Description of compensated and uncompensated disseminated

intravascular coagulation (DIC) responses (non-overt and overt DIC)

in baboon models of intravenous and intraperitoneal Escherichia coli
sepsis and in the human model of endotoxemia: Toward a better
definition of DIC
Fletcher B. Taylor, Jr, MD; Hideo Wada, MD; Gary Kinasewitz, MD

Objective: Work toward a better definition of disseminated
intravascular coagulation (DIC) by characterizing the difference
between compensated and uncompensated responses of the he-
mostatic system to inflammatory stress in baboons and human
subjects using global coagulation and molecular marker assays of
hemostatic, inflammatory, and endothelial perturbation.
Design: We conducted prospective evaluation of the response
of baboons to increasing concentrations of intravenous Esche-
richia coli, human subjects to intravenous endotoxin, and ba-
boons to intraperitoneal E. coli.
Setting: Animal laboratory and medical intensive care facili-
ties, University of Oklahoma Medical School laboratories.
Subjects: Cynocephalus baboons; normal healthy male human
subjects (age, 24 37 yrs).
Measurements and Main Results: Global coagulation assays,
white blood cell counts, and molecular marker assays (ELISA) of
components of the inflammatory and hemostatic systems, neu-
trophil release products, and endothelial injury. A fall in both
fibrinogen concentration and platelet counts indicated a decom-
pensated hemostatic response to inflammatory stress (ie, overt
DIC). These responses were observed 2 6 hrs after intravenous
infusion of 109 and 1010 colony-forming units (CFU)/g of E. coli
and after implantation of 1011 CFU/g of E. coli into the peritoneal
cavity. However, 6 hrs after E. coli challenge, these tests were
much less reliable as markers of overt DIC because the fibrinogen
underwent an acute phase response and the platelet count fell
and remained depressed for 48 hrs in the face of a coagulopathic
response that was already beginning to resolve, as reflected by a
rising fibrinogen concentration. This lack of reliability was par-
ticularly evident in the E. coli peritonitis studies, in which one
third of the animals recovered, one third remained sick for up to
14 days, and one third died.
In contrast, fibrin degradation products and the molecular
markers thrombin/antithrombin, soluble fibrin monomer, protein
C, and activated protein C/inhibitor complexes responded consis-
tently in a dose-dependent manner regardless of the length of
time after challenge. These variables exhibited this dose response
to 106 and 108 CFU/g of E. coli in absence of a fall in fibrinogen
concentration. This was defined as a compensated hemostatic
response to inflammatory stress (ie, non-overt DIC). The values of
these variables correlated closely with rising concentrations of
markers of neutrophil activation (elastase/ 1 antitrypsin) and
endothelial injury (soluble thrombomodulin). This was particularly
evident in the human response to endotoxin, in which there was
abundant evidence of hemostatic marker response in absence of
a fall in platelet or fibrinogen concentration, both immediately
after endotoxin infusion (first stage, 0 8 hrs after endotoxin) and
later (second stage, 12 48 hrs after endotoxin).
Conclusion: Global coagulation tests are most useful in detect-
ing overt consumptive coagulopathy (overt DIC) near the time of
challenge or injury (1 to 6 hrs). Molecular markers can detect and
grade the degree of hemostatic stress of a non-overt consumptive
coagulopathy (nonovert DIC). These markers correlate with de-
gree of endothelial cell injury and reveal a reperfusion injury stage
(second stage) in the human endotoxin model of compensated
hemostatic stress after all clinical symptoms have subsided and
the subjects have returned to work. (Crit Care Med 2000;
28[Suppl.]:S12S19)
KEY WORDS: disseminated intravascular coagulation; sepsis;
septic shock; reticuloendothelial system; microvasculature; intra-
venous endotoxemia; systemic inflammation response syndrome;
overt disseminated intravascular coagulation; non-overt dissem-
inated intravascular coagulation; microvascular reperfusion

