Professional Documents
Culture Documents
Sepsis
a, b
Steven M. Opal, MD *, Xavier Wittebole, MD
KEYWORDS
Rapid Culture-independent Microbial diagnostic techniques
Host-derived biomarkers for early diagnosis of sepsis and septic shock
Machine-learning systems to identify susceptible patients with sepsis
Rapid genotypic and phenotypic methods to measure antimicrobial
susceptibility patterns
KEY POINTS
Promising molecular diagnostic techniques are under development, but not yet displaced
the traditional rather insensitive, and slow microbial diagnostic measures in use today.
Identifying a pathogen and determining its susceptibility to antimicrobial agents generally
takes more than 24 hours, forcing clinicians to give empiric, broad-spectrum, antimicro-
bial agents to cover the likely pathogen(s).
Clear evidence exists that correct administration of antimicrobial agents can be life-saving
in patients with sepsis.
Two major strategies are in development to assist treating physicians to make expeditious
decisions on appropriate antibiotic choices that specifically targets the invading patho-
gens in the septic patient.
This article reviews the current status of rapid molecular markers to optimally define which
patients need urgent treatment.
INTRODUCTION
Intensive care unit (ICU) practitioners and emergency department clinicians are in
urgent need of a major upgrade in their access to accurate and fast microbial diag-
nosis (Table 1). The diagnostic techniques for bloodstream infection available to
most ICU clinicians today do not fundamentally differ from Robert Koch’s laboratory
methods circa 1887.1,2 That was the year when 2 co-workers in Koch’s laboratory,
Fanny Hess and Richard Petri, first suggested using semisolid agar and 2 glass plates
of slightly different size for isolation of bacteria in pure culture.
a
Infectious Disease Division, Alpert Medical School of Brown University, Ocean State Clinical
Coordinating Center at Rhode Island Hospital, 1 Virginia Avenue Suite 105, Providence, RI
02905, USA; b Critical Care Department, (Pr Laterre), Cliniques Universitaires Saint-Luc, Avenue
Hippocrate 10, 1200 Brussels, Belgium
* Corresponding author.
E-mail address: Steven_Opal@brown.edu
Table 1
Rapid detection methods and molecular biomarkers to improve the diagnosis of sepsis
Abbreviations: DEGs, differentially expressed genes; LDLR, low-density lipoprotein receptor; LPS,
lipopolysaccharide; PCSK9, proprotein convertase subtilisin kexin type 9; SNP, single nucleotide
polymorphism.
Most clinical microbiology laboratories still use the same basic method of single
colony isolation before conducting antimicrobial susceptibility testing. The drug sensi-
tivity results are usually not available for 24 to 72 hours. However, it should be noted
that current treatment guidelines for treating sepsis promulgate initiation of effective
antibiotics within 1 to 3 hours!3–5 Although clinicians at large medical centers might
now have access to early microbial identification via innovations in mass spectrometry
(MS) and nuclear magnetic resonance methods, rarely do they have antimicrobial sus-
ceptibility results at the critical moment of ordering the antibiotic treatment regimen.
Antibiotic therapy is therefore empiric by nature and purposely given as broad-
spectrum treatment to avoid missing the drug susceptibility profile of the causative
pathogen.
