Download acute systemic inflammation

ACUTE SYSTEMIC INFLAMMATION
IN
HEALTH AND DISEASE
Cover: Changes of TNF, IL-6, PCT, SAA, CRP, α1AT, WBC, PC, IgM and IgG respectively
after an acute inflammatory event.
Design by E. Nijsten based on an idea of T.E. Feltkamp
Electronic ISBN 90-367-1453-2
Printed ISBN 90-367-1454-0
© M.Nijsten 2001
RIJKSUNIVERSITEIT GRONINGEN
ACUTE SYSTEMIC INFLAMMATION
IN
HEALTH AND DISEASE
Proefschrift
ter verkrijging van het doctoraat in de
Medische Wetenschappen
aan de Rijksuniversiteit Groningen
op gezag van de
Rector Magnificus, dr. D.F.J. Bosscher,
in het openbaar te verdedigen op
woensdag 19 september 2001
om 16.00 uur
door
Maarten Willem Nicolaas Nijsten
geboren op 2 april 1960
te Scheveningen
2
Promotores
Prof. Dr. H.J. ten Duis
Prof. Dr. T.H. The
Referent
Dr. R.J. Porte
Beoordelingscommissie Prof. Dr. R.J. Goris
Prof. Dr. R. van Schilfgaarde
Prof. Dr. L.G. Thijs
Paranymfen
H. Makkinga
Dr. B. de Smet
Aan Anja,
Susan en Peter
Abbreviations:
α1AT
α1AP
APACHE-II
APP
C3a
C3dg
C4
CD-14
CRP
DIC
EDTA
ELISA
ESR
EURICUS-II
FES
GP IIb/IIIa
HGF
ICU
IgG, IgM
IL-1
(rh)IL-6
ISS
KD
LPS
MODS
MPS
NASCIS
PC
PC2
∆PC/∆t
∆PC/∆t0→2
∆PC/∆t2→10
PCT
PFO
SAA
SIRS
SOFA
TBSA
(rh)TNF
WBC
α1-antitrypsin, acute phase protein
α1-antiprotease inhibitor, acute phase protein, identical with α1AT
acute physiological and health evaluation score, ICU score
acute phase protein
activation product of complement factor 3
rest product after activation of C3
complement factor 4
lipopolysaccharide receptor
C-reactive protein, acute phase protein
disseminated intravascular coagulation
ethylenediamine tetraacetic acid, an anticoagulant
enzyme linked immuno sorbent assay
erythrocyte sedimentation rate
European ICU studies
fat embolism syndrome
glycoprotein IIb/IIIa receptor, mediates platelet adhesion and aggregation
hybridoma growth factor (identical with IL-6)
intensive care unit
immunoglobulin G, immunoglobulin M
interleukin-1, cytokine
(recombinant human) interleukin-6, cytokine
injury severity score
kilodalton
lipopolysaccharide, major bacterial endotoxin component
multiple organ dysfunction score, ICU score
methyl prednisolone sodium succinate
national acute spinal cord injury studies
platelet count
platelet count at admission day 2
daily rate of change in platelet count
daily rate of change in platelet count between admission day 0 and 2
daily rate of change in platelet count between admission day 2 and 10
procalcitonin
patent (or persisting) foramen ovale
serum amyloid A, acute phase protein
systemic inflammatory response syndrome
sequential organ failure score, ICU score
total body surface area
(recombinant human) tumor necrosis factor
white blood cell count
CHAPTER 1
INTRODUCTION AND SCOPE OF THIS THESIS
Introduction
INTRODUCTION
Wounds, healing and inflammation
When the skin is damaged, for example by a small cut, a blood clot will start forming within seconds,
and within minutes bleeding will stop. We should realize that the process of wound repair has then
already started. Under normal circumstances, healing will be the end result of overlapping processes:
inflammation, tissue formation and tissue remodelling [1]. Inflammation is a response of the immune
system, since the immune system guards the individual against infection. Inflammation is defined in
Webster's Medical Dictionary as "A local response to cellular injury that is marked by capillary
dilatation, leukocyte infiltration, redness, heat, pain, swelling, and often loss of function and that
serves as a mechanism initiating the elimination of noxious agents and damaged tissue[2]."
Since minor wounds are sustained very regularly, the local inflammatory responses initiated by the
immune system are as crucial part of body homeostasis as is feeding or sleeping. Inflammation in an
extremely complex homeostatic response that involves neutrophils, platelets, macrophages,
endothelial cells and the coagulation and complement systems. The immune system can be divided
into the innate immune system and the adaptive immune system. Furthermore, the system has been
divided into a humoral and a cellular system. Combined, these two divisions result in the four major
parts (Table 1.1) of the immune system.
Table 1.1. Major divisions in the immune system
Humoral
Innate immune system
Adaptive immune system
Cytokines, complement,
Immunoglobulins
acute phase proteins
Cellular
Phagocytes, Natural killer
T-lymphocytes
cells
This thesis focuses on a number of aspects of the innate humoral immune response. The early
inflammatory response is part of the innate immune system. In contrast to the adaptive immune
system, which provides improved responses upon repeated infection, the innate response is not based
on immune memory, but on evolutionary highly preserved mechanisms of pattern recognition of
potential pathogens. For example, bacterial endotoxin (lipopolysaccharide; LPS) is recognized by
the innate immune system in even very low concentrations, and thus serves as a very strong stimulus
of subsequent responses. Acting together, CD-14 and so-called Toll-like receptors (e.g. TLR-4)
present on myeloid cells, are believed to form a single pathway common to all mammals to
transduct the LPS-signal [3]. The importance of these Toll-like receptors is underscored by the
discovery that the receptors are coupled to a pathway that activates genes mediating innate
immune defenses in mammals, insects, and even plants [4,5].
2
Chapter 1
Homeostasis
recovery
disturbance
response
stimulus
Immune response phases
III
III
II
I
Amplification
Induction
Humoral
Cellular
Cellular
Figure 1.1. To preserve a stable healthy situation (homeostasis), like other physiologic systems, the
immune system reacts to disturbances such as a trauma with phased responses. The purpose of these
responses is to eliminate the disturbance and restore homeostasis. The induction phase the presence of
tissue damage or infection (the stimulus) is recognized and transducted into cytokine or chemokine
signals that direct the response that eliminates the injury. (Adapted from T.H.The et al. 1995)
Whereas local inflammation in the setting of tiny wounds may not have significant systemic effects,
at a certain size of the wounds, systemic effects will become evident.
Clinical signs of systemic inflammation
The systemic manifestations of the innate immune response that result after trauma, burns or ICUadmission are the subject of this thesis. When an inflammatory stimulus is sufficiently strong to
cause systemic effects (Table 1.2), the so-called systemic inflammatory response syndrome (SIRS)
will develop. SIRS has been defined by a consensus conference [6] as present if two of the clinical
manifestations listed in Table 1.3 are observed. At the ward many patients may have SIRS, at the
intensive care unit nearly all patients have SIRS.
Grades and definitions of systemic inflammation
3
Introduction
The American consensus statement contained, apart from the definition of SIRS three additional
definitions. The main purpose of these definitions has been to illuminate the distinction between
inflammation and infection:
SIRS= fever + leukocytosis;
Sepsis= SIRS + infection;
Severe Sepsis= Sepsis + multiorgan dysfunction;
Septic shock= Severe Sepsis + refractory hypotension [6].
Table 1.3. SIRS (Systemic Inflammatory Response Syndrome) criteria
[6]
Temperature >380 or <360 C
Heart rate > 90 /min
Respiratory rate > 20
Hyperventilation (PaCO2 < 4.3 kPa)
Leukocytes > 12 109/l or < 4 109/l
Immature neutrophils > 10%
Organ failure
Multiple organ failure has been defined in 1985 by Goris as 'generalized, autodestructive inflammation' [7]. As more organs are affected by the systemic inflammatory process, the chances of
survival decrease.
Two examples of how organ failure has been graded are shown in table 1.4, a minimal total score has
a high probability of survival, a maximal score has a very low probability of survival.
Biochemical signs of systemic inflammation - acute phase proteins
Although not strictly defined, the acute phase response can be considered to consist of the initial
systemic clinical and biochemical responses that follow for trauma or infection. Fever is the clinical
hallmark of the acute phase response, and it is usually accompanied by tachycardia [8]. These
clinical signs are paralleled by many biochemical changes. Many of the proteins induced by the
innate immune response are called acute phase proteins.
An elevated erythrocyte sedimentation rate (ESR) [9] was the first widely used laboratory
Table 1.4. Examples of two multiple organ failure scores
Organ/system
Goris’ Multiple organ failure
score [7]
Cardiovascular
0–2
Pulmonary
0–2
Renal
0–2
Neurologic
0–2
Liver
0–2
Hematologic/ coagulation
0–2
Gastrointestinal
0–2
Total score
0 – 14
4
Sequential organ failure
assessment (SOFA) [52]
0–4
0–4
0–4
0–4
0 – 4\a
0-4
0 - 24
Chapter 1
parameter of inflammation. Increased levels of the acute phase protein fibrinogen are a main
determinant of an elevated ESR [10]. A large number of proteins have turned out to be acute phase
proteins (APP). What constitutes an acute phase protein has not clearly been defined. An
increased levels after inflammation and some sort of effector function (as opposed to signalling
cytokines) are the most important characteristics of acute phase proteins. Kushner [11] somewhat
sweepingly defined APP as those proteins whose plasma concentration rises 25% or more
following an inflammatory stimulus. Albumin, the most abundant protein in the circulation, is
sometimes called a negative acute phase protein since it is down-regulated during inflammation.
The metabolic effects and the reasons why albumin levels are decreased, are only partly
understood [12]. Although not subject of this thesis, decreased albumin levels are an important
illustration of the relation between inflammation and metabolism.
Nowadays, C-reactive protein (CRP) is the most widely directly measured acute phase protein. In
health it is hardly detectable in plasma, but with inflammation it can rise rapidly to 10- or 100-fold
levels. Although CRP’s function is not clearly identified, CRP is probably involved in early nonspecific antibacterial defences [13]. Thanks to reliable assays and CRP’s half-life which is in the
order of one day, CRP has gained wide popularity for monitoring inflammation in many disease
states.
On the basis of the rapidity and magnitude of increase in concentration during the acute phase
response, acute phase proteins can be divided into 3 groups (Table 1.5).