From the Oklahoma Medical Research Founda-
tion (Dr. Taylor), and the University of Oklahoma Health
Sciences Center (Dr. Kinasewitz), Oklahoma City, OK; and
Mie University School of Medicine, Tsu-City, Japan (Dr.
Wada).
Supported, in part, by National Institutes of Health
5 R01 GM37704 12.
Presented, in part, at the Margaux Conference on
Critical Illness, Margaux, France, November 1113,
1999.
Address requests for reprints to: Fletcher B. Taylor,
Jr, Cardiovascular Biology Research Program, Okla-
homa Medical Research Foundation, 825 NE 13th
Street, Oklahoma City, OK 73104.
Copyright 2000 by Lippincott Williams & Wilkins

S12 Crit Care Med 2000 Vol. 28, No. 9 (Suppl.)

I deally, one would like to predict, integrity of the microvascular trans- vasculature but contributes to that dys-
identify, and assess patients in- port organ. function. Although not always directly
flammatory/hemostatic response Third, an acute inflammatory distur- linked to the lethal chain of events,
to sepsis, trauma, burns, etc., as bance as caused by E. coli can result in activation of hemostatic factors occurs
well as follow the full course of the re- one or more of the following three re- in the capillary leak, consumptive, and
sponse. This would include recognizing sponses: a) loss of fluid and shock (cap- thrombotic coagulopathic responses of
the changes characteristic of recovery as illary leak syndrome, vascular col- this organ. Therefore, it should be pos-
well as the development of the inflamma- lapse); b) loss of blood (hemorrhage); sible to assess changes in the hemo-
tory/hemostatic response. c) obstruction of flow (reversible or ir- static factors and use these to diagnose
In studying the baboon model of Esch- reversible microvascular thrombosis). dysfunction of the microvascular organ
erichia coli sepsis, we have kept the fol- Malfunction of this transport organ in much as we use changes in the serum
lowing points in mind: one of these ways can lead to hypoxic creatinine to diagnose renal dysfunc-
stress and failure of other organs (ie, tion.
First, the systemic inflammatory/
multiple organ failure). Finally, because activation of hemo-
hemostatic response to E. coli com-
Fourth, activation of hemostatic fac- static factors is common to all three
prises (at least initially) an acute in-
tors reflects stress on the microvascu- variants of the response to E. coli, one
flammation of the microvascular
lar (transport) organ that is supplying is tempted to view them not only as
circulatory system, including circulat-
all the other organs because the hemo- markers of perturbation of the reticu-
ing blood components and the fixed
static system is an integral part of this loendothelial system microvascular or-
reticuloendothelial components that
larger microvascular organ. The dys- gan but also as one of the links in the
are in contact with the blood.
function of the hemostatic system not lethal chain of events in all three vari-
Second, these circulating and fixed
only marks dysfunction of the micro- ants. It would be reasonable to assume
components taken together constitute
an organ, the reticuloendothelial/
microvasculature. This extensive organ
can be viewed as consisting of blood
and the fixed microvascular elements
(endothelium, fixed macrophages,
Kupffer cells) that are in contact with
the blood, which together exhibit mul-
tiple functions ranging from endocrine
and immune functions to transport.
The transport function requires not
only that the blood flow freely but also
that the fixed vascular structures do
not leak. A constant dynamic interac-
tion at the blood/endothelial interface
involving inflammatory and hemo-
static mediator and regulatory net-
works maintains both the flow and the

Table 1. Diagnostic tests

Hemostatic system activation and product

formation
1. Global coagulation tests (fibrinogen,
platelet, fibrinogen degradation products)
2. Mediatorthrombin as measured by
thrombin-antithrombin complex
3. Regulatoractivated protein C as
measured by activated protein C/protein C
inhibitor complexes
4. Productsdimerized plasmin fragment D,
soluble fibrin monomer
5. Inhibitorantithrombin (AT)
6. Substrateprotein C
Neutrophil activation
1. Elastase/ 1AT complex
2. Oxidative metabolism (luminescence)
Endothelial (microvascular) injury
Figure 1. Global coagulation tests (dose response to intravenous Escherichia coli). Baboons were
1. Tissue plasminogen activator, von
infused with four different concentrations of E. coli: 106 colony-forming units (CFU)/kg (n 2), 108
Willebrand Factor
2. Soluble thrombomodulin CFU/kg (n 2), 109 CFU/kg (n 3), and 1010 CFU/kg. This figure shows the responses of fibrin
3. Plasma tissue factor antigen degradation products (FDP) (left), platelets (center), and fibrinogen (right) followed over a 48-hr
period.

