Heterogeneous Phenotypes of Acute Respiratory Distress
Syndrome after Major Trauma
John P. Reilly1,2, Scarlett Bellamy2, Michael G. S. Shashaty1,2, Robert Gallop2, Nuala J. Meyer1, Paul N. Lanken1,
Sandra Kaplan1, Daniel N. Holena3, Addison K. May4, Lorraine B. Ware5, and Jason D. Christie1,2
1Division of Pulmonary, Allergy, and Critical Care, Department of Medicine,2Center for Clinical Epidemiology and Biostatistics,
and3Division of Traumatology, Surgical Critical Care, and Emergency Surgery, Department of Surgery, University of Pennsylvania,
Perelman School of Medicine, Philadelphia, Pennsylvania;4Department of Surgery, and5Divison of Allergy, Pulmonary, and
Critical Care Medicine, Department of Medicine, and Department of Pathology, Microbiology, and Immunology, Vanderbilt University,
Rationale: Acute respiratory distress syndrome (ARDS) is
a heterogeneous syndrome that can develop at various times after
Objectives: To identify and characterize distinct phenotypes of
ARDS after trauma, based on timing of syndrome onset.
Methods: Latent class analyses were used to identify patterns
of ARDS onset in a cohort of critically ill trauma patients.
Identified patterns were tested for associations with known
ARDS risk factors and associations were externally validated at
a separate institution. Eleven plasma biomarkers representing
pathophysiologic domains were compared between identified
patterns in the validation cohort.
Measurements and Main Results: Three patterns of ARDS
were identified; class I (52%) early onset on Day 1 or 2, class II
(40%) onset on Days 3 and 4, and class III (8%) later onset on Days
4 and 5. Early-onset ARDS was associated with higher Abbreviated
Injury Scale scores for the thorax (P , 0.001), lower lowest systolic
blood pressure before intensive care unit admission (P = 0.003),
and a greater red blood cell transfusion requirement during
resuscitation (P = 0.030). In the external validation cohort,
early-onset ARDS was also associated with a higher Abbreviated
Injury Scale score for the thorax (P = 0.001) and a lower lowest
systolic blood pressure before intensive care unit enrollment
(P = 0.006). In addition, the early-onset phenotype demonstrated
higher plasma levels of soluble receptor for advanced glycation
end-products and angiopoietin-2.
Conclusions: Degree of hemorrhagic shock and severity of
thoracic trauma are associated with an early-onset phenotype of
ARDS after major trauma. Lung injury biomarkers suggest a
dominant alveolar–capillary barrier injury pattern in this phenotype.
Keywords: acute respiratory distress syndrome; critical illness;
phenotype; trauma; hemorrhagic shock
(Received in original form August 22, 2013; accepted in final form February 18, 2014)
Supported by National Institutes of Health grants HL60290, HL079063, HL007891, HL081332, HL112656, HL103836, HL115354.
Author Contributions: Conception and design: J.P.R., S.B., M.G.S.S., N.J.M., P.N.L., L.B.W., J.D.C.; acquisition of data: M.G.S.S., R.G., N.J.M., P.N.L., S.K.,
D.N.H., A.K.M., L.B.W., J.D.C.; analysis and interpretation of data: J.P.R., S.B., M.G.S.S., N.J.M., P.N.L., D.N.H., A.K.M., L.B.W., J.D.C.; drafting or revising
the manuscript for important intellectual content: all authors; final approval of the version to be published: all authors.
Correspondence and requests for reprints should be addressed to John P. Reilly, M.D., M.S., Division of Pulmonary, Allergy, and Critical Care
Medicine, University of Pennsylvania, Perelman School of Medicine, 844 West Gates Building, 3600 Spruce Street, Philadelphia, PA 19104.
This article has an online supplement, which is available from the issue’s table of contents at www.atsjournals.org
Ann Am Thorac Soc Vol 11, No 5, pp 728–736, Jun 2014
Copyright © 2014 by the American Thoracic Society
Internet address: www.atsjournals.org
Acute respiratory distress syndrome
(ARDS) is estimated to affect more than
190,000 patients annually in the United
States, with a mortality exceeding 35%
(1). In 1994, the American–European
Consensus Conference (AECC) developed
standardized criteria for diagnosing acute
lung injury and ARDS that are now widely
used both in research and clinically (2).
Acute lung injury was defined as acute-onset
bilateral pulmonary infiltrates on chest
radiograph, a ratio of PaO2to fraction of
inspired oxygen (FIO2) less than or equal
to 300, and the absence of left atrial
hypertension. More recently, experts have
modified the AECC criteria to improve
the validity and reliability of the ARDS
definition, using The “Berlin definition” (3).
