Threshold-free genomic cluster detection to track transmission pathways in healthcare settings

Abstract

AbstractBackgroundOver the past decade, whole-genome sequencing (WGS) has become the gold standard for tracking the spread of infections in healthcare settings. However, a critical barrier to the routine application of WGS for infection prevention is the lack of reliable criteria for determining if a genomic linkage is consistent with transmission.MethodsHere, we sought to understand the genomic landscape in a high-transmission healthcare setting by performing WGS on 435 carbapenem-resistant Enterobacterales (CRE) isolates collected from 256 patients through admission and biweekly surveillance culturing of virtually every hospitalized patient over a 1-year period.FindingsOur analysis revealed that the standard approach of employing a single-nucleotide variant (SNV) threshold to define transmission would lead to both false-positive and false-negative inferences. False positive inferences were driven by the frequent importation of closely related strains, which were presumably linked via transmission at a connected healthcare facility. False negative inferences stemmed from the diversity of colonizing populations being spread among patients, with multiple examples of hypermutator strains emerging within patients and leading to putative transmission links separated by large genetic distances. Motivated by limitations of an SNV threshold, we implemented a novel threshold-free transmission cluster inference approach whereby each of the 234 acquired CRE isolates were linked back to the imported CRE isolate with which it shared the most variants. This approach yielded clusters that varied in levels of genetic diversity but were highly enriched in patients sharing epidemiologic links. Holistic examination of clusters highlighted extensive variation in the magnitude of onward transmission stemming from the more than 100 importation events and revealed patterns in cluster propagation that could inform improvements to infection prevention strategies.InterpretationOverall, our results show how the integration of culture surveillance data into genomic analyses can overcome limitations of cluster detection based on SNV-thresholds and improve our ability to track pathways of pathogen transmission in healthcare settings.FundingCDC U54 CK000481, CDC U54 CK00016 04S2. S.E.H was supported by the University of Michigan NIH Training Program in Translational Research T32-GM113900 and the University of Michigan Rackham pre-doctoral fellowship.Research in contextEvidence before this studyWe searched PubMed for studies published before May 1, 2021, with no start date restriction, with the search “transmission AND whole-genome AND (snp OR snv) AND (cut-off OR threshold) NOT (SARS-CoV-2 OR virus or HIV)”. We identified 18 reports that used whole genome sequencing to study transmission, primarily in healthcare settings. Several of these studies attempted to identify optimal single nucleotide variant (SNV) cutoffs for delineating transmission. These studies were all single-site and had only partial sampling of healthcare facilities. Moreover, even when the same species was considered, different optimal SNV thresholds were reported.Added value of this studyTo understand the limitations of an SNV threshold approach for tracking transmission we leveraged a data set that comprised admission and every-other-week CRE surveillance culturing for every patient entering a hospital over the course of one year. By performing genomic analysis of 435 isolates from the 256 CRE colonized patients we systematically demonstrated pitfalls with the use of SNV thresholds for transmission inference that stem from the importation of closely related strains from connected healthcare facilities, variation in genetic heterogeneity of colonizing populations and uneven evolutionary rates of CRE strains colonizing patients. We went on to implement an alternative approach for tracking transmission in healthcare facilities that relies on genetic context, instead of genetic distance to group patients into intra-facility transmission clusters. We applied this approach to our CRE genomes and demonstrated that the resultant transmission clusters are strongly enriched in patients with spatiotemporal overlap, and that clusters can be interrogated to identify putative targets to interrupt transmission.Implications of all the available evidenceAdvances in the speed and economy of genome sequencing are making it increasingly feasible to perform routine sequencing to track transmission in healthcare settings. However, a critical barrier to these efforts is the lack of clear criteria for inferring transmission that generalizes to diverse strains of healthcare pathogens and that are robust to variation in organism prevalence and differences in connectivity of local healthcare networks. Here, we show that by combining genome sequencing with surveillance data that healthcare transmission can be inferred in a threshold-free manner. The success of this approach in a setting with high importation and transmission rates bodes well for its generalizability to less challenging healthcare settings.