Compelling but not uniformly accepted evidence exists in support of
urgent administration of empiric antibiotics in patients with severe sepsis of septic
shock.5–9 Ideally, antibiotic intervention should be completed within 1 hour of
initial diagnosis of sepsis or septic shock.6–8 Living up to this 1-hour imperative
is hampered by the fact that there is no single, rapid, and accurate diagnostic
test for the syndrome of sepsis or septic shock. The diagnosis rests on a
diagnosis of severe infection accompanied by a constellation of signs and
symptoms, blood studies, and pathophysiologic findings clinically recognized as
sepsis.10
Biomarkers of Infection and Sepsis 3
Advances in molecular diagnosis without the need for traditional, time-consuming cul-
ture methods is finally reaching the stage for actual clinical usage. This advance could
revolutionize the clinical care and treatment decisions of critically ill, patients with
sepsis. Knowing the molecular identity of the causative organism, and perhaps
even the susceptibility data, in a few hours rather than 24 to 72 hours could allow
the ICU physician to make more informed decisions on antibiotic choices.16–23 Most
of these rapid diagnostic methodologies are based on the polymerase chain reaction
(PCR) to amplify the microbial DNA signal. This is followed by various analytical ap-
proaches to determine the genomic sequence identity by MS16–23 or nuclear magnetic
resonance.14,24 The most common system currently used in clinical microbiology lab-
oratory at present is matrix assisted laser desorption/ionization time of flight mass
spectroscopy. Many studies report reduced mortality rates in comparative studies
with traditional microbiologic diagnostic measurements.16,19,21,22 Fast detection of
specific pathogens such as Staphylococcus aureus versus coagulase-negative staph-
ylococci by peptide–nucleic acid fluorescence in situ hybridization can inform the clini-
cian with a rapid determination if the infection is a highly virulent or low virulence
microorganism.25–27 Linking rapid diagnostics results with antimicrobial stewardship
programs with the hospital is particularly useful in realizing the maximum benefits of
rapid microbial diagnostic testing16,19,22
Screening for the presence of antibiotic resistance genes is also certainly feasible as
a part of rapid genomic testing, but there are hundreds of known antibiotic resistance
genes and the list of new resistance genes continues to grow. Moreover, the
resistance gene sequence might be present in the genomic DNA of the pathogen
without being fully expressed.16 Clinical correlation between genotype and pheno-
typic expression in vivo during infection needs further investigation.
4 Opal & Wittebole
Another unmet medical need in managing critically ill patients with severe infection is
the lack of ability to provide rapid and directed antibiotic therapy guided by specific,
antimicrobial susceptibility testing.21,22 In this era of progressive loss of antimicrobial
activity from bacterial evolution of antibiotic resistance genes, rapid susceptibility
testing is urgently needed but rarely available. There are exceptions of course. Strep-
tococcus pyogenes (group A strep) remains uniformly susceptible to penicillin,28 but
for almost everything else direct antibiotic sensitivity testing to guide therapy is desir-
able if possible. Unfortunately, laboratory testing usually entails time-consuming incu-
bation times to fully express and define the susceptibility pattern for each identified
pathogen.
Progress is being made in the provision of novel methodologies to bring rapid sus-
ceptibility testing into clinical practice. Microfluidic systems with single cell imaging
assays,29 laser light scatter rapid imaging technologies,30 and time-lapse microscopic
imaging analyses31 are only some of the antimicrobial sensitivity testing systems un-
derway to fill the current void in rapid and clinically accessible susceptibility testing.
These rapid diagnostic systems are best applied in normally sterile fluids like blood,
cerebrospinal fluid, pleural fluid etc. and are not particularly useful at present for
microbiologic diagnosis from tissues with an endogenous microbiome (sputum, urine,
and gastrointestinal or genitourinary tract pathogens).32–35
Genomics
Advances in DNA sequencing technologies now make it feasible to sequence the exon
segments of the human genome and search for single nucleotide polymorphisms.
Myriads of single nucleotide polymorphisms are already identified in the human
genome that affect susceptibility to infection and alter the expression of the host
response during sepsis.53 We are just beginning to decipher the magnitude of interact-
ing genomic elements induced by sepsis, and the computational biology, which will be
essential to fully understanding the complexity of host susceptibility and resistance to
infection.
At present we can only scratch the surface of the host–pathogen interactions that
participate in the daily standoff between our genomes and an array of potential
microbial pathogens.53 Single nucleotide polymorphisms have already been identi-
fied in complement factors, immunoglobulin structure, intracellular signaling net-
works, and clearance capacity to eliminate lipopolysaccharides and other toxic
biolipids.57–59 Much basic research and applied science research is needed to
bring advances in functional genomics to the bedside in support of sepsis
management.