Source of acute phase proteins
The liver is quantitatively and qualitatively the major source of acute phase proteins. In several
hepatocyte models for different species, the induction of many acute phase protein-production has
been demonstrated. For example in 1966 Hurlimann et al. already performed in vitro studies that
showed that hepatocytes produce CRP [15].
Table 1.5. Classification of some acute phase proteins
Increase by factor 1.5
Slow response
Increase by factor 2 to 4
Intermediate response
Increase by up to factor 1000
Rapid response
Antitrombin III
Complement C3, C4
Ceruloplasmin
C1-inhibitor
α2-antiplasmin
Adapted from Heinrich et al. [14]
α1-proteinase inhibitor
α1 acid glycoprotein
α1 anti-chymotrypsin
Haptoglobin
Fibrinogen
C-reactive protein (CRP)
Serum Amyloid A (SAA)
5
Introduction
Function of acute phase proteins
A close look at the functions - as far as understood - of the acute phase proteins illustrates the
purpose of the acute phase response: provide the organism with those proteins that are necessary to
keep up both an adequate and a restrained inflammatory response. Table 1.5 shows that proteins for
the complement system, the coagulation system and the fibrinolytic system are provided by the acute
phase response. Most acute phase proteins have larger molecular weights (>50,000 Dalton) and
higher concentrations (mg/L to g/L range) when compared to cytokines which have a lower
molecular weight and much lower concentrations. This is due to fact that acute phase proteins are
often effector molecules (e.g. thrombus formation, protease inhibitor, bactericidal factor) as
opposed to cytokines that are hormone-like proteins. Understandably, effector functions require
much higher concentrations than necessary for signalling functions.
Function unknown or no function
Currently it is possible to detect many biological substances and study their interactions with many
cell types. This results in a huge number of possible relations that can be studied. It is helpful to try
to understand these mechanisms in a teleological manner. Teleology is the use of design or purpose
as an explanation of natural phenomena [2]. In understanding inflammation this implies that
evolution should have provided the organism with defenses that can plausibly by expected to help the
organism survive. A mammal will certainly have a survival benefit from adequate protection against
minor trauma, since every animal will continuously sustain such traumas during its life. But an
animal cannot logically be expected to have an elaborate defense against the results of septic shock.
An animal with septic shock or severe trauma will in all probability not survive anyway, so evolution
has no interest in providing for defenses in ‘lost cases’. Of course the animal (or man) with septic
shock may survive with medical intervention, but at this stage we cannot rely on inflammation
responses behaving in a ‘logical’ way. Thus, we may expect every protein found near a small
uncomplicated wound to have a relevant function, and search aggressively for this function if we do
not know it yet. We also may understand why TNFα levels are very high in septic patients, but we
should not expect that these high levels have a specific function. In fact in these patients many
molecules with elevated or depressed levels will not have a function, may have lost their function or
may even function inappropriately.
That a distinct degree of systemic inflammation might be beneficial to the host, was already believed
centuries ago [16,17], when patients who did not properly recover from wounds or infection were
treated by inducing additional inflammation. The widely practiced cauterization of wounds with hot
irons or hot oil was also performed on healthy skin to enhance recovery of wounds elsewhere (see
illustration). This practice of counter-irritation was also performed in the 20th century, albeit in the
slightly more humane form of turpentine abscesses. In fact turpentine in still used in animal models to
induce a sterile acute phase response [18].
6
Chapter 1
History of endogenous pyrogen and IL-6
The genesis of fever has long been a subject of investigation. Decades ago it became apparent that an
'endogenous pyrogen' had to exist. This substance is made by the body in response to tissue damage
or infection and it induces fever by changing the temperature setpoint in the hypothalamus - a process
that can be inhibited by prostaglandin inhibitors like aspirin. With modern molecular biological
technology is became clear that several cytokines not only induce fever but also induce acute phase
proteins. They can thus be considered not only endogenous pyrogens but turned out to be true
Figure 1.2. Patient who is cauterized after injury. Drawing by Johannes
Wechtlin in Feldtbuch der Wundarznei by Hans von Gersdorff, 1540.
hormones of inflammation.
In the 1980’s interleukin-1 (IL-1) became the first contender for the role of endogenous pyrogen.
Experiments by Aarden et al. showed that hybridoma growth factor (HGF) was active in the IL-1
7
Introduction
assay [19]. A very sensitive bio-assay for HGF was developed, and this assay did not respond to IL-1,
indicating that HGF had to be another cytokine. After cloning of cDNA, HGF was discovered to be
identical with a new cytokine named interleukin-6 [20]. In many in vitro and in vivo studies that
followed, the B9.9 bioassay for HGF/IL-6 has been very important in defining the pleiotropic role of
IL-6, to be later replaced by ELISAs [21].
The cytokines TNFα, IL-6 and IL-1 are elevated within hours after a major inflammatory stimulus
[22]. TNFα peaks earlier than IL-6 and also induces IL-6. For example van Gameren and
colleagues [23] showed that administration of IL-6 to patients induced most of the responses listed in
Tables 1.2 and 1.3. Several studies have assessed the predictive value of cytokine levels for clinical
outcome. In general, cytokine levels in patients with sepsis are much higher than in trauma patients
[24]. In addition it was found by Hack et al. that higher IL-6 levels were associated with mortality in
sepsis patients [25]. Partly due to the fact that cytokine levels can rapidly change, and due to
difficulty in routinely assaying cytokines, these measurements are not used in clinical practice. IL-6
levels measured at the bedside (positive at > 1000 pg/ml) have been used to triage patients for antiTNFα sepsis trials [26].
Methodological advantages of studying patients with mechanical trauma or burns
The pathophysiology of inflammation has received intensive attention in patients that are critically
ill, but it is useful as well to study the physiology of inflammation in patients that are not critically ill,
e.g. in healthy persons that sustain a nonlethal trauma. Trauma patients usually are healthy before the
injury and show a responses roughly proportional to the severity of the injury [27]. Since the time of
the injury is a well-defined moment as opposed to sepsis, the timing and order of the elicited
responses can be studied in more detail.
Concerning inflammatory responses in trauma patients, patients with burns represent a special group.
Thermal injury is a one-dimensional trauma, with the percentage of burned total body surface
(TBSA) being the key parameter. With only the age of the patient and TBSA it is possible to make a
good prediction of mortality and hospital stay [28]. Severely burned patients can have extensive
tissue damage without accompanying organ damage. Early after sustaining burn injury, patients
show a pronounced inflammatory response. This occasionally extreme response has made burns
patients a subject of many studies on virtually all known aspects of host responses.
Differentiating infectious from non-infectious causes of inflammation.
Both trauma and infection can induce an acute phase reaction, clinically characterized by SIRS.
When SIRS persists for a number of days after admission to the hospital, this can be due to the
trauma itself or secondary infection. In the first case a wait-and see policy is often justified. In the
second case it may be crucial to search for a focus, or to start antibiotics and even perform an
operation. Since a majority of trauma patients at the ICU develop fever, with only proven
infection in half the these patients [29] - it is very useful to possess additional tools to recognize
bacterial infection. Especially in immune suppressed patients (e.g. after transplantation) it would
be very useful to possess a rapid assay to detect bacterial infection, since in these patients the
margins of error are small. Now that many inflammatory markers can be measured in the
circulation, researchers have looked for those markers that can discriminate bacterial infection
8
Chapter 1
from other causes of systemic inflammation. But discriminating causes of inflammation on
patterns of cytokine or acute phase protein responses have proven an elusive goal. Measuring the
bacterial product endotoxin itself would seem to be an obvious solution to identify bacteremia and
endotoxemia early [30]. Unfortunately circulating endotoxin assays are difficult to use and
reproduce [31].
In accordance with the concept of the spectrum of increasing systemic inflammatory responses
with increasing infection: SIRS → sepsis → severe sepsis → septic shock, systemic inflammation
with infection is associated with higher cytokine levels and higher levels of CRP. Thus, serial
quantitative measurements of the extent of the acute phase response, usually in the form of CRP
have been a major clinical tool in detecting infection. The acute phase response appears to
modulate mainly in intensity and duration, not in the relative enhancement or inhibition of
components of the response. This is of course in agreement with the non-specific nature of the
acute phase reaction.
Procalcitonin to differentiate infection and non-infection?
The recent introduction of a convenient assay for procalcitonin (PCT) and the first experiences
with measurements of circulating PCT have been promising. Some authors have proposed that a
practical parameter that differentiates bacterial infection from other causes of inflammation is now
at hand. Although PCT is biochemically a precursor of calcitonin (a hormone involved in calcium
homeostasis) it is functionally not related to calcitonin. Both the cellular origin and function of
PCT are unknown. Nevertheless, circulating PCT measurements in many patients groups [32]
have shown that:
- Circulating PCT is elevated proportional to the inflammatory stimulus.
- PCT can be induced by infusing endotoxin in volunteers. Elevated PCT-levels are detectable
at 6 hours and disappear with a half-life of about a day [33].
- The dynamic range of PCT is very large: PCT rises at least a factor 400 within 6 hours in
volunteers receiving endotoxin. Where one report [34] claims that PCT levels are <0.01 ug/l in
normal persons, PCT can reach 1000 ug/l in sepsis [35]. Thus PCT's dynamic range may be
even greater than that of CRP or SAA.
The most interesting aspect of PCT is the evidence that PCT compared to CRP is superior in
discriminating bacterial from non-bacterial inflammation. De Werra [36] compared patients with
septic shock, cardiogenic shock, and bacterial pneumonia. The best predictive value for septic
shock was found for PCT in septic shock, with PCT varying from 72 to 135 ng/mL, compared
with approximately 1 ng/mL in the other groups. The monograph on PCT by Meisner [32]
contains a number of early studies that indicate that PCT is better than CRP in discriminating
serious infection from other causes of inflammation. More recently Gendrel et al. reported that
PCT was more specific and sensitive than CRP for the differentiation of bacterial and viral
infections in children [37]. In 236 children with viral infections they found a median PCT of 0.2
ug/l and a median CRP of 10 mg/l. In 46 children with bacterial sepsis or meningitis PCT was 18
ug/l and CRP 144 mg/l. Thus PCT is about a 100-fold higher in sepsis compared to viral infection,
whereas CRP is 'only' 15 times higher.
