utilise a variety of plasmid types to facilitate dissemination.
Finally, four of the VREfm-positive rectal swabs (patients P1, P3, P5 and P8) were selected to provide a snap-shot of VSEfm carriage in the patient cohort. Thirty VSEfm were identified among 80 colonies screened. All 30 VSEfm belonged to Clade A1, the majority belonged to ST80 (25/30) with the remainder belonging to ST203. An MST based on cgMLST analysis of the study VREfm (N=150) and VSEfm (N=30) revealed two distinct CT clusters (≤20 allelic differences) comprising mixtures of VSEfm and VREfm from the same patient. Hybrid assembly of four representative VSEfm from distinct CTs revealed highly similar plasmidomes compared to VREfm. However, in some cases, genetically closely related VSEfm and VREfm from the same patient exhibited plasmidome differences.
In conclusion, this study constitutes the first comprehensive investigation of the epidemiology and population structure of clade A1 E. faecium from hospital patients in the Republic of Ireland using a WGS approach. The study revealed that ST80 has emerged as the predominant endemic E. faecium lineage in Ireland. A diverse and polyclonal population was identified within the major STs including ST80, which were segregated further into genetically distinct sub-lineages using cgSNP or SKA analysis. Irish VREfm were found to harbour multiple distinct iterations of the vanA TN1546-like transposon enriched with IS1216E elements which likely facilitate spread. Large linear plasmids were found to be the major vehicle of vanA dissemination in Irish VREfm. The diversity of VREfm clones identified and their widespread intra- and inter-hospital dissemination, highlight the need for improved surveillance and review of national IPC guidelines. The extensive within-patient VREfm diversity identified warrants the urgent development of novel screening tests to rapidly identify co-colonized patients to inform outbreak investigations. | en |