Transcriptomics
The major methodology now in use to analyze components of gene activation or
inhibition of the host response in sepsis is with upregulation or downregulation of
differentially expressed genes from a given specific baseline determination. Thou-
sands of genes are differentially expressed in response to a powerful and patho-
logic pattern recognition molecule like bacterial endotoxin. Complicating matters
further is that transcription gene profiles between individuals even with single site
of infection (eg, pneumonia or blood stream infections) are markedly heteroge-
neous.60 In response, investigators have endeavored to use unsupervised searches
for common transcriptional clusters to detect recognizable and reproducible
signatures of gene activation. This process requires a careful analysis for primary
driver transcripts from different individuals with the same or similar acute disease
process.61–63
Initial unsupervised clusters of transcripts are then linked in together from patients
with similar disease features to determine recognizable, organ-specific, tissue-spe-
cific, and circulating blood transcript patterns. Meta-clustering and data pooling
methods have improved the reproducibility of detection of common gene clusters
associated with specific diseases. Research teams have now shown it is possible
to distinguish infection-related systemic inflammatory states from noninfectious
inflammatory states by analyzing differences in the transcription profiles.64–67
McHugh and colleagues64 have been able to reduce the number of transcribed
genes necessary to predict the presence or absence of infection in critically ill patients
with only 4 targeted enzymes. The 4 gene products are LAMP1 (lysosomal-associated
membrane protein), PLAG7 (phospholipase A2, group VII), PLAC8 (placenta-specific
gene 8), and CEACAM 4 (carcinoembryonic antigen-related cell adhesion molecule
4). This work is of extreme importance in antibiotic stewardship practices in the ICU
by avoiding the use of antibiotics if no infection is present.67–69
Efforts are underway to streamline this process and develop diagnostic tests by nar-
rowing down the gene search to a small number of critical transcriptional determinants
of infection versus noninfection. This promising approach is ongoing in many labora-
tories around the globe.70,71 Although still in the developmental phase, these analyses
are beginning to decipher patterns of host response genes that might help guide
Biomarkers of Infection and Sepsis 7
Metabolomics
Another potentially rapid diagnostic methodology is the assessment of the metabolic
state of tissues within the host by state-of-the-art 1H nuclear magnetic resonance and
gas chromatographic MS metabolomics.73 Using principal components analysis for
pattern recognition of unsupervised clustering behaviors of patient samples, it is
possible to generate a 2-dimensional principal components analysis to rapidly recog-
nize the presence of infection versus sterile inflammation and predict the risk of mor-
tality in acutely ill patients. Both host-derived and pathogen-derived molecules can be
measured by this technique and provide early distinction between bacterial and viral
pathogens.74
SUMMARY
The role of biomarkers for the detection of sepsis has come a long way in the field of
sepsis management and enrollment of patients into clinical trials. Molecular bio-
markers are taking front stage at present but machine learning and other computa-
tional measures using big datasets is promising as well. Clinical research in sepsis
is hampered but lack of specificity of the diagnosis itself as sepsis is a syndrome
with no uniformly agreed definition.
One consequence of the lack of diagnostic precision in sepsis is who or what is the
gold standard for the final diagnosis of sepsis. Currently the final conclusion is expert
opinion, which is not bad but not perfect.80,81 Perhaps machine learning (or some
highly specific biomarker) will displace expert opinion as the final and most accurate
definition for sepsis.
REFERENCES
1. Opal SM. The evolution of our understanding of the nature of infection and sepsis.
Crit Care Clin 2009;25:637–63.
2. Artenstein A, Higgins T, Opal SM. Sepsis and scientific revolutions. Crit Care Med
2013;41(12):2770–2.
3. Pruinelli L, Westra BL, Yadav P, et al. Delay within the 3-hour Surviving Sepsis
Campaign guideline on mortality for patients with severe sepsis and septic
shock. Crit Care Med 2018;46:500–5.
4. Chen AX, Simpson SQ, Pallin DJ. Sepsis guidelines. N Engl J Med 2019;380(14):
1369–71.
5. Whiles BB, Deis AS, Simpson SQ. Increased time to initial antimicrobial adminis-
tration is associated with progression to septic shock and in severe sepsis pa-
tients. Crit Care Med 2017;45:623–9.