AnnalsATS Volume 11 Number 5|June 2014
The establishment of clear definitions of
ARDS has led to significant advances in the
standardization of populations in research
studies; however, a number of studies
have shown significant heterogeneity
within the population of patients meeting
consensus criteria for ARDS (4, 5).
Heterogeneity has been described on the
basis of predisposing insult, such as sepsis
or trauma, and mechanism of injury,
such as direct or indirect pulmonary injury
(6, 7). Within trauma populations, different
patterns of ARDS onset with distinct
ARDS risk factors have been empirically
described and may have different
pathogenesis (8–11). We hypothesized that
there are clinically distinct phenotypes of
ARDS based on the timing of ARDS onset
after severe trauma with distinct clinical
risk factors and pathogenesis.
Latent class analysis (LCA) is
a statistical method used to identify
unobserved (latent) patterns underlying
the observed heterogeneity in a population.
The latent class approach assumes that the
disease of interest represents a mixture of
distinct subgroups that are not directly
observed but can be determined on the basis
of variables of interest, allowing for an
empiric derivation of phenotypes of disease
rather than using prespecified groups.
Latent class models have been applied to
identify various syndrome phenotypes,
including classes of asthma (12–15).
The primary goal of this study was
to define clinically meaningful subgroups
within the diagnosis of ARDS by first
applying a latent classmodeling approach to
a cohort of major trauma patients with
ARDS based on the certainty of the ARDS
diagnosis each day over 5 days. Second,
we aimed to characterize the construct
validity of resulting latent classes of ARDS
by testing differences in both clinical ARDS
risk factors and biomarkers known to be
associated with ARDS risk and outcomes.
Some of the results of this study were
previously reported in the form of abstracts
Please see the online supplement for further
details of the study methods.
A cohort study of acutely injured trauma
patients was used to derive ARDS classes.
Subjects were enrolled from 1999 to 2002
and from 2005 to 2008 at the Hospital of the
University of Pennsylvania (Philadelphia,
PA) (18, 19). All trauma patients presenting
to the emergency department (ED) and
admitted to the surgical intensive care unit
(ICU) were screened. Inclusion criteria
included age greater than 13 years, Injury
Severity Score (ISS) greater than or equal
to 16 (20), and traumatic injury within
24 hours of presentation. Outside hospital
transfers were included only if they were
transferred from another ED on the day of
acute trauma. Exclusion criteria included
discharge or death within 24 hours of
admission, current or past evidence of
congestive heart failure, and isolated head
trauma. Patients with isolated head
trauma were excluded because neurogenic
pulmonary edema is believed to have
a distinct pathogenesis from trauma-related
ARDS (21), and the early mortality of
patients with isolated head trauma and an
ISS greater than or equal to 16 is extremely
high, resulting in patients dying before
All patients identified for the cohort
were prospectively screened for ARDS
and had clinical data collected during the
first 5 days after admission. We monitored
subjects for ARDS development for 5 days
to capture ARDS cases directly related to
acute incident trauma (22). Using our
previous published methods, patients were
classified during each 24-hour period from
the time of admission as (1) definite ARDS,
(2) equivocal, or (3) definite non-ARDS
(23). Patients classified as definite ARDS
met all of the AECC defining criteria
for acute lung injury while intubated and
receiving mechanical ventilation. The
criteria include acute-onset confluent
infiltrates on chest radiograph consistent
with pulmonary edema, a PaO2/FIO2ratio
less than or equal to 300, and absence of
left atrial hypertension (2). All available
chest radiographs and arterial blood
gases ordered for clinical purposes were
evaluated. The timing of ARDS onset was
determined by identifying the time
when both arterial blood gas and chest
radiograph criteria had been met, according
to the consensus definition. Chest
radiograph classification was determined
by two physician-investigators, blinded
to all data and each other’s interpretations,
who reviewed all available chest radiographs
with adjudication by a third reviewer
if necessary. Chest radiographs were
considered “positive” if bilateral confluent
opacities consistent with pulmonary
edema were present, “negative” if
pulmonary opacities were not consistent
with pulmonary edema or were only
unilateral, and “equivocal” if reviewers
could not confidently classify the chest
radiograph (e.g., the infiltrates were difficult
to distinguish from atelectasis or effusion,
or the chest radiograph was technically
deficient) (23). Patients who had a
definitive diagnosis of ARDS on at least
one of the first 5 days, defined as 24-hour
periods from admission, were considered
for inclusion in the LCA.