9
Introduction
Still the literature is not conclusive on this very important issue. Those disease categories where
determining PCT may not be reliable need to be better defined. For example de Bont [38] found
that in neutropenic patients PCT responses are much lower than would be expected on the basis of
other signs and elevated CRP-levels. The various results that suggest that preferential induction of
PCT by bacterial infection exists, have led some authors to hypothesize that endotoxin can induce
PCT directly. A logical question is whether cytokines like TNFα or IL-6 are necessary and
sufficient for the induction of PCT. And although increased levels of IL-6 and TNFα were also
observed after endotoxin infusion, this does not prove that TNFα and IL-6 are necessary for the
induction of PCT synthesis, as they are for the acute phase protein induction. Does PCT originate
in the liver? Or more general, should PCT be viewed as an acute phase protein?
Platelets and endothelium in inflammation
Platelets can be considered inflammatory agents as much as coagulatory agents. In systemic
inflammation platelets first disappear from the circulation (thrombocytopenia), and as the patient
recovers uneventfully platelets will reappear in increased amounts [39] (thrombocytosis). Several
cytokines can induce this sequence. After administration to humans, IL-6 will induce a complete
acute phase response, including the characteristic sequence of a nadir platelet count at day 2 or 3
and a maximal platelet count at day 12 [23]. Concerning the causes of premature disappearance of
platelets from the circulation, two important possibilities exist. The platelet end up as part of a clot
or the platelet can (temporarily) adhere to the endothelium. In both cases platelets can amplify
inflammation through the release of powerful mediators.
Consumption of platelets is an integral part of the syndrome of diffuse intravascular coagulation
(DIC). The reverse, that DIC always accompanies platelet consumption, is not true. DIC is
characterized by extensive intravascular formation of fibrin, in severe cases resulting in vascular
occlusion and organ failure. DIC occurs secondary to diverse range of serious diseases, like sepsis
or trauma [40]. It is important to note that significant decreases in platelet count are observed in
virtually all patients with trauma or sepsis. Although a significant proportion of these patients may
have indicators of 'low-grade' DIC, many do not have DIC, indicating that platelet sequestration
and DIC are not the same [41].
Endothelium, the single cell layer that separates the blood from the organs and tissues, is a very
active cell system. Endothelium regulates hemodynamics and the transport of molecules and cells
between blood and the tissues. Activation of endothelium occurs early and universally [42] in the
process of inflammation. Endothelial cell activation [43] is characterized by the loss of vascular
integrity, expression of adhesion molecules, transition from an antithrombotic to a prothrombotic
state and cytokine production. Exposure of the subendothelium with tissue factor and von
Willebrand Factor and rapid expression of P-selectin can initiate the binding of platelets as well as
the adhesion molecules such as GPIIb/IIIa, P-selectin, CD-31, LFA-1, CD-36 [44] and CD-87
(urokinase plasminogen activator receptor; uPAR) [45]. Cardiovascular research has produced
extensive evidence of the importance of platelet-endothelium interaction. This is underscored by
the clinical effectiveness of aspirin and the recently introduced GPIIb/IIIa-inhibitors in limiting or
preventing coronary thrombosis [46]. Despite the evidence of the importance of the plateletendothelial interaction in inflammation, direct studies on this interaction are methodologically
10
Chapter 1
very difficult. Ex vivo, endothelial cells and platelets are not the same as in vivo. Whereas the
erythrocyte has a circulating precursor that can be clearly identified and quantified (the
reticulocyte), no such equivalent exists for the circulating platelet. Therefore, in thrombocytopenia
it is difficult to decide if decreased synthesis or increased aggregation or adhesion is present, even
when one resorts to a bone marrow biopsy or radioactively labeled platelet studies [47]. Reinjected radioactively labeled autologous platelets are not identical to native circulating platelets.
Serial platelet counts as an indicator of endothelial activation
Instead of performing highly specialized platelet studies in a limited set of patients, in this thesis
is was preferred to study serial platelet counts and study them in larger patient groups. The
assumption was that changes in the platelet count are to a large extent correlated to the magnitude
of endothelial activation, which in turn is related to systemic inflammation. A low admission
platelet count is known to be a relatively strong predictor of adverse outcome in a variety of
disease states. In meningococcal sepsis [48], after ruptured aortic aneurysm [49], at admission to
the intensive care [50] the early platelet count is one of the strongest predictors of outcome. As a
result the platelet count has been introduced in several intensive care scoring systems, while the
leukocyte count has been eliminated from some scoring systems. The multi-organ dysfunction
score (MODS) [51], the sequential organ failure assessment score (SOFA) [52] and the pediatric
risk of mortality score (PRISM-III) [53] use the platelet count and not the leukocyte count as one
of its component parameters. In contrast to the well-described importance of initial platelet
counts, the relevance of subsequent changes has received little attention. Within the healthy
individual the platelet count is quite stable with an intra-individual variation that is only 20% to
30% (or 60 ·109/L) of the inter-individual variation [54,55]. Thus serial platelet counts might in
principle provide additional information. As long as the platelet count is within the normal range
(i.e. 150 to 350 ·109/L) it has been implicitly assumed that such a count is normal. Yet, in patients
with an uneventful clinical recovery after a moderate trauma, thrombocytosis normally develops
during the second week. Thrombopoietin has been identified as a major regulator of
thrombopoiesis [56]. After trauma IL-6 is released, which is also capable of inducing
thrombocytosis [23]. In critically ill patients thrombocytosis is often not present, although such
patients typically have cytokine levels (including IL-6) much higher than those observed after
uncomplicated trauma. Thus despite the fact that many ICU patients are in a more advanced state
of inflammation (ranging from SIRS to sepsis, severe sepsis or even septic shock) compared to
uncomplicated trauma patients (usually SIRS), they have lower platelet counts. Since in many
instances these decreased platelet counts are still in the 'normal range' these changes have received
little attention. That platelet counts decrease as a result of increased endothelial activation may be
the explanation of the inverse relation between inflammation and platelet count in critically ill
patients.
11
Introduction
Fat embolism syndrome
This thesis originated in studies on the incidence and causes of the fat embolism syndrome (FES).
FES is typically seen in young patients who sustain isolated long bone fractures [57]. After an
interval of 8 hours to 2 days the three cardinal symptoms of the syndrome appear:
• petechiae with a typical upper-body distribution, unlike that seen in severe thrombocytopenia
• respiratory distress
• cerebral disturbances
As the name of the syndrome denotes, fat is involved in these three cardinal symptoms, since fat
emboli of bone marrow origin can be recovered from skin lesions, the lung and the brain during
pathological examination. In the process of embolization, the venous fat apparently easily
(by)passes the lungs. Aggressive, early operative stabilization of fractures, and the improved
level of supportive care are assumed to have contributed to the decrease in FES-incidence.
Although FES in seen less frequently than for example 25 years ago, both its causes and the
reasons for its decreased incidence are unclear. Fever and tachycardia are among the 'minor
symptoms' described for FES [57], indicating a link of FES with inflammatory responses. Instead
of being a result of FES, systemic inflammation might be a factor in inducing FES. CRP can
agglutinate fat globules into emboli [58] - a property of CRP that has even led to the development
of a bed-side CRP-test based on fat-agglutination [59]. Here we looked at the relation of early
inflammatory signs with the subsequent development of FES.
Prolonged increases in pressures in (closed) wounds around the fractured bone may assist the
process of fat intravasation and subsequent embolization. Early operation may decompress the
fractured bone and prevent further embolization of fat. The potential detrimental effect of delayed
treatment on the incidence of FES, and the relevance of a persisting open foramen ovale [60] as a
shunt for fat globules are explored in this thesis.
12
Chapter 1
OUTLINE AND AIMS OF THIS THESIS
In this thesis the systemic counterparts of acute local inflammation were studied. As a part of the
innate immune response, systemic manifestations of inflammation were primarily studied in
trauma patients
Hypothesis
The various stages of the systemic inflammatory response are expressed by distinct markers (proinflammatory cytokines as well as acute phase proteins and platelet counts) in a typical time
sequence. They reflect the underlying pathophysiological mechanism and may offer possibilities
for monitoring (new) intervention strategies by providing better clinical prognostic and diagnostic
tests.
The clinical model used to study these marker kinetics were trauma patients since this patient
category allows to note the starting point of the (pathophysiologic) chain of events.
Trauma patients are young and in general have no comorbity. By a better understanding of the
time sequence of the marker responses in trauma patients with single injuries physician may
subsequently better interpret responses in critically ill patients, most of whom have preexistent comorbidity.
Aims of this thesis
To study the various stages of the systemic inflammatory response, we selected a number of
(circulating) markers that we assumed would adequately reflect and be correlated with
(patho)physiological signs and symptoms. The protein responses of IL-6, PCT, CRP, SAA and
α1-antiproteinase as well as the platelet count and leukocyte count and physiological parameters
as temperature or heart rate were studied.
The first studies concern:
- Fat embolism syndrome (chapters 2 and 3)
The subsequent chapters are ordered according to the sequence of inflammatory events:
- Interleukin-6 (chapters 4 and 5)
- Procalcitonin (chapter 6)
- Platelet counts (chapters 7, 8 and 9)
Two studies (chapters 7 and 9) were an intervention studies; all studies were retrospective in
design, although in chapters 4, 5, 6 and 9 some samples and data were prospectively collected.
Not studied in this thesis were: animals, local responses, adaptive immune responses.
Fat embolism syndrome
Chapter 2 is a retrospective study of the incidence of FES in 172 patients with an isolated fracture
of the femoral shaft. The goal was to find associations of the incidence of FES with the type of
fracture, the timing of operation and an early inflammatory response.
Chapter 3 verifies if a right-to-left shunt in the heart has a causal role in FES. Such a shunt could
explain the transit of large fat globules from the venous to the arterial circulation.
13
Introduction
Interleukin-6 and acute phase responses
Chapter 4 describes IL-6 levels in patients with burns at a time that IL-6 had never been
measured in patients with acute systemic inflammation. The aim was to correlate IL-6 levels with
basic acute phase responses: i.e. fever, CRP-levels and α1-antiproteinase levels, and address the
question if IL-6 was the long sought endogenous pyrogen.
Chapter 5 examines IL-6 and acute phase responses in burns patients in detail. The goal was to
study time dependent changes of phenomena in which IL-6 could play a causal role. Also
correlations of these parameters with IL-6 and the potential causal relations on the basis of
published evidence (as it was published at that time) are formulated.