6. Levy MM, Evans LE, Rhodes A. The surviving sepsis campaign bundle: 2018 up-
date. Intensive Care Med 2018;44:925–8.
7. Seymour CW, Gesten F, Prescott, et al. Time to treatment and mortality during
mandated emergency care of sepsis. N Engl J Med 2017;376:2235–44.
8. Kumar A, Roberts D, Wood KE, et al. Duration of hypotension before initiation of
effective antimicrobial therapy is the critical determinant of survival in human sep-
tic shock. Crit Care Med 2006;34:1589–96.
9. Sterling SA, Miller WR, Pryor J, et al. The impact of timing of antibiotics on out-
comes in severe sepsis and septic shock: a systematic review and meta-anal-
ysis. Crit Care Med 2015;43:1907–15.
10. Singer M, Deutschman CS, Seymour CW, et al. The third international definitions
for sepsis and septic shock (Sepsis -3). JAMA 2016;315(8):801–10.
11. Pop-Vicas A, Opal SM. The clinical impact of multidrug resistant gram-negative
bacilli in the management of septic shock. Virulence 2014;5(1):1–7.
12. Opal SM. Non-antibiotic treatments for pan-resistant bacterial pathogens. Crit
Care 2016;20:397.
13. Opal SM, Pop-Vicas A. Molecular mechanisms of antibiotic resistance in bacte-
ria. In: Bennett JE, Dolin R, Blaser M, editors. Principles and practice of infectious
diseases. 9th edition. Elsevier Publishers: Philadelphia; 2019. p. 222–39.
14. Pfaller MA, Wolk DM, Lowery TJ. T2MR and T2 candida: novel technology for the
rapid diagnosis of candidemia and invasive candidiasis. Future Microbiol 2016;
11:103–7.
15. Patel GP, Simon D, Scheetz M, et al. The effect of time to antifungal therapy on
mortality in candidemia associated septic shock. Am J Ther 2009;16:508–11.
Biomarkers of Infection and Sepsis 9
16. Beganovic M, McCreary EK, Mahoney MV, et al. Interplay between rapid diag-
nostic tests and antimicrobial stewardship programs among patients with blood-
stream and other severe infections. J Appl Lab Med 2019;3(4):601–16.
17. Kerremans JJ, Verboom P, Stijnen T, et al. Rapid identification and antimicrobial
susceptibility testing reduce antibiotic use and accelerate pathogen-directed
antibiotic use. J Antimicrob Chemother 2008;61:428–35.
18. Timbrook TT, Morton JB, McConeghy KW, et al. The effect of molecular rapid
diagnostic testing on clinical outcomes in bloodstream infections: a systematic
review and meta-analysis. Clin Infect Dis 2017;64:15–23.
19. Cairns KA, Doyle JS, Trevillyan JM, et al. The impact of a multidisciplinary antimi-
crobial stewardship team on the timeliness of antimicrobial therapy in patients
with positive blood cultures: a randomized controlled trial. J Antimicrob Chemo-
ther 2016;71:3276–83.
20. Banerjee R, Teng CB, Cunningham SA, et al. Randomized trial of rapid multiplex
polymerase chain reaction-based blood culture identification and susceptibility
testing. Clin Infect Dis 2015;61:1071–80.
21. Bookstaver PB, Nimmich EB, Smith TJ, et al. Cumulative effect of an antimicrobial
stewardship and rapid diagnostic testing bundle on early streamlining of antimi-
crobial therapy in gram-negative bloodstream infections. Antimicrob Agents Che-
mother 2017;61 [pii:e00189-17].
22. Rivard KR, Athans V, Lam SW, et al. Impact of antimicrobial stewardship and
rapid microarray testing on patients with gram-negative bacteremia. Eur J Clin
Microbiol Infect Dis 2017;36:1879–87.
23. Vincent JL, Brealey D, Libert N, et al. Rapid diagnosis of infection in the critically
ill, a multicenter study of molecular detection in bloodstream infections, pneu-
monia, and sterile site infections. Crit Care Med 2015;43(11):2283–91. https://
doi.org/10.1097/CCM.001249.
