Research ArticleSepsis

Robust classification of bacterial and viral infections via integrated host gene expression diagnostics

See allHide authors and affiliations

Science Translational Medicine  06 Jul 2016:
Vol. 8, Issue 346, pp. 346ra91
DOI: 10.1126/scitranslmed.aaf7165
  • Fig. 1. Summary ROC curves for discovery and direct validation data sets for the bacterial/viral metascore.

    Summary ROC curve is shown in black, with 95% CIs in dark gray.

  • Fig. 2. Bacterial/viral score in COCONUT-conormalized whole-blood validation data sets.

    The global AUC across all whole-blood discovery data sets is 0.93. Top: Score distribution by data set (blue, bacterial; red, viral). Middle: Individual gene expression (exp.). Bottom: Housekeeping genes (grayscale). The dotted line at the top shows a possible global threshold for discriminating infection type.

  • Fig. 3. IADM across COCONUT-conormalized public gene expression data that matched inclusion criteria.

    (A) IADM schematic. (B) Distribution of scores and cutoffs for IADM in COCONUT-conormalized data. SIRS, systemic inflammatory response syndrome. (C) Confusion matrix for diagnosis. Bacterial infection sensitivity, 94.0%; bacterial infection specificity, 59.8%; viral infection sensitivity, 53.0%; viral infection specificity, 90.6%.

  • Fig. 4. Targeted NanoString gene expression data for children with SIRS/sepsis from the GPSSSI cohort never tested with microarrays.

    Total n = 96, of which SIRS = 36, bacterial sepsis = 49, and viral sepsis = 11. (A) Breakdown of infected patients by organism type. (B and C) ROC curves for the SMS and the bacterial/viral metascore. (D) Distribution of scores and cutoffs for IADM. (E) Confusion matrix for IADM. Bacterial infection sensitivity, 89.7%; bacterial infection specificity, 70.0%; viral infection sensitivity, 54.5%; viral infection specificity, 96.5%.

  • Table 1. Data sets used in the discovery and direct validation of the bacterial/viral metascore.

    CAP, community-acquired pneumonia; HHV6, human herpesvirus 6; RSV, respiratory syncytial virus; HSV, herpes simplex virus; CMV, cytomegalovirus; MPV, metapneumovirus; PICU, pediatric intensive care unit; LRTI, lower respiratory tract infection; ARI, acute respiratory infection; COPD, chronic obstructive pulmonary disease.

    AccessionAuthorTissuePlatformDemographicBacteriaVirusesNumber
    healthy
    Number
    bacterial
    Number
    viral
    A. Discovery data sets
    GSE6269RamiloPBMCGPL2507Children
    admitted
    with
    infection
    Escherichia coli,
    Staphylococcus aureus,
    Streptococcus
    pneumoniae
    Influenza0168
    GPL570S. aureus,
    S. pneumoniae
    Influenza01210
    GPL96S. aureus,
    S. pneumoniae
    Influenza67318
    GSE20346ParnellWhole bloodGPL6947Adults with
    CAP
    Unknown
    bacterial
    pneumonia
    Influenza36128
    GSE40012ParnellWhole bloodGPL6947Adults with
    CAP
    Unknown bacterial
    pneumonia
    Influenza183611
    GSE40396HuWhole bloodGPL10558Febrile
    children in
    emergency
    department
    MultipleAdenovirus,
    enterovirus,
    rhinovirus,
    HHV6
    22835
    GSE42026HerbegWhole bloodGPL6947Children admitted
    with infection
    Streptococcus and
    Staphylococcus spp.
    Influenza, RSV331841
    GSE66099WongWhole bloodGPL570Septic children
    in PICU
    MultipleInfluenza, HSV,
    CMV, BK,
    adenovirus
    4710911
    B. Validation data sets
    GSE15297PopperWhole bloodGPL8328Febrile
    children
    Scarlet fever
    (Streptococcus)
    Adenovirus058
    GSE25504SmithWhole bloodGPL13667Septic
    neonates
    MultipleRhinovirus,
    CMV
    6113
    GPL6947MultipleCMV35261
    GSE60244SuarezWhole bloodGPL10558Adults hospitalized
    with LRTI
    Gram-positive and atypicalInfluenza, RSV, MPV402271
    GSE63990TsalikWhole bloodGPL571Adults with ARIMultipleMultiple070115
    E-MEXP-3589AlmansaWhole bloodGPL10332Adults with
    COPD with
    infection
    Gram-positive,
    Gram-negative,
    atypical
    Influenza, RSV, MPV445
  • Table 2. Validation data sets that matched inclusion criteria and have a single known pathogen type (viral or bacterial).