Procalcitonin
Chapter 6 it is assumed that endotoxin is not necessary for the induction of PCT. Thus, the direct
effects of TNFα and IL-6 on the expression of PCT, SAA and CRP were measured. In vitro,
human liver slices were used. In vivo, two groups of cancer patients, treated with TNFα and IL-6
respectively, were studied.
Primary and secondary thrombocytopenia
Early and late decreased platelet counts in trauma and ICU patients may reflect increased
sequestration due to systemic endothelial activation.
The aim in chapter 7 was to examine early changes in platelet count as they occur during the first 2
days after trauma. We also investigated if high-dose methylprednisolone administered shortly after
the injury affected platelet consumption. This may indicate if the pleiotropic effects steroids are able
to affect inflammation related platelet sequestration.
In chapter 8 the aim was relate late changes (i.e. > 2 days) in platelet count with mortality, since
persisting systemic inflammation is known to be associated both with platelet sequestration and
mortality. Patients from one surgical ICU were studied. The value of platelet counts was also
compared to leukocyte counts.
The goal in Chapter 9 was to generalize the observations of the previous study and address changes
in platelet counts and outcome in a very large heterogeneous European multi-center population of
ICU patients. The predictive power for mortality of early and late changes in platelet count
respectively, was investigated. For this purpose a simple mathematical model to describe timedependent changes in platelet counts was used. We also investigated the behavior of platelet counts
according to admission groups (medical, surgery scheduled or unscheduled.)
Finally Chapter 10 attempts to integrate the results and give suggestions for further research. In
particular the timing of all responses observed in the various studies is combined into one scheme.
The question what would constitute an ideal inflammatory marker is discussed, as well if such a
marker exists.
14
Chapter 1
REFERENCES
1.
Singer AJ, Clark RA. Cutaneous wound healing. N Eng J Med 1999;341:738-746.
2.
Webster's Medical Desk Dictionary. Merrian-Webster. 1986.
3.
Hirschfeld M, Ma Y, Weis JH, Vogel SN, Weis JJ. Cutting edge: Repurification of
lipopolysaccharide eliminates signaling through both human and murine Toll-like receptor 2. J
Immunol 2000 165:618-622.
4.
Means TK, Golenbock DT, Fenton MJ. The biology of Toll-like receptors. Cytokine Growth
Factor Rev 2000;11:219-32.
5.
MacKay I, Rosen FS. Innate immunity. N Eng J Med 2000;338-344.
6.
American College of Chest Physicians/Society of Critical Care Medicine consensus conference:
definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis.
Crit Care Med 1992;20:864-74.
7.
Goris RJ, te Boekhorst TP, Nuytinck JK, Gimbrere JS. Multiple-organ failure. Generalized
autodestructive inflammation? Arch Surg 1985;120:1109-1115.
8.
Karjalainen J, Viitasalo M. Fever and cardiac rhythm. Arch Intern Med 1986;146:1169-71
9.
Brigden M. The erythrocyte sedimentation rate. Still a helpful test when used judiciously. Postgrad
Med 1998;103:257-274.
10.
Bedell SE, Bush BT. Erythrocyte sedimentation rate. From folklore to facts. Am J Med
1985;78:1001-1009.
11.
Kushner I. The phenomenon of the acute phase response. Ann N Y Acad Sci 1982;389:39-48.
12.
Haupt W, Holzheimer RG, Riese J, Klein P, Hohenberger W. Association of low preoperative
serum albumin concentrations and the acute phase response. Eur J Surg 1999;165:307-13.
13.
Szalai AJ, Agrawal A, Greenhough TJ, Volanakis JE. C-reactive protein: structural biology and
host defense function. Clin Chem Lab Med 1999;37:265-270.
14.
Heinrich PC, Castell JV, Andus T.Interleukin-6 and the acute phase response. Biochem J 1990
265:621-636.
15.
Hurlimann J, Thorbecke GJ, Hochwald GM. The liver as the site of C-reactive protein formation. J
Exp Med 1966 123;1966:365-378.
16.
Gillies HC. Theory and practice of counter-irritation. MacMillan Oxford 1896.
17.
van Gool J. Farewell speech. May 18 1990. University of Amsterdam.
18.
Leon LR, Kozak W, Peschon J, Kluger MJ Exacerbated febrile responses to LPS, but not
turpentine, in TNF double receptor-knockout mice. Am J Physiol 1997;272:R563-9
19.
Helle M, Boeije L, Aarden LA IL-6 is an intermediate in IL-1-induced thymocyte proliferation. J
Immunol 1989;142:4335-4338 .
20.
Brakenhoff JP, de Groot ER, Evers RF, Pannekoek H, Aarden LA. Molecular cloning and
expression of hybridoma growth factor in Escherichia coli. J Immunol 1987;139:4116-4121.
21.
Helle M, Boeije L, de Groot E, de Vos A, Aarden L. Sensitive ELISA for interleukin-6. Detection
of IL-6 in biological fluids: synovial fluids and sera. J Immunol Methods 1991;138:47-56
22.
Hack CE, Aarden LA, Thijs LG Role of cytokines in sepsis. Adv Immunol 1997;66:101-95
15
Introduction
23.
van Gameren MM, Willemse PHB, Mulder NH, et al.:Effects of recombinant human interleukin-6
in cancer patients: a phase I-II study. Blood 1994; 84:1434-1441.
24.
Martin C, Boisson C, Haccoun M, Thomachot L, Mege JL. Patterns of cytokine evolution (tumor
necrosis factor-alpha and interleukin-6) after septic shock, hemorrhagic shock, and severe trauma.
Crit Care Med 1997;25:1813-1819.
25.
Hack CE, De Groot ER, Felt-Bersma RJ, Nuijens JH, Strack Van Schijndel RJ, Eerenberg-Belmer
AJ, Thijs LG, Aarden LA. Increased plasma levels of interleukin-6 in sepsis. Blood 1989;74:17041710.
26.
Abraham E. Why immunomodulatory therapies have not worked in sepsis. Intensive Care Med
1999;25:556-566.
27.
Giannoudis PV, Smith MR, Evans RT, Bellamy MC, Guillou PJ. Serum CRP and IL-6 levels after
trauma. Not predictive of septic complications in 31 patients. Acta Orthop Scand 1998;69:184-188
28.
Vico P, Papillon J. Factors involved in burn mortality: a multivariate statistical approach based on
discriminant analysis. Burns 1992;18:212-215.
29.
Hene IZ, Bendien SA, Nijsten MWN, ten Duis HJ, Zwaveling JH. Fever and severe trauma. Intensive
Care Med 1998;24 Suppl1:S11.
30.
van Deventer SJ, Buller HR, ten Cate JW, Sturk A, Pauw W Endotoxaemia: an early predictor of
septicaemia in febrile patients. Lancet 1988;i (8586):605-9.
31.
Novitsky TJ. Limitations of the Limulus amebocyte lysate test in demonstrating circulating
lipopolysaccharides. Ann N Y Acad Sci 1998;851:416-21
32.
Meisner M. Procalcitonin. A new innovative infection parameter. Biochemical and clinical aspects.
Berlin: Brahms Diagnostica 1996. ISBN 3-00-000803-9
33.
Dandona P, Nix D, Wilson MF, Aljada A, Love J, Assicot M, Bohuon C. Procalcitonin increase
after endotoxin injection in normal subjects. J Clin Endocrinol Metab 1994;79:1605-1608.
34.
Snider RH Jr, Nylen ES, Becker KL. Procalcitonin and its component peptides in systemic
inflammation: immunochemical characterisation. J Investig Med 1997; 45:552-560.
35.
Assicot M, Gendrel D, Carsin H, Raymond J, Guilbaud J, Bohuon C. High serum procalcitonin
concentrations in patients with sepsis and infection. Lancet 1993; 341:515-518.
36.
de Werra I, Jaccard C, Corradin SB, Chiolero R, Yersin B, Gallati H, Assicot M, Bohuon C,
Baumgartner JD, Glauser MP, Heumann D. Cytokines, nitrite/nitrate, soluble tumor necrosis factor
receptors, and procalcitonin concentrations: comparisons in patients with septic shock, cardiogenic
shock, and bacterial pneumonia. Crit Care Med 1997;25:607-613.
37.
Gendrel D, Raymond J, Coste J, Moulin F, Lorrot M, Guerin S, Ravilly S, Lefevre H, Royer C,
Lacombe C, Palmer P, Bohuon C. Comparison of procalcitonin with C-reactive protein, interleukin
6 and interferon-alpha for differentiation of bacterial vs. viral infections. Pediatr Infect Dis J
1999;18:875-881.
38.
de Bont E, Vellenga E, Swaanenburg J, Visser-van Brummen P, Kamps W. Procalcitonin used as a
marker of bacterial infection in neutropenic cancer patients with fever. Infection 2000;28:398-400.
39.
Griesshammer M, Bangerter M, Sauer T, Wennauer R, Bergmann L, Heimpel H. Aetiology and
clinical significance of thrombocytosis: analysis of 732 patients with an elevated platelet count. J
Int Med 1999;245:295-300.
40.
16
Levi M, ten Cate H. Disseminated Intravascular coagulation. N Eng J Med 1999;341: 586-92.
Chapter 1
41.
Neame PB, Kelton JG, Walker IR, Stewart IO, Nossel HL, Hirsch J. Thrombocytopenia in
septicemia: the role of disseminated intravascular coagulation. Blood 1980;56:88-92.
42.
Lentsch AB, Ward PA Regulation of inflammatory vascular damage. J Pathol 2000;190:343-348.
43.
Hunt BJ, Jurd KM. Endothelial cell activation. A central pathophysiological process.
BMJ
1998;316:1328-1329
44.
Mannel DN, Grau GE. Role of platelet adhesion in homeostasis and immunopathology. Mol Path
1997;50:175-185.
45.
Piguet PF, Vesin C, Donati Y, Tacchin-Cottier F, Belin D, Barazzone C. Urokinase receptor
(uPAR,CD-87) is a platelet receptor important for TNF-induced endothelial adhesion in mice.
Circulation 1999;99:3315-3321.
46.
Cannon CP. Incorporating platelet glycoprotein IIb/IIIa inhibition in critical pathways: unstable
angina/non-ST-segment elevation myocardial infarction. Clin Cardiol 1999;22(Suppl):IV30-36.
47.