24. Wilson NM, Alangaden G, Tibbetts RJ, et al. T2 magnetic resonance assay im-
proves timely management of candidemia. JAMS 2017;1(1):12–8.
25. Forrest GN, Mankes K, Jabra-Rizk MA, et al. Peptide nucleic acid fluorescence
in situ hybridization-based identification of Candida albicans and its impact on
mortality and antifungal costs. J Clin Microbiol 2006;44:3381–3.
26. Forrest GN, Mehta S, Weekes E, et al. Impact of rapid in situ hybridization testing
on coagulase-negative staphylococci positive blood cultures. J Antimicrob Che-
mother 2006;58:154–8.
27. Wenzler E, Wang F, Goff DA, et al. An automated, pharmacy-driven initiative im-
proves quality of care for Staphylococcus aureus bacteremia. Clin Infect Dis
2017;65:194–200.
28. Horn DL, Zabriskie JB, Austrian R, et al. Why have group A Streptococci re-
mained susceptible to penicillin? Clin Infect Dis 1998;26:1341–5.
29. Baltekin O, Boucharin A, Tano E, et al. Antibiotic susceptibility testing in less than
30 minutes using direct single-cell imaging. Proc Natl Acad Sci U S A 2017;
114(34):9170–5.
30. Choi J, Jeong HY, Lee GY, et al. Direct, rapid antimicrobial susceptibility test from
positive blood cultures based on microscopic imaging analysis. Sci Rep 2017;
7(1148):1–26.
31. Hayden RT, Clinton LK, Hewitt C, et al. Rapid antimicrobial susceptibility testing
using forward laser light scatter technology. J Clin Microbiol 2016;54(11):2701–6.
32. Mulpuru S, Aaron SD, Ronksley PE, et al. Hospital resource utilization and patient
outcomes associated with respiratory viral testing in hospitalized patients. Emerg
Infect Dis 2015;21:1366–71.
10 Opal & Wittebole
33. Potage CR, Cohen SH. State-of-the-art microbiologic testing for community-
acquired meningitis and encephalitis. J Clin Microbiol 2016;54:1197–202.
34. Leber AL, Everhart K, Balada-Llasat JM, et al. Multicenter evaluation of Biofire
Film Array meningitis/encephalitis panel for detection of bacteria, viruses and
yeast in cerebrospinal fluid specimens. J Clin Microbiol 2016;54:2251–61.
35. Hanson KE, Couturier MR. Multiplexed molecular diagnostics for respiratory,
gastrointestinal and central nervous infections. Clin Infect Dis 2016;63:1361–7.
36. Whitesides GM. The origins and the future of microfluidics. Nature 2006;
442(7101):368–73.
37. Crawford K, DeWitt A, Brierre S, et al. Rapid biophysical analysis of host immune
cell variations associated with sepsis. Am J Respir Crit Care Med 2018;198(2):
280–2.
38. Khismatullin DB. The cytoskeleton and deformability of white blood cells. Curr
Top Membr 2009;64:47–111.
39. Gossett DR, Tse HT, Lee SA, et al. Hydrodynamic stretching of single cells for
large population mechanical phenotyping. Proc Natl Acad Sci U S A 2012;
109(20):7630–5.
40. Cho SY, Choi JH. Biomarkers for sepsis. Infect Chemother 2014;46(1):1–2.
41. Reinhart K, Meisner M. Biomarkers in the critically ill patient: procalcitonin. Crit
Care Clin 2011;27:253–63.
42. Kopterides P, Siempos II, Tsangaris A, et al. Procalcitonin-guided algorithms of
antibiotic therapy in the intensive care unit: a systematic review and meta-
analysis of randomized controlled trials. Crit Care Med 2010;38:2229–41.
43. Anand D, Das S, Bhargava S, et al. Procalcitonin as a rapid diagnostic biomarker
to differentiate between culture-negative bacterial sepsis and systemic inflamma-
tory response syndrome: a prospective, observational, cohort study. J Crit Care
2015;30:218.e7-12.
