ReportsPATHOGEN TRANSMISSION

Integrated genomic and interfacility patient-transfer data reveal the transmission pathways of multidrug-resistant Klebsiella pneumoniae in a regional outbreak

See allHide authors and affiliations

Science Translational Medicine  22 Nov 2017:
Vol. 9, Issue 417, eaan0093
DOI: 10.1126/scitranslmed.aan0093
  • Fig. 1. Ancestral reconstruction of the 2008 regional carbapenem-resistant K. pneumoniae outbreak.

    Phylogenetic and ancestral facility reconstruction was performed in BEAST using variants, indels, health care facility locations, and dates of isolation for 41 isolates from 31 patients. Colors of branches correspond to the most probable ancestral facility, with the branch thickness indicating the relative probability as compared to all other considered facilities. Numbers on branches correspond to nine predicted interfacility transmission events based on the intermixing of isolates from different facilities on the tree. Branches labeled A, B, and C highlight the early branching of outbreak isolates into subclades that have varying concordance with the patient-sharing network.

  • Fig. 2. A subset of patient transfers explains the genomic transmission network.

    The patient-transfer network from (17) was reconstructed. Blue circles that contain numbers represent patients, gray nodes that contain letters represent facilities, and directed edges represent patients coming from or going to a given facility. Patient-transfer events consistent with the genomic transmission network (red edges) were identified by searching for common facility exposures among patients in the subclades flanking each interfacility transmission event on the tree (see Materials and Methods and table S1). Numbers on the edges correspond to the branch labels from Fig. 1 and are only shown for the six genomic linkages that can be explained by specific patient-transfer events. Note that although a single patient-sharing explanation is displayed, in many cases, more than one transfer event could explain a genetic linkage, because patients were often transferred among common sets of facilities (for example, an acute care hospital, a nursing home, and a long-term acute care hospital) as their clinical care needs changed.

  • Fig. 3. Comparison between genomic and patient-transfer networks.

    (A) The number of interfacility linkages observed in the patient-transfer network (left) and genomic transmission network (right) is shown as undirected networks. For the genomic network, the number of predicted intrafacility transmissions is also indicated on loops circling back to each facility label. Genomic transmissions were extracted from the phylogenetic reconstruction that included the set of surveillance isolates from NH-B (fig. S4). Note that the patient-sharing network does not consider the additional set of surveillance isolates from patients from NH-B and LTACH-A, because the patients’ history of health care exposures was not known. (B) The number of connections between each pair of facilities from the patient transfer and genomic networks in (A) is compared in a scatterplot, with each point representing a pair of facilities. Points on the plot are jittered, such that overlapping points can be distinguished. The number of interfacility connections in the two networks was compared using a Mantel test (P = 0.027).

  • Fig. 4. Comparison of real-time and retrospective genomic predictions of patients’ isolate origins.

    Genomic predictions for the origin of each patient’s isolate were compared by using the complete data set and real-time subsets of the data. Real-time data subsets consisted of only isolates that predated a patient’s isolate for which a prediction was being made. Patients are ordered from left to right by the date on which the specimen from which their isolate was grown was collected. The top color bar indicates the facility where the patient resided when the culture from which their carbapenem-resistant K. pneumoniae isolate was collected, the middle color bar indicates the predicted source facility by using only real-time data, and the bottom color bar indicates the predicted source facility when all data were considered. Differences from the top color bar indicate a predicted importation, and conversely, identities indicate a predicted intrafacility transmission. Note that the prediction for the first isolate is trivial, because there was no preceding isolate to link it to.

Supplementary Materials

  • www.sciencetranslationalmedicine.org/cgi/content/full/9/417/eaan0093/DC1

    Fig. S1. Epidemic curve of outbreak across regional health care facilities.

    Fig. S2. Maximum likelihood tree of outbreak isolates.

    Fig. S3. Distribution of overlaps between genomic and patient-sharing networks for randomized data.

    Fig. S4. Ancestral facility reconstruction for collection including surveillance isolates from NH-B and LTACH-A.

    Fig. S5. Variants identified among outbreak isolates.

    Fig. S6. Correlation between date of isolation and genetic distance from common ancestor.

    Fig. S7. Maximum likelihood ancestral reconstruction of outbreak isolates.

    Table S1. Patient-transfer events associated with genomically inferred interfacility transmissions.

    Table S2. Summary of whole-genome sequencing in the current study.

  • Supplementary Material for:

    Integrated genomic and interfacility patient-transfer data reveal the transmission pathways of multidrug-resistant Klebsiella pneumoniae in a regional outbreak

    Evan S. Snitkin,* Sarah Won, Ali Pirani, Zena Lapp, Robert A. Weinstein, Karen Lolans, Mary K. Hayden*

    *Corresponding author. Email: esnitkin{at}umich.edu (E.S.S.); mhayden{at}rush.edu (M.K.H.)

    Published 22 November 2017, Sci. Transl. Med. 9, eaan0093 (2017)
    DOI: 10.1126/scitranslmed.aan0093

    This PDF file includes:

    • Fig. S1. Epidemic curve of outbreak across regional health care facilities.
    • Fig. S2. Maximum likelihood tree of outbreak isolates.
    • Fig. S3. Distribution of overlaps between genomic and patient-sharing networks for randomized data.
    • Fig. S4. Ancestral facility reconstruction for collection including surveillance isolates from NH-B and LTACH-A.
    • Fig. S5. Variants identified among outbreak isolates.
    • Fig. S6. Correlation between date of isolation and genetic distance from common ancestor.
    • Fig. S7. Maximum likelihood ancestral reconstruction of outbreak isolates.
    • Table S1. Patient-transfer events associated with genomically inferred interfacility transmissions.
    • Table S2. Summary of whole-genome sequencing in the current study.

    [Download PDF]

Stay Connected to Science Translational Medicine

Navigate This Article