Sigurdsson GH, Christenson JT, el-Rakshy MB, et al: Intestinal platelet trapping after traumatic and
septic shock. An early sign of sepsis and multiorgan failure in critically ill patients? Crit Care Med
1992;20:458-467.
48.
van Deuren M, Neeleman C, van 't Hek LGFM, van der Meer JWM. A normal platelet count at
admission in acute meningococcal disease does not exclude a fulminant course. Intensive Care Med
1998;24:157-161.
49.
Bradbury AW, Bachoo P, Milne AA, Duncan JL. Platelet count and the outcome of operation for
ruptured abdominal aortic aneurysm. J Vasc Surg 1995;21:484-491.
50.
Bonfiglio MF, Traeger SM, Kier KL, Martin BR, Hulisz DT, Verbeck SR. Thrombocytopenia in
intensive care patients: a comprehensive analysis of risk factors in 314 patients. Ann Pharmacother
1995;29:835-842.
51.
Marshall JC, Cook DJ, Christou NV, Bernard GR, Sprung CL, Sibbald WJ. Multiple organ
dysfunction score: a reliable descriptor of a complex clinical outcome. Crit Care Med
1995;23:1638-1652.
52.
Vincent JL, de Mendonca A, Cantraine F, et al. Use of the SOFA score to assess the incidence of
organ dysfunction/failure in intensive care units: results of a multicenter prospective study. Crit Care
Med 1998; 26:1793-1800.
53.
Pollack MM, Patel KM, Ruttimann UE. PRISM III: An updated pediatric risk of mortality score.
Crit Care Med 1996;24:743-752.
54.
Buckley MF, James JW, Brown DE, Whyte GS, Dean MG, Chesterman CN, Donald JA. A novel
approach to the assessment of variations in the human platelet count. Thromb Haemost
2000;83:480-484.
55.
Ross DW, Ayscue LH, Watson J, et al: Stability of hematologic parameters in healthy subjects.
Intraindividual versus interindividual variation. Am J Clin Pathol 1988;90:262-267.
56.
Kaushansky K. Thrombopoietin. N Engl J Med 1998;339:746-754.
57.
ten Duis HJ. Fat embolism syndrome. Injury 1997;28:77-85.
58.
Rowe IR, Soutar AK, Pepys MB. Agglutination of intravenous lipid emulsion ('Intralipid') and
plasma lipoproteins by C-reactive protein. Clin Exp Immunol 1986;66:241-247.
17
Introduction
59.
Richter D, Rohricht AM, Nurnberger W, Wahn V, Schroten H. The fat emulsion agglutination test:
a reliable and cost effective alternative to the latex agglutination test for rapid bedside CRP
measurement. Clin Chim Acta 1997; 261:141-148.
60.
Webster MW, Chancellor AM, Smith HJ, Swift DL, Sharpe DN, Bass NM, Glasgow GL. Patent
foramen ovale in young stroke patients. Lancet 1988;ii:11-12.
18
CHAPTER 2
FAT EMBOLISM IN PATIENTS WITH
AN ISOLATED FRACTURE OF THE FEMORAL SHAFT
H.J. ten Duis, M.W.N. Nijsten, H.J. Klasen, B. Binnendijk
J Trauma 1988; 28: 383-390
Fat embolism in patients with an isolated fracture of the femoral shaft
ABSTRACT
Analysis of basic pathophysiologic variables in fat embolism patients is often restricted by the
complexity of the different injuries present in each individual patient. To avoid this problem we
investigated the presence of the fat embolism syndrome in patients with an 'isolated' fracture of the
femoral shaft. Two groups were distinguished: those who had an open fracture or a closed fracture
treated operatively within 24 hours after the accident (decompression group), and those who were
treated initially conservatively (non-decompressed group). Clinical fat embolism occurred only in
patients in the non-decompressed group (3.5%). They showed significantly higher temperatures,
lower pulse rates, a progressive hemoglobin decrease, and a fracture localization more proximal
(p<0.025) than the other patients in the non-decompressed group; they also showed pathophysiologic
patterns significantly different from the patients in the decompressed group. Although the
pathophysiologic mechanism of the onset of clinical fat embolism remains unclear, initial
temperature elevations in combination with 'typical' fracture localization and fracture type appear to
have a predictive value.
INTRODUCTION
The fat embolism syndrome is considered to be a symptom complex of acute respiratory distress,
cerebral disturbances, and petechiae after one or more long bone fractures [1]. Many researchers have
investigated the pathophysiology of this syndrome, though essential etiologic items remain unclear.
Neither quantitative nor qualitative studies on fat embolization have been able to indicate consistent
risk factors for the fat embolism syndrome. The clinical picture of full-blown fat embolism may
develop after a single long bone fracture, although the incidence increases with the number of
fractures. The investigation of a category of patients with single injuries enabled us to perform an
analysis of possible etiologic factors without being influenced by accompanying injuries.
We studied a group of patients with an isolated fracture of the femoral shaft to determine which
factors might predispose to the development of the fat embolism syndrome. Especially fracture site,
fracture type and degree of comminution were analyzed and also minor fat embolism features like
pyrexia, tachycardia, and thrombocytopenia.
20
Chapter 2
MATERIALS AND METHODS
In the period 1967 through 1985, 172 consecutive patients, varying from 16 and 65 years of age, with
an 'isolated' fracture of the femoral shaft, were studied from the moment they arrived at the
Table 2.1
Accident cause in isolated fractures
of the femoral shaft
Cause
N
%
Fall
18
11
Pedestrian
3
2
Bicycle
16
9
Autobike
72
42
Motorcycle
9
5
Automobile
30
17
Hit by heavy object
7
4
Miscellaneous
17
10
Driver/passenger of
Department of Traumatology at the University Hospital Groningen. Individuals with moderate and
severe accompanying injuries were excluded, while
patients with one minor additional injury (patella
fracture, fracture of the clavicle or less) were included
in the study. The investigation was performed on
patients with a closed as well as an open fracture of
the femoral shaft. Patients with a pathological fracture
of the femoral shaft were excluded from this
investigation.
Figure 2.1. Division of the femoral shaft into
seven equal segments.
Each patient was examined daily for the symptoms of
fat embolism. Records of temperature, pulse and
blood pressure were maintained as well as water and
salt balance. The following investigations were performed daily for 6 days: supine chest X-ray,
arterial blood gas analysis, hemoglobin, hematocrit, platelet- and white blood count, coagulograms,
fibrinogen concentration levels, and serum elektrolyte and triglyceride determinations. The X-rays of
the femur, taken in the emergency room, were used to determine fracture localization in the femoral
shaft, degree of comminution and fracture type. For fracture localization the shaft was divided into
seven equal segments according to Kootstra [2] (Fig. 2.1).
The following fracture types were distinguished (we made use of the fracture quotient Q=x/y;
x:length of fracture; y:width of femoral shaft at place of fracture): transverse Q=1.0-1.1; short
oblique Q=1.2-1.3; oblique Q=1.4-1.7; longitudinal Q>1.7; spiral. The degree of comminution was
21
Fat embolism in patients with an isolated fracture of the femoral shaft
noted: none, slight comminution with two or more fragments of less than 7 mm, moderate
comminution with fragments from 7 mm to 2 cm, and serious comminution with multiple fragments
greater than 2 cm with or without fragments of smaller size. Open fractures were treated, preferably
with open reduction and internal fixation, on the day of the accident. In patients with a closed
fracture, the fracture was reduced and immobilized in skeletal traction within 6 hours after
admission. During the period 1967-1972 most closed fractures were treated conservatively; in the
period 1972-1985 open reduction and internal fixation from day 10 onwards was usually the first
choice of treatment. The diagnosis of fat embolism was made when at least two major symptoms as
described by Gurd [3] were observed: 1) petechial rash, 2) respiratory distress with bilateral clinical
and/or radiologic signs of pulmonary involvement, and 3) evidence of cerebral involvement unrelated
to head injuries (primarily excluded). Two main groups of patients were distinguished: those who
had a closed fracture and were initially (10 days) treated conservatively (non-decompressed group)
and patients who had an open fracture, or a closed fracture but were treated initially (within 24 hours)
operatively (decompression group).
No corticosteroids, antipyretics, or antiphlogistic medicaments were given in the initial phase of
treatment. Student's t-test for unpaired values was used in the statistical evaluation of the results.
100
90
80
Number of patients
70
60
50
40
30
20
10
0
16-20
21-25
26-30
31-35
36-40
41-45
46-50
51-55
56-60
61-65
Years
Figure 2.2. Ages of the patients at the time of the accident.
RESULTS
Of the 172 patients who satisfied the conditions of this study, 133 were males (77%) and 39 (23%)
were females. The accident causes are given in Table 2.1. The average delay between accident and
time of admission was 45 ± 35 minutes. The distribution of ages is given in Fig. 2.2. None of the
patients showed any sign of shock at admission or during the initial phase of treatment.
22
Chapter 2
There were 21 open and 151 closed fractures. Twenty-two patients with closed fractures were
treated operatively on the day of the accident. Six patients (3.5%) were diagnosed as having fat
embolism (Table 2.2), of whom four were given supplementary oxygen, and two patients had to be
treated with mechanical ventilation. These patients all belonged to the primarily conservatively
treated group with a closed fracture (129 patients). Neither the patients with an open fracture nor the
patients with a closed fracture, initially treated operatively, developed two or more major signs of
clinical fat embolism.
The fat embolism patients showed a fracture localization in segment two or three of the femoral shaft
(mean 2.3). The average localization in the other patients was significantly more distal (mean 4.2;
p<0.025). Fracture localization and degree of comminution are given in Table 2.3.
Temperature recordings revealed significantly higher temperatures on the day of the accident and day
1 in the fat embolism patients compared to the other patients in the non-decompressed group (p<0.01,
day 0; p<0.005, day 1). Low initial temperatures were found in the decompressed group compared to
Table 2.2.
Major fat embolism symptoms present in patients of non-decompressed
and decompression group.
Non-decompressed
Decompressed
FES (6)
No FES (123)
(43)
Petechial rash
4
1
0
Pulmonary dysfunction
5
4
0
Mental dysfunction
4
1
0
the non-decompressed group (p<0.01) (Fig. 2.3).
In the fat embolism group the initial pulse rate was as low as in the remaining patients of the nondecompressed group, but increased to significantly higher values during the rest of the observation
period (p<0.025). Pulse rate values in the decompressed group generally followed the temperature
pattern (Fig. 2.4).