Crit Care Med 2000 Vol. 28, No. 9 (Suppl.) S13

 that because all the same players ap-
pear on the field in all three types of
responses, they share a common mech-
anism that can be blocked if treated
early enough or at the right time with
a single agent; however, this is not
true. The status of the hosts effector
and target cells (tissues) at the time of
the challenge determines which of the
three variants and which of several pos-
sible effector mechanisms will be dom-
inant. This in turn determines whether
the host dies within hours of capillary
leak syndrome or within days of irre-
versible microvascular thrombosis.
Even although the same players (cyto-
kines, activated coagulation factors) do
appear on the field, depending on the
status of the host at the time of chal-
lenge or the stage of the disease, they
run different plays against cellular de-
fenses, which are themselves adjusting
and reprogramming on the fly. This
puts a premium on diagnosis and on a
treatment that is customized to a par-
ticular host response, much as we cus-
tomize antibacterial treatment to a
particular bacterial organism.
only after the infusion of 109 and 1010
colony-forming units (CFU)/kg concen-
trations of E. coli, and in the case of 109
CFU/kg concentration of E. coli, the fi-
brinogen, being an acute phase protein,
recovered and rose above baseline by 48
hrs. Infusion of lower concentrations of
E. coli produced either no change or a
rise in fibrinogen. Therefore, this acute
phase response can mask evidence of
overt DIC if given time.
Platelets present a different problem.
Figure 1 shows that they were more sen-
sitive than fibrinogen and responded to
concentrations of E. coli as low as 108
CFU/kg. In fact, they responded identi-
cally to all concentrations of E. coli and
remained depressed for 48 hrs regard-
less of the concentration of E. coli in-
fused. Furthermore, experience with in-
strumented baboons has shown that
platelet counts often become depressed in
the absence of any infusion of E. coli (3).
This sensitivity, nonspecificity, and pro-
longed depression can give false indica-
tion of a consumptive coagulopathy
(overt DIC) because other factors besides
thrombin (eg, neutrophil protease) asso-
ciated with injury of the microvascula-
ture may account for sustained thrombo-
cytopenia (4). As a result, critical care
physicians often consider sustained
thrombocytopenia after acute injury or
trauma as a useful although nonspecific
marker of microvascular injury, and a
rise in platelet count as a positive prog-
nostic sign (5).
Fibrin degradation products gave the

Comparative Dose-Response Studies

of Global Coagulation and Molecular
Marker Tests in the Baboon Model of
Intravenous Escherichia coli Sepsis. The
first step in reviewing our experience
with diagnostic tests in the baboon E. coli
sepsis models is to examine their re-
sponse patterns as one progresses from
decompensated to uncompensated injury
of the microvascular (endothelial) trans-
port system. The tests used in these stud-
ies fall into three categories: those that
measure activation of hemostatic compo-
nents and product formation, those that
measure neutrophil activation and re-
lease, and those that measure endothelial
cell injury (Table 1). Each test has limits,
advantages, and specific uses as follows.
Global Coagulation Tests. Global tests
of coagulation factors including fibrin
degradation products (FDP), platelets,
and fibrinogen are diagnostic for overt
disseminated intravascular coagulation
(DIC); they have been in widespread clin-
ical use and are available worldwide. Clin-
ical investigators have shown that if these Figure 2. Molecular markers (dose response to intravenous Escherichia coli). Baboons were infused
with four different concentrations of E. coli: 106 colony-forming units (CFU)/kg (n 2), 108 CFU/kg
tests are scored as a group in conjunction
(n 2), 109 CFU/kg (n 3), 1010 CFU/kg E. coli. This figure shows the responses of fibrin degradation
with clinical criteria, they are helpful in
products (left), platelets (center), and fibrinogen (right) followed over a 48-hr period. The left column
the diagnosis and characterization of shows the responses of thrombin/antithrombin complex (TAT, ), and soluble fibrin monomer (sFM,
overt DIC, particularly in trauma patients circles). The center column shows the response of protein C (PC, ), and activated protein C/protein
(1, 2). Individually, however, each test C inhibitor complex (APC/PCI, circles). The right column shows the responses of soluble thrombo-
has specific limitations. Figure 1 shows modulin (sTM, ), and elastase/alpha 1AT (ELAS, circles). All observations were made over a 48-hr
that the concentration of fibrinogen fell period.