44. Huang DT, Yealy DM, Filbin MR, et al. Procalcitonin–guided use of antibiotics for
lower respiratory tract infection. N Engl J Med 2018;379:236–49.
45. Thompson D, Pepys MB, Wood SP. The physiological structure of human C-reac-
tive protein and its complex with phosphocholine. Structure 1999;7(2):169–77.
46. Pepys MB, Hirschfield GM. C-reactive protein: a critical update. J Clin Invest
2003;111(12):1805–12.
47. Danesh J, Wheeler JG, Hirschfield GM, et al. C-reactive protein and other circu-
lating markers of inflammation in the prediction of coronary heart disease. N Engl
J Med 2004;350(14):1387–97.
48. Linder A, Arnold R, Boyd JH, et al. Heparin-binding protein measurement im-
proves the prediction of severe infection with organ dysfunction in the emergency
department. Crit Care Med 2015;43(11):2378–86.
49. Linder A, Akesson P, Inghammar M, et al. Elevated plasma levels of heparin-
binding protein in intensive care unit patients with severe sepsis and septic
shock. Crit Care 2012;16:R90.
50. Que YA, Delodder F, Guessous I, et al. Pancreatic stone protein as an early
biomarker predicting mortality in a prospective cohort of patients with sepsis
requiring ICU management. Crit Care 2012;16:R114.
51. Garcı́a de Guadiana-Romualdo L, Berger M, Jiménez-Santos E, et al. Pancreatic
stone protein and soluble CD25 for infection and sepsis in an emergency depart-
ment. Eur J Clin Invest 2017;47(4):297–304.
52. Tapia P, Gatica S, Cortez-Rivera C, et al. Circulating endothelial cells from septic
shock patients convert to fibroblasts are associated with the resuscitation fluid
Biomarkers of Infection and Sepsis 11
dose and are biomarkers for survival prediction. Crit Care Med 2019. https://doi.
org/10.1097/CCM 003778.
53. Hotchkiss RS, Moldawer LL, Opal SM, et al. Sepsis and septic shock. Nat Rev Dis
Primers 2016;2:1–21.
54. Dellinger RP, Bagshaw SM, Antonelli M, et al. Effect of targeted polymyxin B he-
moperfusion on 28-day mortality in patients with septic Shock and elevated endo-
toxin level. JAMA 2018;320(14):1455–63.
55. Gibot S, Cravoisy A, Levy B, et al. Soluble triggering receptor expressed on
myeloid cells and the diagnosis of pneumonia. N Engl J Med 2004;350:451–8.
56. Oved K, Cohen A, Bioco O, et al. A novel host-proteome signature for distinguish-
ing between acute bacterial and viral infections. PLoS One 2015;10:e0120012.
57. Walley KR, Francis GA, Opal SM, et al. The central role of proprotein convertase
subtilisin/kexin type 9 in septic pathogen lipid transport and clearance. Am J Re-
spir Crit Care Med 2015;192(11):1275–86.
58. Boyd JH, Fjell CD, Russel JA, et al. Increased plasma PCSK9 levels are associ-
ated with reduced endotoxin clearance and the development of acute organ fail-
ures during sepsis. J Innate Immun 2016. https://doi.org/10.1159/000442976.
59. Giamarellos-Bourboulis E, Opal SM. The role of genetics and antibodies in
sepsis. Ann Transl Med 2016;4(17):328, 1-8.
60. Tsalik EL, Langley RJ, Dinwiddlie DL, et al. An integrated transcriptome and ex-
pressed variant analysis of sepsis survival and death. Genome Med 2014;6:111.
61. Maslove DM, Wong HR. Gene expression profiling in sepsis: timing, tissue and
translational considerations. Trends Mol Med 2014;20:204–13.
62. Sweeney TE, Shidham A, Wong HR, et al. A comprehensive time-course-based
multicohort analysis of sepsis and sterile inflammation reveals a robust diagnostic
gene set. Sci Transl Med 2015;7:287ra271.
63. Geiss RK, Bumgarner RE, Birditt B, et al. Direct multiplexed measurement of
gene expression with color-coded probe pairs. Nat Biotechnol 2008;26(3):
317–25.