All patients showed comparable decreases in hemoglobin and hematocrit values in the first 24 hours
after the accident. Significantly lower values were observed in the fat embolism group from day 2
onwards (p<0.025) compared to the remaining patients in the non-decompressed group. This
decrease was comparable to those of the open and/or primarily operatively treated patients (p<0.001)
(Fig. 2.5).
23
Fat embolism in patients with an isolated fracture of the femoral shaft
All groups showed a temporary decrease in platelet count during the observation period. Platelet
counts were slightly lower in the fat
embolism group (p<0.05, day 1) (Fig.
Table 2.3.
2.6)
and
in
the
primarily
Fracture localization and degree of comminution in
decompressed group.
patients in the fat embolism group, non-decompressed
Arterial blood gas analysis revealed
group and decompression group.
about equal values directly after the
Fat embolism
Non
Decompression
trauma in all groups. The fat
decompression
embolism
patients
showed
a
N S M S N S M S N S M S
significant decrease in PaO2 and
o
l
o e o l
o e o l
o e
n
i
d
r
n
i
d
r
n
i
d r
oxygen saturation levels on day 1 after
e
g e i
e g e i
e g e i
the trauma (p<0.001) (Fig. 2.7). The
h r o
h r o
h r o
t
a
u
t
a
u
t
a u
fat embolism group had a PaO2 of 8.0
t
s
t
s
t
s
kPa and an oxygen saturation of 90%,
e
e
e
Segmen
compared to 11.2 kPa and 96% for the
t
non-decompressed group and 14.1
0 0 0 0 6 1 1 1 0 0 1 0
1
1 0 2 1 4 7 8 1 0 2 2 1
2
kPa and 97% in the decompressed
0 1 0 1 7 9 11 1 0 3 4 1
3
group. After (symptomatic) treatment
0 0 0 0 7 6 6 0 3 1 7 3
4
of the pulmonary dysfunction in the
0 0 0 0 10 6 13 2 2 0 3 1
5
0 0 0 0 3 2 6 0 0 1 0 1
6
fat embolism patients, this group
0 0 0 0 1 0 2 0 1 0 2 2
7
reached similar PaO2 and oxygen
1 1 2 2 38 31 47 5 6 7 19 9
total
saturation levels to the other groups
around day 2 after the trauma. No
important differences between the groups were observed in leukocyte counts, electrolytes, or BUN
and creatinine levels, plasma protein, and albumin, as well as coagulograms. All patients survived
and left the hospital in good health.
DISCUSSION
Fat embolism is a syndrome which includes the symptoms of respiratory failure, sensorium
disturbances, and petechiae. The onset of the syndrome is unpredictable and there are no hard
indications known which invariably
24
Chapter 2
39.0
39.0
Hemoglobin
140
8
120
Hemoglobin
9
140
8
38.0
mmol/l
g/l
38.0
Temperature
38.5
9
degrees Celcius
g/l
38.5
degrees Celcius
160
Temperature
mmol/l
160
120
7
7
37.5
37.5
100
100
6
6
37.0
37.0
0
1
2
3
4
0
5
80
5
0
1
2
3
4
1
2
3
4
3
4
5
80
5
5
0
1
2
5
Figure 2.3. Mean with standard error of the mean of temperature values in patients of the fat embolism group (open
squares), non-decompressed group (open circles) and decompressed group (filled circles) during the 6 day period
Figure
2.5. Mean
with
standard error
of the differences
mean of hemoglobin
patients of the fat embolism group (open
following
trauma.
Statistically
significant
are markedvalues
with ainstar.
squares), non-decompressed group (open circles) and decompressed group (filled circles) during the 6 day period
following trauma. Statistically significant differences are marked with a star.
120
300
120 300
Heart Rate
Heart Rate
Platelet count
110
110
250
250
200
90
10E9/l
100
/minute
10E9/l
/minute
100
200
90
150
80
70
100
Platelet count
150
80
0
0
1
1
2
2
3
3
4
4
5
5
70 100
0
0
1
1
2
2
3
3
4
4
5
5
Figure
Figure2.4.
2.6.Mean
Meanwith
withstandard
standarderror
errorofofthethemean
meanofofheart
platelet
rate counts
values in patients of the fat embolism group (open
squares),
squares),non-decompressed
non-decompressedgroup
group(open
(opencircles)
circles)and
anddecompressed
decompressedgroup
group(filled
(filledcircles)
circles)during
duringthe
the6 6day
dayperiod
period
following
followingtrauma.
trauma.Statistically
Statisticallysignificant
significantdifferences
differencesare
aremarked
markedwith
witha star.
a star.
lead to its development. The incidence - relatively rare in patients with single fractures - increases
with the number of fractures or injuries [4].
In this study six cases of clinical fat embolism (3.5%) were observed. This percentage is comparable
to other authors [5], considering the fact that this study only included solitary fractures. Clinical fat
25
15
15
14
14
13
13
12
12
11
11
KPa
KPa
Fat embolism in patients with an isolated fracture of the femoral shaft
10
9
10
9
PaO2
8
PaO2
8
7
7
6
6
5
5
0
1
2
3
4
5
0
1
2
3
4
5
Figure 2.7. Mean with standard error of the mean of arterial pO2 values in patients of the fat embolism group (open
squares), non-decompressed group (open circles) and decompressed group (filled circles) during the 6 day period
following trauma. Statistically significant differences are marked with a star.
embolism developed in patients who had a closed fracture and had not been operated on within 24
hours (non-decompressed group). In contrast, patients with an open fracture or those who had been
operated upon early (within 24 hours; decompression group) using open reduction and internal
fixation showed no clear evidence of pulmonary or cerebral disturbances or petechiae. This
discrepancy between decompressed and non-decompressed patients, especially in the patients with
closed fractures who were operated upon within 24 hours, suggests that induction phenomena for the
onset of fat embolism occur in the first 24 hours. In this regard a well known factor is pulmonary
embolization of marrow fat caused by fat globules, which are invariably found in the venous
circulation after injury [5]. Although authors have tried to correlate the amount of fat globules with
the incidence of fat embolism, it is unlikely that the syndrome is caused by this mechanism only.
None of the patients with an open fracture or with a closed fracture who had been operated upon
within 24 hours showed any evidence of fat embolism symptoms. Therefore, bone marrow
embolization alone is unlikely to provoke fat embolism but it may compound the effect of other
stresses in an injured patient to induce clinically apparent pulmonary problems. Especially local
tissue ischemia by fracture hematoma may play an important role. Opening of the fracture evacuates
the fracture hematoma and reduces tissue pressure around the fracture. In support of this
(decompression) hypothesis is the investigation by Kallio in 1941 [6,7], who described a lower
incidence of fat embolism in patients with open fractures. Concomitantly, this might be the reason
why cases of postoperative clinical fat embolism were observed after acute 'closed' intramedullary
nailing of the femoral shaft fractures [8].
Is the fat embolism syndrome related to fracture type or localization? Until now less attention has
been paid to this criterion. All fat embolism patients in this series had a fracture localization in
segment two or three of the femoral shaft, just cranial of the isthmus. At this place there several
venous sinusoids within the marrow with large capillaries which do not have any wall musculature.
26
Chapter 2
The walls of these veins are very fragile here and are only embedded in depots of marrow fat [4].
This might explain why marrow fat just in this place might be released into the venous circulation.
The fat embolism patients were shown to have a short oblique or oblique fracture with a moderate
degree of comminution (except the eldest patient with a spiral fracture) especially of fragments up to
2 cm. These fracture characteristics are often related to a certain degree of energy impact and are
designated as high-energy injuries. Consequently, this moderate degree of bone marrow destruction
will invariably be accompanied by fracture hematoma formation, soft-tissue injury, and
consequently, in closed fractures, by local tissue hypoxia.
Interesting in this regard was that differences in blood loss reflected in hemoglobin and hematocrit
decrease could not be found during the first 24 hours between the fat embolism group and the nondecompressed group. Significant differences were detected from day 2 onwards and seem therefore
not related to the initial amount of hematoma formation.
The temperature pattern in fat embolism patients generally showed significantly elevated values
within 24 hours after the accident, before any sign of pulmonary insufficiency was noticed. Here
there was a significant difference to the main (non-decompressed) group with a highest mean of
37.9°C and 37.6°C in the group with any form of decompression. This pyrexia is a well known
feature in patients with clinically apparent fat embolism, and it is probably induced through the
release of interleukin-1 (endogenous pyrogen) by activated macrophages [9]. Fat globuli or toxic
products released from the local ischemic tissues could serve as activators. The low initial pulse rate
values (day of accident) in the fat embolism patients are of special interest because a rise in
temperature is normally accompanied by an increase in pulse rate.
During the entire 6-day observation period the decompressed group was the least compromised
concerning respiratory dysfunction. As expected, arterial PaO2 and oxygen saturation were low at the
onset of the fat embolism syndrome, although on admission differences between the groups could not
be detected. As only two patients had to be treated by mechanical ventilation, the degree of
respiratory distress was limited. Following the assumption that the pulmonary distress is an integral
part of complement-induced PMN and platelet activation, the moderate degree of respiratory distress
is in accordance with the only slight decrease in platelet numbers and the absence of severe
disturbances in coagulograms.
The form of fat embolism described in this article is considered the nonfulminant subacute or
classical form classified by Sevitt [11]. The fulminant fat embolism syndrome, which is often
misdiagnosed and develops suddenly within a few hours after the accident, with an often fatal
outcome was not observed in our selective series. It is unlikely that there will be any predictable sign
for this form of fat embolism.
To our knowledge the patients described represent the first large series of patients investigated for the
occurence of fat embolism in a 'standardized' injury.
In this series a temperature increase to levels of 38.5°C or higher within 24 hours combined with a
fracture of the femoral shaft just proximal to the isthmus seems to have some predictive value. If a
patient shows these symptoms within 24 hours after the accident, one should be aware of the danger
of a fat embolism syndrome during the following few days.
27
Fat embolism in patients with an isolated fracture of the femoral shaft
REFERENCES
1.
Peltier LF. Fat embolism: a current concept. Clin Orthop 1969;66:241-253.
2.
Kootstra G. Femoral shaft fractures in adults. A study of 329 consecutive cases with a statistical
analysis of different methods of treatment. Thesis, Assen, the Netherlands, van Gorcum, 1973.
3.
Gurd AR. Fat embolism: an aid to diagnosis. J Bone Jt Surg 1970;52-B;732-737.