S14 Crit Care Med 2000 Vol. 28, No. 9 (Suppl.)

most consistent dose response to E. coli.
Figure 1 shows that fibrin degradation
products appeared in a dose-dependent
manner within 6 hrs after E. coli infu-
sion. This figure also shows that after
infusion of 108 and 109 CFU/kg of E. coli,
a second fibrin degradation product peak
appeared at 24 hrs after the E. coli chal-
lenge, although fibrinogen levels were at
or above baseline levels.
Molecular Markers of Activation of
Hemostatic Factors and Neutrophils and
of Injury to the Endothelium. Figure 2
(left tier) shows that thrombin activation
as reflected by assays of thrombin/
antithrombin (TAT) complex responded
to as little as 106 CFU/kg of E. coli. This
response was dose dependent, but like the
cytokines (not shown), the TAT values
returned to baseline before the response
to E. coli had run its full course. This is
shown by soluble fibrin monomer (sFM),
which exhibited a small but significant
peak soon after the infusion of 10 8
CFU/kg of E. coli at 2 hrs and a much
larger peak at 24 hrs. This two-stage re-
sponse of sFM and the rapid decline of
TAT toward baseline began to merge after
infusion of 109 CFU/kg E. coli and then
merged completely after infusion of 1010
CFU/kg. We concluded that the two-stage
sFM response pattern was unique to the
infusion of 108 CFU/kg of E. coli. This
response pattern reflects an initial hemo-
static response secondary to the direct
effects of E. coli followed by a second
greater hemostatic response secondary to
reperfusion of the microvasculature. The
stronger direct effects of the higher con-
centrations of E. coli overrode any evi-
dence of reperfusion and/or produced
conditions in which reperfusion did not
occur. The conclusion that the second
peak of sFM represented reperfusion was
supported by the rising lactate levels, the
Figure 3. Qualitative correlation of phagocyte elastase production and lactate concentration with
appearance of plasma tissue factor anti-
markers of the inflammatory and hemostatic system responses to endotoxin. Top, after a bolus
gen, and a fall in factor VIIa at 1224 hrs
infusion of endotoxin (4 ng/kg), the elastase/alpha 1AT complexes (circles) appeared and peaked twice
(not shown), long after the cytokines had at 4 and 8 hrs before rapidly returning to baseline. In contrast, the lactate concentration (), like that
returned toward baseline and the animals of the soluble thrombomodulin, remained elevated before slowly returning toward baseline at 48 hrs
had recovered normal activity and appe- after endotoxin infusion. Second from top, The concentrations of inflammatory mediators tumor
tite. necrosis factor (TNF) (solid circles) and interleukin-6 (IL-6) () peaked at 2 hrs and 4 hrs, respectively,
Figure 2 (center tier) shows that the before rapidly returning to normal, whereas the inflammatory regulators IL-10 (triangles) and cortisol
protein C concentration changed in a re- (open circles) peaked at 3 hrs and 4 hrs followed by a second peak of IL-10 at 7 8 hrs and a sustained
ciprocal manner to the production of elevation of cortisol to 8 hrs after endotoxin infusion. Second from bottom, the concentrations of
thrombin. Protein C concentration fell hemostatic mediators inhibitor complexes thrombin/antithrombin (TAT) (solid circles) and plasmin/
antiplasmin complex (PAP) () peaked at 4 and 2 hrs respectively before returning toward baseline,
after the infusion of as little as 106
whereas the concentration of the protein C hemostatic regulator inhibitor complex (open circles)
CFU/kg of E. coli, and there was a dose-
peaked twice at 2 hrs and 7 hrs after endotoxin infusion. Bottom, the concentration of factor VIIa ()
response generation of activated protein did not reach a nadir until 12 hrs after endotoxin infusion, long after the initial inflammatory and
C (APC) as indicated by the formation of hemostatic mediator and regulator responses had returned toward baseline. This delayed response was
increasing amounts of APC/protein C in- mirrored by a reciprocal rise in the concentration of soluble tissue factor (TF) (open circles) and
hibitor (PCI) complexes. Although the soluble fibrin (S.Fib) (solid circles), which also peaked at 12 hrs. Note that these late events occurred
protein C network remained functional, after the subjects had become asymptomatic and had returned to work the following day.