    DHF, dengue hemorrhagic fever; DSS, dengue shock syndrome.

    AccessionAuthorTissuePlatformDemographicSpecific
    pathogens
    Number
    healthy
    Number
    bacterial
    Number
    viral
    E-MEXP-3567IrwinWhole
    blood
    GPL96Malawian children with
    bacterial meningitis or
    pneumonia
    S. pneumoniae,
    Neisseria meningitidis,
    or Haemophilus
    influenzae
    3120
    GSE11755EmontsWhole
    blood
    GPL570Children in PICU with
    meningococcal sepsis
    N. meningitidis360
    GSE13015PanklaWhole
    blood
    GPL6106Adults with bacterial sepsisBurkholderia pseudomallei
    and others
    10450
    GPL694710150
    GSE22098BerryWhole
    blood
    GPL6947Children with Gram-positive
    infections
    Staphylococcus and
    Streptococcus
    81520
    GSE28750SutherlandWhole
    blood
    GPL570Adults with community-acquired
    bacterial sepsis
    Multiple bacteria20100
    GSE29161ThunyWhole
    blood
    GPL6480Adults with native valve-infected
    endocarditis
    Staphylococcus and
    Streptococcus
    550
    GSE33341AhnWhole
    blood
    GPl571Adults with septic
    bloodstream infections
    S. aureus or E. coli43510
    GSE40586LillWhole
    blood
    GPL6244Bacterial meningitisMultiple bacteria18210
    GSE42834BloomWhole
    blood
    GPL10558Bacterial pneumoniaUnknown118190
    GSE57065CazalisWhole
    blood
    GPL570Adults with bacterial
    septic shock
    Multiple bacteria25820
    GSE69528ConejeroWhole
    blood
    GPL10558Adults with
    bacterial sepsis
    B. pseudomallei and others55830
    E-MTAB-3162van de WegWhole
    blood
    GPL570Indonesian patients >14 years old with
    uncomplicated and severe dengue
    Dengue15030
    GSE17156ZaasWhole
    blood
    GPL571Volunteers with viral
    challenge peak symptoms
    Influenza, RSV, rhinovirus56027
    GSE21802Bermejo-MartinWhole
    blood
    GPL6102Adults with
    septic influenza
    Influenza (H1N1)4012
    GSE27131BerdalWhole
    blood
    GPL6244Adults with septic influenza with
    mechanical ventilation
    Influenza (H1N1)707
    GSE38900MejiasWhole
    blood
    GPL10558Children with acute LRTIRSV8028
    GPL6884Influenza, RSV, rhinovirus310153
    GSE51808KwissaWhole
    blood
    GPL13158Children and adults with
    uncomplicated dengue
    and DHF
    Dengue9028
    GSE68310ZhaiWhole
    blood
    GPL10558Adults with ARIsMostly influenza
    and rhinovirus
    2430211
    GSE16129ArduraPBMCGPL6106Children with invasive
    staph infections
    S. aureus990
    GPL9610460
    GSE23140LiuPBMCGPL6254Children with acute otitis mediaS. pneumoniae440
    GSE34205IoannidisPBMCGPL570Infants and children with ARIsInfluenza, RSV22079
    GSE38246PopperPBMCGPL15615Nicaraguan children with un
    complicated dengue, DHF, and DSS
    Dengue8095
    GSE69606BrandPBMCGPL570Children with mild-to-severe RSVRSV17026

Supplementary Materials

  • www.sciencetranslationalmedicine.org/cgi/content/full/8/346/346ra91/DC1

    Fig. S1. The SMS and pathogen type.