4.
Gossling HR, Pelligrini VD Jr. Fat embolism syndrome: a review of the pathophysiology and
6.
physiological basis of treatment. Clin Orthop 1982;165:68-82.
Allardyce DB, Meek RN, Woodruff B, Cassim MM, Ellis D. Increasing our knowledge of the
pathogenesis of fat embolism:a prospective study of 43 patients with fractured femoral shafts. J
Trauma 1974;14:955-962.
Editorial comment to (7) J Trauma 1982;22:894.
7.
Riska EB, Myllynen P. Fat embolism in patients with multiple injuries. J Trauma 1982;22:891-894.
8.
King KF, Rush J. Closed intramedullary nailing of femoral shaft fractures: a review of 112 cases
5.
treated by Kuntscher technique. J Bone Jt Surg 1981;63-A:1319-1323.
9.
Dinarello CA. Interleukin-1. Rev Infect Dis 1984;6:51-95.
10.
Manning JB, Bach AW, Herman CM, Carrico CJ. Fat release after femur nailing in the dog.
J Trauma 1983;23:322-326.
11.
28
Sevitt S. Fat embolism. London, Butterworths, 1962.
CHAPTER 3
FAT EMBOLISM AND PATENT FORAMEN OVALE
M.W.N. Nijsten, J.P.M. Hamer, H.J. ten Duis, J.L. Posma
Part of this chapter was published as a Letter in
the Lancet 1989; i: 1271
Fat embolism and patent foramen ovale
INTRODUCTION
Fat is an integral constituent of the soft tissues and the bone marrow. After skeletal trauma, bone
marrow emboli can rapidly enter the circulation. Especially long bone fractures are known to
generate circulating, abnormally large fat globules that sometimes have diameters>10 µm, in
contrast to the normal fat that circulates in the form of small lipoproteins [1].
Acute massive fat embolism
As forensic physicians know, a dead person with multiple fractures found three floors below an
open window should have his lungs examined. If fat emboli are present in his lung he was still
alive at the moment he hit the ground. However, if no fat emboli or marrow emboli are present he
may have been dead before he fell, and thus may have been murdered [1]. The sometimes
hyperacute nature of this embolization process is illustrated by a large post-mortem study done
on aircraft accident victims. In 1979 a New Zealand jetliner crashed into Mount Erebus on
Antarctica, killing all persons aboard. Detailed post-mortem analysis [2] in 205 of the 257 victims
showed that despite an nearly instantaneous death, many had fat emboli (65%) and bone marrow
emboli (29%) in their pulmonary capillary beds.
In a different study of 56 patients who died within 24 hours of blunt injury (including long bone
fractures) 68% had significant pulmonary fat emboli upon post-mortem examination [3]. Acute
massive fat embolism, as described above, should not be confused with classic fat embolism
syndrome (FES) as defined below.
Classic fat embolism syndrome (FES)
Some trauma patients can develop the fat embolism syndrome (FES). The classical fat embolism
syndrome is generally defined by three major symptoms that occur with a delayed onset of hours
to days after the trauma [4]:
• A petechial rash on the anterior chest, face, axillae, conjunctivae and other non-dependent
parts in general.
• Respiratory distress with diffuse bilateral abnormalities on the chest X-ray.
• Cerebral disturbances varying from mild to coma.
In addition to these three major criteria aspecific minor signs are usually observed, the most
important being fever, tachycardia and thrombocytopenia.
Whereas acute massive fat embolism is often diagnosed post-mortem in very severely injured
patients who rapidly succumb, patients with FES will usually survive. Apart from severity, the
crucial difference between these two entities is the delayed onset that is seen in FES. In apparently
similar patients with long bone fractures, only a minority will actually develop FES after the
symptom-free period.
Two central problems in FES-research are :
• What is the precise origin and what are the effects of abnormal circulating fat that is observed in
most patients after trauma?
32
Chapter 3
•
Why does only a minority develop FES?
Origin of circulating fat
Two main theories on the genesis of circulating fat emboli exist: the mechanical and the chemical
theory [1,4]. The mechanical theory explains the strong association between long bone fractures and
FES through the entrance of bone marrow fat into the circulation. The chemical theory [5,6] is
inspired by observations that fat emboli can occur in patients who have no trauma, such as patients
treated by steroids or patients with sickle cell disease [7]. Also some biochemical changes can lead to
the agglutination of fat and fat emboli, as has been observed in the early days of total parenteral
nutrition [8]. The acute phase protein CRP can agglutinate fat - a property that has even been used for
a new rapid bedside CRP-assay [9].
Although the mechanical theory is now the most popular, the mechanical and chemical models are
not mutually exclusive and may act in concert. The mechanical theory may not only involve fat
embolization from the bones, but also from the soft tissues, as confirmed in a recent study [3].
Predisposing factors for FES
It has proven very difficult to predict who will develop FES. Many investigators have wondered why
FES only occurs in a minority of apparently similar patients. In a previous study we identified
delayed stabilization of femoral fractures and early development of fever as a risk factors for the
development of FES [10]. When venous fat particles reach the arterial circulation, the lungs must
have been (by)passed somehow. In addition to exchanging oxygen and carbon dioxide with the
surroundings, the lung has an important filtering function. The lungs continuously interact with
and filter many particles that would otherwise enter the systemic arterial circulation. This includes
thrombo-emboli, bone marrow emboli, air emboli and tumor emboli.
The most prevalent occurring right-to-left shunt at the level of the heart is the patent foramen ovale.
The presence of a patent foramen ovale (PFO) as detected by echocardiography is strongly increased
in young patients who sustained ischaemic stroke [11,12] and divers who developed decompression
sickness [13]. By analogy, it could be hypothesized that a patent foramen ovale might allow entry of
fat globules into the systemic circulation. Might patients with a history of post-traumatic fat
embolism have a patent foramen ovale? We investigated the presence of PFO by echocardiography
in patients who developed FES after sustaining isolated skeletal trauma.
33
Fat embolism and patent foramen ovale
PATIENTS AND METHODS
We selected otherwise healthy patients with a history of fat embolism syndrome after isolated longbone fractures. Of all trauma patients with isolated fractures of the long extremities that were seen at
our hospital between 1968 and 1985, 12 met this condition. In a retrospective series of 172 patients
with an isolated femoral fracture admitted to our hospital between 6 developed FES [10]. An
additional 6 patients with FES were identified after screening patients with tibial, combined tibial and
femoral or bilateral femoral fractures without significant additional injury [4]. To detect the presence
of PFO, transesophageal color-coded Doppler echocardiography with a Toshiba 65A 3.75 MHz
transducer was performed as an outpatient procedure. The Valsalva maneuver was performed during
echocardiography to maximize the sensitivity for detecting a shunt across a patent foramen ovale.
Table 3.1.
Patients who developed FES and were later investigated by
trans-esophageal echocardiography.
Male/Female
Mean Age (+/-SD)
Tibial fracture
Isolated femoral fracture
Bilateral femoral fracture
5/1
24 +/- 11
1
4
1
RESULTS
Of the total of 12 patients with diagnosed FES, 6 patients (Table 3.1) volunteered to undergo
echocardiography. Echocardiography was performed without complications, and resulted in
adequate imaging of the atria in all cases. In 5 patients no right-to-left shunt was present. 1 patient
had a minute right-to-left shunt during the Valsalva maneuver. This short-lasting and barely
detectable shunt was of no pathophysiological significance.
DISCUSSION
The fact that no patients in our series had relevant right-to-left shunting as measured with a sensitive
technique [14], contrasts with the reported frequencies of PFO in patients who sustained idiopathic
ischaemic stroke and serious air embolism (50% and 61% respectively) [11-13]. Whereas PFO
obviously is of major importance in stroke and air embolism, the absence of a relevant foramen ovale
in the patients we studied makes it unlikely that the presence PFO has a causal role in the
development of FES. Although our series of FES patients is small, the observed incidence of 0 out of
6 makes PFO very unlikely as a necessary factor for the development of FES. It is no surprise that
we found no relevant PFO in patients who had well-defined FES. In fact an important role of
PFO in the pathogenesis of FES has long ago already been proposed and challenged [1]. In postmortem investigations in seven patients who died with FES, Sevitt found no PFO in any of the
patients [15]. With the availability of trans-esophageal echocardiography, assessing PFO after
FES was a logical step. Our study is the first that confirms this observation in patients who are
still alive (long) after recovering from FES.
34
Chapter 3
Another argument against a causal role of PFO in FES is the presence of circulating fat globules
in the majority of trauma patients [16], while the population incidence of PFO is below 50% even
when measured according to the most sensitive criteria. Thus systemically circulating fat is
present regardless of a PFO. The fat emboli somehow (repeatedly) pass the lungs. Possibly the fat
emboli deform to pass the lungs' capillaries (Fig 3.1). In the majority of patients with skeletal
injury fat excretion in the urine can be shown according to some investigators [17], underscoring
the ease with which fat globules reach the systemic circulation, and pass the glomerular filter as
well.
Pell's case report in 1993
The discussion of the relevance of PFO would probably have been closed, had not a paper on this
subject appeared in the New England Journal Medicine [18]. The editorial [19] accompanying
this brief report was even optimistically titled "Unravelling the fat embolism syndrome". In their
paper Pell and colleagues suggested a major role of PFO in the pathogenesis of FES on the basis
of a report on a single patient who died shortly after intramedullary nailing of a femoral shaft
fracture. Transesophageal echocardiography was performed during the operation. Showers of
echogenic masses were seen passing through the right heart. Even a lump of 7 cm was seen.
Material was seen passing through a PFO. The fatal outcome the authors vividly describe is one
of acute cor pulmonale with secondary paradoxical embolism. The floating objects imaged by
transesophageal echocardiography, must have been large thrombi or bone marrow fragments.
Finally, the patient described was not previously healthy, and he did not display the characteristic
symptom-free interval.
The case report by Pell illustrates the devastating consequences of acute cor pulmonale in patients
who often are already in a compromised cardiopulmonary condition. Unfortunately, the patient
described in the case report in the New England Journal of Medicine did not have typical FES.
Contrary to what might be inferred from the report's title, this patient did in all likelihood not die
from FES or from PFO, and he certainly did not show the classic picture of FES. Whereas the
genesis of FES is not understood, acute right heart failure in patients where embolic material is
dislodged during operative procedures is well-known and well-understood. In fact today more
patients may die from this form of acute pulmonary embolism than patients who die from the
classical FES.