Crit Care Med 2000 Vol. 28, No. 9 (Suppl.) S15

it was increasingly overridden as the con- Table 2. Molecular markers of dysfunction
centration of E. coli rose from 108 to 1010
Subject
CFU/kg as suggested by the sustained de-
pression of protein C concentrations to Mild Responders Moderate Responders
48 hrs after the E. coli challenge. The fact
that the protein C concentration was not 1 2 4 3
driven to 35% after 10 9 and 10 10
CFU/kg of E. coli raises the question of a Neutrophil activity
rate-limiting step (receptor inactivation) COR/P 41 35 77 60
Elastase/1AT (ng/mL) 157 247 272 389
caused by the inflammatory response. Endothelial and hemostatic perturbation
Finally, Figure 2 (right) shows that sThrombomodulin (ng/mL) 16 22 33 39
markers of neutrophil activation (elas- Soluble fibrin (g/mL) 13 20 64 96
tase/1AT) and endothelial injury (solu- Hepatic, metabolic, and symptomatic response
SGPT (units/L) 17 20 47 25
ble thrombomodulin, sTM) also respond Lactate (meq/L) 2.1 1.4 2.3 2.8
to increasing concentrations of E. coli in Rigors (04) 0.2 0.2 1.4 1.4
a dose-dependent manner. The response
of these markers, like those of TAT and COR/P, basal oxidase activity/phagocyte; AT, antithrombin; SGPT, serum glutamic-pyruvic
8 transaminase.
sFM to 10 CFU/kg of E. coli, was unique.
Values returned toward baseline after 24 Comparison of the intensity of the neutrophil response (COR/P) to endotoxin with the degree of
to 48 hrs, suggesting a recovery that cor- endothelial (thrombomodulin) and hemostatic system (soluble fibrin) perturbation, and with the
hepatic (SGPT), metabolic (lactate), and symptomatic (rigors) responses. The concentrations or scores
responded with evidence of reperfusion as
of each of the four subjects between 0 and 48 hrs after endotoxin infusion were added and then divided
reflected by the second soluble fibrin by the number of measurements taken (10 12).
monomer peak and rising lactate during
the same interval. In contrast, the re-
sponse to the higher E. coli concentra-
tions was characterized by higher peak
concentrations of elastase/1AT and by
sTM concentrations that continued to rise
during the full observation period. By
comparison, Figure 1 shows that the
global coagulation tests (with the excep-
tion of platelet count) were above or near
baseline up to 48 hrs, at which time the
fibrinogen averaged 223%, FDP averaged
23 g/dL, white blood cell count averaged
7.8 103/mm3, and body temperature av-
eraged 38.1C. This is striking evidence
that the above-described molecular mark-
ers registered inflammatory/hemostatic ac-
tivity not detected by the global coagulation
tests. Figure 1, however, also shows that by
6 hrs, the global coagulation tests did ap-
propriately reflect consumption of hemo-
static factors after 109 and 1010 CFU/kg of
E. coli, at which time the fibrinogen aver-
aged 35%, FDP averaged 80 g/dL, white
blood cell count averaged 1.9 103/mm3,
and bocy temperature averaged 39C.
Although these studies of the baboon
intravenous model of E. coli sepsis show
the limits of global coagulation tests and
offer a rationale supporting the use of
molecular marker tests, two questions re-
main. Can molecular markers be used to
detect dysfunction of the microvascular
hemostatic network under stress before it
becomes decompensated in the human
model of endotoxemia? And will these Figure 4. Global coagulation tests (response to intraperitoneal Escherichia coli). All baboons received 1011
markers also reveal a two-stage response? colony-forming units (CFU)/kg of E. coli intraperitoneally as a fibrin clot. Three different responses were
Global and Molecular Marker Tests in observed: complete recovery (Well); sustained illness over 14 days (Sick); and death within 48 hrs (Died).
the Human Model of Compensated Intra- This figure shows the fibrin degradation product (FDP) (left), the platelet (center), and the fibrinogen
venous Endotoxemia. Figure 3 (bottom (right) responses of animals in each of these three categories. WBC, white blood cell count.