To enhance clinical usefulness, the classes
resulting from the derivation population
LCA were simplified as early- and late-onset
ARDS with a 48-hour cutoff, based on data
inspection. This 48-hour cutoff for early
and late onset was used for validation in
a separate cohort of patients admitted to an
ICU at Vanderbilt University Medical
Center (Nashville, TN) from 2006 to 2011
(the VALID [Validating Acute Lung Injury
Biomarkers for Diagnosis] Study). Patients
were eligible if they were at least 18 years of
age, suffered an acute traumatic injury
within 24 hours of ED presentation, and
were admitted to the ICU for at least
48 hours, regardless of ISS. Subjects were
enrolled on the morning after ICU
admission. Details of this cohort have
been previously described (24, 25). Eleven
biomarkers representing various domains
of ARDS pathogenesis (see Table E1 in
the online supplement) were measured in
plasma samples obtained from a subgroup
of 112 subjects in the validation cohort
at the time of enrollment (26–28).
in the validation cohort daily, using all
available chest radiographs and arterial
blood gases in the prior 24 hours obtained
at the discretion of the treating physicians
(24, 25). Two physician-investigators
uninvolved in the patients’ care interpreted
the chest radiographs by consensus.
Patients were first evaluated for ARDS at
enrollment in the ICU and then again
daily for three subsequent days. Similar to
the derivation cohort, the timing of ARDS
onset was defined as the time that both
arterial blood gas and chest radiograph
criteria were met.
The Institutional Review Boards of the
University of Pennsylvania and Vanderbilt
Reilly, Bellamy, Shashaty, et al.: Timing of Onset Phenotypes of ARDS
acute lung injury on hospital mortality in critically ill trauma patients.
Crit Care Med 2008;36:2309–2315.
19 Christie JD, Ma SF, Aplenc R, Li M, Lanken PN, Shah CV, Fuchs B,
Albelda SM, Flores C, Garcia JG. Variation in the MYLK gene is
associated with development of acute lung injury after major trauma.
Crit Care Med 2008;36:2794–2800.
20 Civil ID, Schwab CW. The Abbreviated Injury Scale, 1985 revision:
a condensed chart for clinical use. J Trauma 1988;28:87–90.
21 Sedý J, Zicha J, Kunes J, Jendelová P, Syková E. Mechanisms of
neurogenic pulmonary edema development. Physiol Res 2008;57:
22 Hudson LD, Milberg JA, Anardi D, Maunder RJ. Clinical risks for
development of the acute respiratory distress syndrome. Am J
Respir Crit Care Med 1995;151:293–301.
23 Shah CV, Lanken PN, Localio AR, Gallop R, Bellamy S, Ma SF, Flores
C, Kahn JM, Finkel B, Fuchs BD, et al. An alternative method of
acute lung injury classification for use in observational studies. Chest
24 O’Neal HR Jr, Koyama T, Koehler EA, Siew E, Curtis BR, Fremont RD,
May AK, Bernard GR, Ware LB. Prehospital statin and aspirin use
and the prevalence of severe sepsis and acute lung injury/acute
respiratory distress syndrome. Crit Care Med 2011;39:1343–1350.
25 Siew ED, Ware LB, Gebretsadik T, Shintani A, Moons KG, Wickersham
N, Bossert F, Ikizler TA. Urine neutrophil gelatinase-associated
lipocalin moderately predicts acute kidney injury in critically ill adults.
J Am Soc Nephrol 2009;20:1823–1832.
26 Fremont RD, Koyama T, Calfee CS, Wu W, Dossett LA, Bossert FR,
Mitchell D, Wickersham N, Bernard GR, Matthay MA, et al.
Acute lung injury in patients with traumatic injuries: utility of a
panel of biomarkers for diagnosis and pathogenesis. J Trauma 2010;
27 Levitt JE, Gould MK, Ware LB, Matthay MA. The pathogenetic and
prognostic value of biologic markers in acute lung injury. J Intensive
Care Med 2009;24:151–167.
28 Cross LJ, Matthay MA. Biomarkers in acute lung injury: insights into the
pathogenesis of acute lung injury. Crit Care Clin 2011;27:355–377.
29 Collins LM, Lanza ST. Latent class and latent transition analysis:
with applications in the social, behavioral, and health sciences.
Hoboken, NJ: John Wiley & Sons; 2010.
30 Conaway MR. Latent class analysis. In: Armitage P, Colton T, editors.
Encyclopedia of biostatistics. Hoboken, NJ: Wiley; 2005.
31 PROC LCA & PROC LTA (version 1.2.7) [Software]. University Park: The
Methodology Center, Pennsylvania State University; 2011.