    Fig. S2. Study schematic.

    Fig. S3. Forest plots of the seven-gene set.

    Fig. S4. Summary ROC forest plots for discovery data.

    Fig. S5. Summary ROC forest plots for direct validation data.

    Fig. S6. Bacterial/viral metascore ROC in GSE53166.

    Fig. S7. Schematic of COCONUT conormalization.

    Fig. S8. COCONUT conormalization of whole-blood discovery data sets.

    Fig. S9. Bacterial/viral score in global ROC of non-conormalized whole-blood discovery data sets.

    Fig. S10. Bacterial/viral score in global ROC of COCONUT-conormalized whole-blood discovery data sets.

    Fig. S11. Bacterial/viral score in global ROC of non-conormalized whole-blood validation data sets.

    Fig. S12. Bacterial/viral score in global ROC of non-conormalized PBMC validation data sets.

    Fig. S13. Bacterial/viral score in global ROC of COCONUT-conormalized PBMC validation data sets.

    Fig. S14. The effects of age on SMS in COCONUT-conormalized data.

    Fig. S15. SMS across all COCONUT-conormalized whole-blood data (both discovery and validation).

    Fig. S16. IADM across COCONUT-conormalized public gene expression data including healthy controls.

    Fig. S17. NPV and PPV for the IADM.

    Fig. S18. GSE63990, adults with ARIs.

    Table S1. Significant gene list.

    Table S2. Test characteristics of the bacterial/viral metascore in direct validation data sets.

    Table S3. Data sets with noninfected inflammatory conditions used to test the IADM.

    Table S4. NanoString data.

    Acknowledgments

  • Supplementary Material for:

    Robust classification of bacterial and viral infections via integrated host gene expression diagnostics

    Timothy E. Sweeney,* Hector R. Wong, Purvesh Khatri*

    *Corresponding author. Email: tes17{at}stanford.edu (T.E.S.); pkhatri{at}stanford.edu (P.K.)

    Published 06 July 2016, Sci. Transl. Med. 8, 346ra91 (2016)
    DOI: 10.1126/scitranslmed.aaf7165

    This PDF file includes:

    • Fig. S1. The SMS and pathogen type.
    • Fig. S2. Study schematic.
    • Fig. S3. Forest plots of the seven-gene set.
    • Fig. S4. Summary ROC forest plots for discovery data.
    • Fig. S5. Summary ROC forest plots for direct validation data.
    • Fig. S6. Bacterial/viral metascore ROC in GSE53166.
    • Fig. S7. Schematic of COCONUT conormalization.
    • Fig. S8. COCONUT conormalization of whole-blood discovery data sets.
    • Fig. S9. Bacterial/viral score in global ROC of non-conormalized whole-blood discovery data sets.
    • Fig. S10. Bacterial/viral score in global ROC of COCONUT-conormalized whole-blood discovery data sets.
    • Fig. S11. Bacterial/viral score in global ROC of non-conormalized whole-blood validation data sets.
    • Fig. S12. Bacterial/viral score in global ROC of non-conormalized PBMC validation data sets.
    • Fig. S13. Bacterial/viral score in global ROC of COCONUT-conormalized PBMC validation data sets.
    • Fig. S14. The effects of age on SMS in COCONUT-conormalized data.
    • Fig. S15. SMS across all COCONUT-conormalized whole-blood data (both discovery and validation).
    • Fig. S16. IADM across COCONUT-conormalized public gene expression data including healthy controls.
    • Fig. S17. NPV and PPV for the IADM.
    • Fig. S18. GSE63990, adults with ARIs.
    • Table S1. Significant gene list.
    • Table S2. Test characteristics of the bacterial/viral metascore in direct validation data sets.
    • Table S3. Data sets with noninfected inflammatory conditions used to test the IADM.
    • Legend for table S4
    • Acknowledgments

    [Download PDF]

    Other Supplementary Material for this manuscript includes the following:

    • Table S4 (.csv format). NanoString data.

Navigate This Article