S16 Crit Care Med 2000 Vol. 28, No. 9 (Suppl.)

two panels) shows the response of hemo- days, or death within 310 days. Although Figure 4 shows that with the excep-
static factors described previously in re- this model does not show the consistent tion of fibrin degradation products, the
lation to elastase, lactate production, and lethal responses favored by investigators global coagulation tests were of limited
cytokine production (top two panels). studying intervention strategies, it does value in discriminating between the three
The enzyme inhibitor complexes that re- offer the opportunity to evaluate the ca- groups over the 14-day period of obser-
flect production of thrombin, plasmin, pacity of potential diagnostic tests to dis- vation. This figure (top center panel)
and activated protein C all peaked during criminate between or predict which of shows that animals that were instru-
the symptomatic (stage 1) response to these three responses to E. coli might mented but not infected exhibited the
bolus infusion (4 ng/kg) of endotoxin. occur. We believed that this model might same fall in platelet counts as did the
Although soluble fibrin monomer and help us determine whether the global co- animals that were infected but remained
soluble tissue factor also began to rise at agulation or molecular marker tests are well. This finding also applied to fibrin
2 4 hrs into this stage, they did not reach of any value in assessing the status of the degradation products but to a lesser ex-
their peak concentrations until 1224 hrs reticuloendothelial system/microvascular tent. Fibrinogen, being an acute phase
after endotoxin challenge. Soluble throm- transport organ, particularly in those an- reactant and relatively insensitive marker
bomodulin followed a similar pattern to imals that have a prolonged/recurrent of DIC, is of value in this model only early
that of soluble tissue factor, peaking when systemic inflammation response syn- in the course of peritonitis.
factor VIIa reached its nadir. Surprisingly, drome-like response to E. coli, and if Figure 5 shows the same molecular
this occurred during the asymptomatic pe- there are any tests that could predict how markers shown in Figures 2 and 3. These
riod when subjects were already back at a given animal might respond to intra- animals again were observed over 14
work, long after the inflammatory and he- peritoneal E. coli. days. The well animals exhibited, with the
mostatic mediators began their return to-
ward baseline (stage 2). At no time were
there significant declines in platelet or fi-
brinogen concentrations.
Table 2 shows that subjects who were
classified into groups 1 or 2 based on the
severity of their symptoms also exhibited
elevated levels of markers of neutrophil,
endothelial, hemostatic, and metabolic/
liver dysfunction that corresponded with
the severity of clinical symptoms (rigors,
fatigue). This raises the possibility that
these molecular markers not only reflect
microvascular/endothelial/hemostatic
stress, but that they also may distinguish
between degrees of stress short of frank
decompensation or overt DIC as would be
reflected by a fall in platelet or fibrinogen
concentrations.
Global Coagulation and Molecular
Marker Tests in the Baboon Model of
Escherichia Coli Peritonitis. Although
the two intravenous models of E. coli
sepsis and endotoxemia outlined above
offer some perspective on the fidelity and
potential of various molecular markers,
neither model mimics what is most often
seen in hospital practice. Recently, we
have developed a conscious baboon
model of interperitoneal E. coli sepsis/
systemic inflammation response syn-
drome, which reflects the highly variable
courses we most commonly see in clini-
cal practice (3). The responses observed
in this model emphasize the importance Figure 5. Molecular markers (response to intraperitoneal Eshcerichia coli). All baboons received 1011
colony-forming units (CFU)/kg of E. coli intraperitoneally as a fibrin clot. Three different responses
of the host in determining which of sev-
were observed: complete recovery (Well); sustained illness over 14 days (Sick); and death within 48
eral different outcomes can arise after
hrs (Died). This figure shows the fibrin degradation product (FDP) (left), the platelet (center), and the
implantation of a fixed concentration of fibrinogen (right) responses of animals in each of these three categories. The left column shows the
E. coli organisms (1 1011 CFU/kg) into pattern of thrombin/antithrombin complex (TAT) () and soluble fibrin monomer (sFM) (solid
the peritoneal cavity. The three responses circles). The center column shows the responses of activated protein C/protein C inhibitor complex
include a full recovery in 23 days, a (APC/PCI) () and protein C (PC) (solid circles). The right column shows the responses of elastase/
sustained or recurrent illness over 14 alpha 1AT complex (ELAS) () and soluble thrombomodulin (sTM) (solid circles).

