First Author: David Roach, University of Washington School of Medicine, Class of 2016
Introduction: In the United States, healthcare-associated bacterial infections affect approximately 2 million patients annually and result in nearly 100,000 deaths. To better understand the pathogenesis and transmission of bacterial pathogens, it will be necessary to employ comprehensive analytic approaches able to biologically characterize the organisms and to reconstruct their molecular epidemiology. Whole genome sequencing (WGS) is emerging as a powerful method for these types of analyses, but is not yet used in routine clinical care. WGS studies to date have typically been limited in that they have been 1) retrospective in nature, 2) limited in scope to particular bacterial species, and 3) extended to subsets of bacterial isolates, rather than globally across an entire patient population. Here, we performed prospective collection and WGS of all bacterial isolates collected from three intensive care units within a single hospital over a period of one year. This enabled us to examine many aspects of organisms infecting our patient population, including the identification of previously unreported pathogens and large-scale, longitudinal reconstruction of bacterial transmission events.
Methods: We sequenced 1,260 clinical isolates collected during routine care across three intensive care units over a 12-month period at the University of Washington Medical Center. After assembly, genomes were BLASTed against NCBI databases to establish their identity, and sequence divergence was quantified with Average Nucleotide Identity by BLAST (ANIb) to identify new species (defined as ANIb value <95%). Finally, a single nucleotide variant analysis was performed to unambiguously identify instances of clonal bacteria found between patients.
Results: 35.2% (445/1262) of bacterial isolates qualified as new species at the genomic level. 77 distinct “species clusters” were identified, with 51 consisting entirely of unknown species while 26 were a mixture of known and novel species. New species were isolated from all anatomic culture sites and consisted of both commensal and pathogenic isolates.
There were 116 instances of clonal bacteria found across multiple patients, representing likely bacterial transmission events. 8 different species were involved. 11 separate Staphylococcus epidermidis strains were found across multiple patients, including 3 strains found exclusively in the neonatal intensive care unit (NICU). All of these isolates were associated with intubated infants, and one strain caused sepsis in 5/9 infants from whom it was cultured.
Conclusions: We found a surprisingly high prevalence of new bacterial species infecting ICU patients, many of which are related to known species but some representing entirely novel pathogens whose clinical and biological significance remain unexplored. Additionally, we identified cryptic outbreaks not recognized by hospital infection-control teams, most notably within the NICU. This study demonstrates the power of using WGS routinely in the clinical management of infectious disease, both for improving infection-control practices and for expanding our understanding of bacterial pathogens.