64. McHugh L, Seldon TA, Brandon RA, et al. A molecular host response assay to
discriminate between acute bacterial infection and infection-negative systemic
inflammation in critically ill patients; discovery and validation in independent co-
horts. PLoS Med 2015;12:e1001916.
65. Miller RR, Lopansri BK, Burke JP, et al. Validation of a host response assay, Steri-
Cycle LAB, for discriminating sepsis from systemic inflammatory syndrome in the
ICU. Am J Respir Crit Care Med 2018;198(7):904–13.
66. Bauer M, Giamarellos-Bourboulis EJ, Kortgen A, et al. A transcriptomic biomarker
to quantify systemic inflammation in sepsis-a prospective multicenter, phase II
diagnostic study. EBioMedicine 2016;6:114–25.
67. Zimmerman JJ, Sullivan E, Yager TD, et al. Diagnostic accuracy of a host gene
expression signature that discriminates clinical severe sepsis syndrome and
infection-negative systemic inflammation among critically ill children. Crit Care
Med 2017;45:e418–25.
68. Herberg JA, Kaforou M, Wright VJ, et al. Diagnostic test accuracy of a 2-tran-
script host RNA signature for discriminating bacterial from viral infection in febrile
children. JAMA 2016;316:835–45.
69. Burham KL, Davenport EE, Radhakrishnan J, et al. Shared and distinct aspects of
the sepsis transcriptomic response to fecal peritonitis and pneumonia. Am J Re-
spir Crit Care Med 2017;196:328–9.
70. Davenport EE, Burnham KL, Radhakristnan J, et al. Genomic landscape and out-
comes in sepsis: a prospective cohort study. Lancet Respir Med 2016;4:257–71.
12 Opal & Wittebole
71. Sweeney TE, Azad TD, Donato M, et al. Unsupervised analysis of transcriptomics
in bacterial sepsis across multiple datasets reveals three robust clusters. Crit
Care Med 2018;46(6):916–25.
72. Scicluna BP, van Vught LA, Zwiderman AH, et al. Classification of patients ac-
cording to blood genomic endotype: a prospective cohort study. Lancet Respir
Med 2017;5:816–26.
73. Coen M, O’Sullivan M, Bubb WA, et al. Proton nuclear magnetic resonance-
based metabolomics for rapid diagnosis of meningitis and ventriculitis. Clin Infect
Dis 2005;41(1):1582–90.
74. Banoei MM, Vogel HJ, Weljie AM, et al. Plasma metabolomics for the diagnosis
and prognosis of H1N1 influenza pneumonia. Crit Care 2017;21:97.
75. Seymour CW, Kennedy JN, Wang S, et al. Novel clinical phenotypes in sepsis:
derivation, validation, and potential treatment implications. JAMA 2019;321(20):
2003–17.
76. Shimabukuro DW, Barton CW, Feldman MD, et al. Effect of a machine-learning-
based severe sepsis prediction algorithm on patient survival and hospital length
of stay: a randomised clinical trial. BMJ Open Respir Res 2017;4:e000234.
77. Desautels T, Calvert J, Hoffman J, et al. Prediction of sepsis in the intensive care
unit with minimal electronic medical record data: a machine learning approach.
JMIR Med Inform 2016;4(3):e28.
78. Calvert J, Desautels T, Chettipally U, et al. High-performance detection and early
prediction of septic shock for alcohol-use disorder patients. Ann Med Surg
(Lond) 2016;8:50–5.
79. Rajkomar A, Dean J, Kohane I. Machine learning in medicine. N Engl J Med 2019;
380(14):1347–57.
80. Lopansri BK, Miller RR, Burke JP, et al. Physician agreement on the diagnosis of
sepsis in the intensive care unit: estimation of concordance and analysis of un-
derlying factors in a multicenter cohort. J Intensive Care 2019;7:13.
81. Klein Klouwenberg PM, Ong DS, Bos LD, et al. Interobserver agreement of the
Centers for Disease Control and Prevention criteria for classifying infections in
critically ill patients. Crit Care Med 2013;41:2373–8.