Crit Care Med 2000 Vol. 28, No. 9 (Suppl.) S17

 P
redictive tests
most likely will
come from assess-
ment of host cell responses
to in vitro lipopolysaccha-
ride stimulation. This pri-
mate model of peritonitis,
because it mimics the vari-
able responses seen in the
clinical practice, provides an
opportunity to determine
whether such an approach
will aid in predicting the out-
come in patients who are at
risk.

ever, as reflected by activated protein

C/protein C inhibitor (APC/PCI) complex
formation appears to coincide with the
severity of the clinical response (died
sick well).
Taken as a group, these molecular
markers offer a view of the status of the
Figure 6. Cytokine in RNA response to endotoxin of monocytes harvested before and 24 hrs after
reticuloendothelial system/microvascular
intraperitoneal Escherichia Coli. The figure shows cytokine (tumor necrosis factor [TNF], interleukin
(IL)-10, IL-12, IL-15, IL-18) messenger RNA responses of a baboon that died and one that survived the organ over the full course of the disorder
intraperitoneal instillation of 1011 colony-forming units (CFU)/kg E. coli in a fibrin clot. The solid and in the sick group of animals that cannot
open bars represent the responses of nonsurvivor and survivor animals, respectively. Top, the be obtained by clinical signs and global
responses of monocytic harvested before (T-0) instillation of intraperitoneal E. coli; bottom, the coagulation tests. These studies suggest
responses of monocytes harvested 24 hrs after instillation of intraperitoneal E. coli. that this combination of molecular mark-
ers could be clinically useful, particularly
if they were available for use at the bed-
exception of protein C, a relatively lim- with the late sFM response. This suggests side.
ited response of markers of hemostatic that there was sustained inflammatory As far as the question of predicting the
and neutrophil/endothelial perturbation, activity and endothelial injury in this sick response to an inflammatory stimulus of
after the implementation of 1 1011 group that may have accounted for the patients at risk, results from the assess-
CFU/kg of E. coli. The sick animals ex- continued production of sFM. Instrumen- ment of the in vitro response of mono-
hibited the familiar two-stage response of tation of these animals did not perturb cytes to endotoxin harvested from ba-
hemostatic factors (TAT and sFM). The these molecular markers. Figure 5 also boons before and early in the course of E.
sFM was especially helpful in this model shows the TAT and sTM responses of the coli peritonitis suggest that this approach
because by 72 hrs, temperature, white nonsurvivors were significantly greater merits further study. Figure 6 (top) sum-
blood cell count, and global coagulation than those of the surviving but sick ani- marizes the cytokine mRNA responses to
test responses differed very little from mals, whereas the response of sFM and endotoxin of 0.6 to 1.2 106 mono-
those of the well animals and because the elastase did not differ significantly be- cytes/mL harvested before induction of
TAT levels (although remaining slightly tween these two groups. Finally, the re- peritonitis from an animal that subse-
elevated) had already fallen toward base- sponse of protein C (anticoagulant) quently died and from one that lived. The
line. The sick animals also exhibited a reached a similar nadir of 45% to 50% in concentrations of mRNA were normal-
strong elastase response and sustained the well, as in the sick and nonsurvivor ized to beta actin (housekeeping gene).
elevation of sTM levels, which coincided groups. The activation of protein C, how- The bars represent the fold increase or

