Genomic insights into carbapenem-resistant Klebsiella pneumoniae transmission and adaptation in the healthcare environment

Seminar Details
Monday, April 26, 2021 - 12:00pm to 1:00pm


Zena Lapp,
Bioinformatics PhD student, Snitkin Lab, Microbiology and Immunology


Advisor:  Evan Snitkin

Multidrug resistant organisms (MDROs) pose a threat to healthcare facilities worldwide due to global prevalence, limited treatment options, and high mortality rates. This dissertation acts as a proof-of-principle study for using whole-genome sequencing and associated clinical metadata to provide actionable insights into MDRO infection prevention and control practices. We focus our analysis on carbapenem-resistant Klebsiella pneumoniae (CRKP) sequence type (ST) 258, an MDRO that is particularly prevalent in long-term acute care hospitals (LTACHs) in the United States (US). Using a comprehensive set of 417 clinical CRKP ST258 isolates collected over the course of a year in 21 US LTACHs, we investigate regional transmission, predictors of infection, and evolution of antibiotic resistance. In addition, we develop three open-source R packages that implement methods developed and applied here: regentrans for studying regional pathogen transmission, mikropml for performing machine learning, and prewas for preprocessing data prior to bacterial genome-wide association studies.

First, we reconstructed regional transmission pathways with genomic data and analyzed this network in the context of patient transfer data and patient-level clinical data to identify potential drivers of regional CRKP transmission. We found high regional CRKP burdens in Los Angeles area LTACHs that were due to a small number of introductions with subsequent proliferation occurring via within-facility transmission and patient transfers among healthcare facilities.

As only a subset of colonized patients develop clinical infection, we next used machine learning to determine whether patient characteristics and CRKP genetic background can predict infection status. We found that patient and genomic features were predictive of clinical CRKP infection to similar extents. Genomic predictors of infection included presence of the ICEKp10 mobile genetic element carrying the yersiniabactin iron acquisition system and disruption of the O-antigen biosynthetic gene kfoC in a CRKP ST258 sublineage (clade IIB). Disrupted kfoC was associated with isolation from the respiratory tract, and subsequent ICEKp10 acquisition was associated with increased virulence. These results highlight the utility of machine learning to provide insight into patient clinical trajectories and ongoing within-lineage pathogen adaptation.

Finally, we investigated the evolution of CRKP resistant to the antibiotic colistin, one of the few remaining treatment options for this MDRO. Two large clusters of resistant strains in clade IIB accounted for over half of the detected colistin resistance, in stark contrast to the sporadic resistance events observed in other clades. Moreover, while resistant isolates from other clades were less fit than susceptible non-revertant isolates, clade IIB resistant isolates were more fit, underscoring the potential for continued regional spread of clade IIB colistin resistant strains.

In summary, we identified an emerging CRKP sublineage that has spread across Los Angeles area LTACHs and appears to have an increased affinity for the respiratory tract, increased transmissibility, and a decreased fitness cost of colistin resistance. Future work should continue to monitor this strain as increased prevalence could reduce the efficacy of colistin as a treatment for CRKP in this region and possibly lead to exportation to other regions. Additionally, these findings highlight the potential impact of incorporating genomic epidemiology into regional infection prevention efforts to identify high-transmission facilities and emerging strains, thereby facilitating the containment of MDROs to the greatest extent possible.