32 Han MK, Agusti A, Calverley PM, Celli BR, Criner G, Curtis JL, Fabbri
LM, Goldin JG, Jones PW, Macnee W, et al. Chronic obstructive
pulmonary disease phenotypes: the future of COPD. Am J Respir Crit
Care Med 2010;182:598–604.
33 Moore WC, Meyers DA, Wenzel SE, Teague WG, Li H, Li X, D’Agostino
R Jr, Castro M, Curran-Everett D, Fitzpatrick AM, et al.; National
Heart, Lung, and Blood Institute’s Severe Asthma Research
Program. Identification of asthma phenotypes using cluster analysis
in the Severe Asthma Research Program. Am J Respir Crit Care Med
34 Vincent JL, Sakr Y, Groeneveld J, Zandstra DF, Hoste E, Malledant Y,
Lei K, Sprung CL. ARDS of early or late onset: does it make
a difference? Chest 2010;137:81–87.
35 Justice AC, Covinsky KE, Berlin JA. Assessing the generalizability of
prognostic information. Ann Intern Med 1999;130:515–524.
36 van Meurs M, K¨ umpers P, Ligtenberg JJ, Meertens JH, Molema G,
Zijlstra JG. Bench-to-bedside review: angiopoietin signalling in
critical illness—a future target? Crit Care 2009;13:207.
37 Fiedler U, Augustin HG. Angiopoietins: a link between angiogenesis
and inflammation. Trends Immunol 2006;27:552–558.
38 Agrawal A, Matthay MA, Kangelaris KN, Stein J, Chu JC, Imp BM,
Cortez A, Abbott J, Liu KD, Calfee CS. Plasma angiopoietin-2
predicts the onset of acute lung injury in critically ill patients.
Am J Respir Crit Care Med 2013;187:736–742.
39 Creagh-Brown BC, Quinlan GJ, Evans TW, Burke-Gaffney A. The RAGE
axis in systemic inflammation, acute lung injury and myocardial
dysfunction: an important therapeutic target? Intensive Care Med
40 Wolfson RK, Chiang ET, Garcia JG. HMGB1 induces human lung
endothelial cell cytoskeletal rearrangement and barrier disruption.
Microvasc Res 2011;81:189–197.
41 Demling N, Ehrhardt C, Kasper M, Laue M, Knels L, Rieber EP.
Promotion of cell adherence and spreading: a novel function
of RAGE, the highly selective differentiation marker of human
alveolar epithelial type I cells. Cell Tissue Res 2006;323:
42 Uchida T, Shirasawa M, Ware LB, Kojima K, Hata Y, Makita K, Mednick
G, Matthay ZA, Matthay MA. Receptor for advanced glycation
end-products is a marker of type I cell injury in acute lung injury.
Am J Respir Crit Care Med 2006;173:1008–1015.
43 Cohen MJ, Carles M, Brohi K, Calfee CS, Rahn P, Call MS, Chesebro
BB, West MA, Pittet JF. Early release of soluble receptor for
advanced glycation endproducts after severe trauma in humans.
J Trauma 2010;68:1273–1278.
44 van der Heijden M, van Nieuw Amerongen GP, Chedamni S, van
Hinsbergh VW, Johan Groeneveld AB. The angiopoietin–Tie2 system
as a therapeutic target in sepsis and acute lung injury. Expert Opin
Ther Targets 2009;13:39–53.
45 Sarraf-Yazdi S, Mi J, Moeller BJ, Niu X, White RR, Kontos CD,
Sullenger BA, Dewhirst MW, Clary BM. Inhibition of in vivo tumor
angiogenesis and growth via systemic delivery of an angiopoietin
2-specific RNA aptamer. J Surg Res 2008;146:16–23.
46 White RR, Shan S, Rusconi CP, Shetty G, Dewhirst MW, Kontos CD,
Sullenger BA. Inhibition of rat corneal angiogenesis by a nuclease-
resistant RNA aptamer specific for angiopoietin-2. Proc Natl
Acad Sci USA 2003;100:5028–5033.
47 Lutterloh EC, Opal SM, Pittman DD, Keith JC Jr, Tan XY, Clancy BM,
Palmer H, Milarski K, Sun Y, Palardy JE, et al. Inhibition of the
RAGE products increases survival in experimental models of severe
sepsis and systemic infection. Crit Care 2007;11:R122.
48 Lutterloh EC, Opal SM. Antibodies against RAGE in sepsis and
inflammation: implications for therapy. Expert Opin Pharmacother
AnnalsATS Volume 11 Number 5|June 2014