S18 Crit Care Med 2000 Vol. 28, No. 9 (Suppl.)

decrease of mRNA over baseline and are
the average of four to five blood samples
collected over a 4 5 day period just be-
fore induction of peritonitis. The cyto-
kine response pattern of the nonsurvivor
differs from that of the survivor. More
TNF (mediator cytokine) and IL-15 (reg-
ulator cytokine) were produced by the
survivor relative to the nonsurvivor. The
survivors response pattern was reversed
at 24 hrs after induction of peritonitis, at
which time less TNF and IL-12, and more
IL-10 and IL-15 were produced by the
survivor as compared with the nonsurvi-
vor. In both animals the circulating cyto-
kine levels at 24 hrs were below level of
detection (eg, 5 ng/mL TNF).
Because these are preliminary results,
one does not know how far this approach
will take us. However, it seems obvious
that there is a partial discrepancy be-
tween what we measure in plasma and
what is actually happening in tissues, al-
Crit Care Med 2000 Vol. 28, No. 9 (Suppl.)
though the plasma measurements are
valuable. This seems particularly true
when one considers the capacity of in-
flammatory cells such as monocytes and
macrophages to reprogram on the fly
such that under one set of conditions
they produce mediator and under an-
other set produce regulatory cytokines
(6). In our opinion, this suggests that
predictive tests most likely will come
from assessment of host cell responses to
in vitro lipopolysaccharide stimulation.
This primate model of peritonitis, be-
cause it mimics the variable responses
seen in the clinical practice, provides an
opportunity to determine whether such
an approach will aid in predicting the
outcome in patients who are at risk.
REFERENCES
1. Wada H, Wakita Y, Nakase T, et al: Increased
plasma-soluble fibrin monomer levels in pa-
tients with disseminated intravascular coag-
ulation. Am J Hematol 1996; 51:255260,
2. Gando S, Kameue T, Nanzaki S, et al: Partic-
ipation of tissue factor and thrombin post-
traumatic systemic inflammatory syndrome.
Crit Care Med 1997; 25:1820 1826
3. Kinasewitz GT, Chang ACK, Peer GT, et al:
Peritonitis in the baboon: A primate model
that simulates human sepsis. Shock 1999;
13:100 109
4. Freund M, Ravanat C, Schwartz C, et al:
Thrombin independent thrombocytopenia
and cleavage of platelet GPV during endo-
toxin shock in the rat. Abstract # 5130 pre-
sented at: XVII Congress of the International
Society on Thrombosis and Haemostasis, Au-
gust 17, 1999, Washington, DC
5. Gilbert JA Jr, Scalzi RP: Disseminated intra-
vascular coagulation. Emerg Med Clin North
Am 1993; 11:465 480
6. Majetschak M, Flach R, Kreuzfelder E, et al:
The extent of traumatic damage determines a
graded depression of the endotoxin respon-
siveness of peripheral blood mononuclear
cells from patients with blunt injuries. Crit
Care Med 1999; 27:313318
S19