Pell later published a series of 24 patients who were monitored with TEE per-operatively [20]:
small emboli were observed in 14 patients; in 6 patients moderately large emboli (up to 10 mm)
and in 4 patients large emboli >10mm were observed. The dimensions of the observed emboli
underscore that 70mm is very large indeed for an embolic 'lump' as was described in the case
report [17].
Animal models of FES
One of the difficulties in investigating FES is finding an appropriate animal model. A correct
animal model should include a trauma or at least a simulation of the effects of trauma, a
subsequent interval that is relatively free of signs that is later followed by organ manifestations of
fat emboli in affected organs. Despite the fact that many animal experiments designed to model
35
Fat embolism and patent foramen ovale
fat embolism have been carried out, no model reproduces the events observed in humans. Either
the animals rapidly succumbed from traumatic shock [21], or immediate pulmonary edema was
induced by fatty acid injury [22,23], or unrealistically massive neutral fat infusions resulted in
obstructive shock [24]. None of the models reproduce the intriguing 'incubation period' that is
observed in classical FES.
Decreasing incidence of FES
Although FES undoubtedly remains a real entity [25], the incidence of clinically manifest FES
has decreased to such extent that many clinicians have never seen a patient with the classical FES.
Probably a combination of changes in modern management of trauma patients has contributed to
this decrease [26]. Except for the more rapid stabilization of fractures [27] is not clear which
other changes have been decisive in reducing the incidence of FES. Potential factors are: more
aggressive fluid infusions, early application of positive end-expiratory pressure (PEEP) or the
routine early administration of low-molecular weight heparins. Although this decreased incidence
of manifest FES may be good news, observational studies into the causes of FES are hampered by
the low incidence of the syndrome. Thus it may be wise to direct research efforts to detect
subclinical FES, for example with modern imaging tools.
Imaging in FES
In a further advance in real-time monitoring of embolic events, Edmonds recently reported on the
use of transcranial Doppler to detect fat or air emboli during total hip arthroplasty [28]. Embolic
signals in the middle cerebral artery were found in 8 of the 20 patients that were monitored. The
number of embolic signals in these patients varied from 1 to 200 - in all cases coinciding with the
impaction of a cemented component or after hip relocation. Another useful application of realtime imaging is the use of peroperative TEE to compare two different techniques of artificial hip
fixation for their potential to generate emboli.
Whereas conventional cementing generated emboli in 95% of the patients, a new 'bone vacuum'
technique resulted in only 5% emboli [30]. Such studies have also shown that patients with preexisting disease are especially prone for complications from iatrogenic emboli. CT-scans of the
lungs have shown that the radiological picture of FES [31] differs from acute pulmonary edema or
ARDS. CT-scanning and magnetic resonance imaging of the brain has shown lesions caused by
cerebral FES [32], indicating that although patients survive FES, it is may leave permanent
defects.
36
Chapter 3
CONCLUSION
The pathogenesis of FES is still not understood, but patent foramen ovale is unlikely to be
relevant in FES. Early inflammatory responses and especially timing of the operation are probably
important in FES, although the mechanisms are not clear. Investigating FES in its isolated form in
patients with solitary long bone injuries has become more difficult due to the decreased incidence
of FES.
Figure 3.1 Illustration from Zenker [29] showing a fat droplet deformed into
sausage shape at the upper right of the capillary bed.
Nowadays, it may be more relevant and fruitful to study subclinical fat embolism or fat embolism
in patients with multiple injuries where the classical manifestations of FES will be overshadowed
by the effects of the injuries. Such an approach will probably benefit from new sensitive
techniques that can determine the behavior of fat emboli.
37
Fat embolism and patent foramen ovale
REFERENCES
1.
Sevitt S. Fat embolism. London, Butterworths, 1962.
2.
Bierre AR, Koelmeyer TD . Pulmonary fat and bone marrow embolism in aircraft accident
victims. Pathology 1983;15:131-135.
3.
Mudd KL, Hunt A, Matherly RC, Goldsmith LJ, Campbell FR, Nichols GR 2nd, Rink RD.
Analysis of pulmonary fat embolism in blunt force fatalities. J Trauma 2000;48:711-715.
4.
ten Duis HJ. The fat embolism syndrome. Injury 1997;28:77-85.
5.
Bergentz SE, Studies on the genesis of post-traumatic fat embolism. Acta Chir Scand 1962:282.
6.
Hulman G. Fat macroglobule formation from chylomicrons and non-traumatic fat embolism. Clin
Chim Acta 1988;177:173-178.
7.
Huang JC, Gay R, Khella SL. Sickling crisis, fat embolism, and coma after steroids. Lancet
1994;344:951-2.
8.
Panter-Brick M, Wagget J, Dale G. Intralipid and thrombocytopenia. Lancet 1975;i:857-858.
9.
Richter D, Rohricht AM, Nurnberger W, Wahn V, Schroten H. The fat emulsion agglutination
test: a reliable and cost effective alternative to the latex agglutination test for rapid bedside CRP
measurement. Clin Chim Acta 1997; 261:141-148.
10.
11.
12.
13.
ten Duis HJ, Nijsten MWN, Klasen HJ, Binnendijk B. Fat embolism in patients with an isolated
fracture of the femoral shaft. J Trauma 1988;28:383-90.
Lechat Ph, Mas JL, Lascault G, Loron P, Theard M, Klimczac M, Drobinski G, Thomas D,
Grosgogeat Y.Prevalence of patent foramen ovale in patients with stroke. N Eng J Med
1988;318:1148-52.
Webster MWI, Chancellor AM, Smith HJ, Swift DL, Sharpe DN, Bass NM, Glasgow GL. Patent
foramen ovale in young stroke patients. Lancet 1988;ii:11-12.
Moon RE, Camporesi EM, Kisslo JA. Patent foramen ovale in decompression sickness in divers.
Lancet 1989;i:513-14.
14.
Mügge A, Daniel WG, Wenzlaff P, Lichtlan PR. Patent foramen ovale or left atrial thrombi in
unexplained arterial embolism. Lancet 1989;i:282-283.
15.
Sevitt S. Pulmonary and systemic fat embolism. Proc Path Soc Gr Britain Ireland. 1956.
16.
Peltier LF. Fat embolism. Detection of fat droplets in the circulating blood. Surgery 1954;36:198.
17.
Musselman MM, Glas WW, Grekin TD. Fat embolism, a clinical investigation. Arch Surg
Chicago 1952;65:551.
18.
Pell AC, Hughes D, Keating J, Christie J, Busuttil A, Sutherland GR. Brief report: fulminating fat
embolism syndrome caused by paradoxical embolism through a patent foramen ovale. N Engl J
Med 1993;329:926-929.
19.
Fabian TC. Unravelling the fat embolism syndrome. N Engl J Med 1993;329:961-963.
20.
Pell AC, Christie J, Keating JF, Sutherland GR The detection of fat embolism by transesophageal
echocardiography during reamed intramedullary nailing. A study of 24 patients with femoral and
tibial fractures. J Bone Joint Surg Br 1993;75:921-925.
21.
Derks CM, Peters RM. The role of shock and fat embolus in leakage from pulmonary capillaries.
Surg Gynecol Obstet 1973;37:945-948.
22.
Scuderi CS. Fat embolism: a clinical and experimental study. Surg Gynec Obstet 1941;72:732.
23.
Tornabene VW, Fortune JB, Wgner PD, Halasz NA. Gas exchange after pulmonary fat embolism
in dogs. J Thorac Cardiovasc Surg 1979;78:589.
38
Chapter 3
24.
Jones JG, Minty BD, Beeley JM, Royston D, Crow J, Grossman RF Pulmonary epithelial
permeability is immediately increased after embolization with oleic acid but not with neutral fat.
Thorax 1982;37:169-174.
25.
van den Brand JG, van der Hoeven JH, Olsman JG. Dyspnea or confusion after trauma? Consider
fat embolism syndrome. Ned Tijdschr Geneeskd 2000;144:1513-1517.
26.
Gossling HR, Pellegrini VD. Fat embolism syndrome: a review of the pathophysiology and
physiological basis of treatment. Clin Orthop 1982; 165:68-82.
27.
Behrman SW, Fabian TC, Kudsk KA, Taylor JC.Improved outcome with femur fractures: early
vs. delayed fixation. J Trauma 1990;30:792-797.
28.
Edmonds CR, Barbut D, Hager D, Sharrock NE. Intraoperative cerebral arterial embolization
during total hip arthroplasty. Anesthesiology 2000;93:315-318
29.
Zenker FA. Beiträge zur Anatomie und Physiologie der Lunge. Braunsdorf 1861. p61.
30.
Pitto RP, Blunk J, Kossler M. Transesophageal echocardiography and clinical features of fat
embolism during cemented total hip arthroplasty. A randomized study in patients with a femoral
neck fracture. Arch Orthop Trauma Surg 2000;120:53-58.
31.
Arakawa H, Kurihara Y, Nakajima Y. Pulmonary fat embolism syndrome: CT findings in six
patients. J Comput Assist Tomogr 2000;24:24-29.
32.
Takahashi M, Suzuki R, Osakabe Y, Asai JI, Miyo T, Nagashima G, Fujimoto T, Takahashi Y.
Magnetic resonance imaging findings in cerebral fat embolism: correlation with clinical
manifestations. J Trauma 1999;46:324-327.
39
Fat embolism and patent foramen ovale
40
CHAPTER 4
INTERLEUKIN-6, FEVER AND ACUTE PHASE RESPONSES IN PATIENTS WITH
BURNS
M.W.N. Nijsten, E.R. de Groot, H.J. ten Duis, H.J. Klasen, C.E. Hack, L.A. Aarden
Part of this chapter was published in as a letter in
the Lancet 1987;2: 921
Serum levels of interleukin-6 and acute phase responses
INTRODUCTION
Fever has been defined as "a state of elevated core temperature, which is often, but not
necessarily, part of the defensive responses of multicellular organisms (host) to the invasion of
live (microorganisms) or inanimate matter recognized as pathogenic or alien by the host" [1]. This
definition clearly recognizes that fever is often a response to foreign substances, called exogenous
pyrogens. The bacterial cell-wall constituent lipopolysaccharide (endotoxin) is an important
exogenous pyrogen. However, many non-infectious diseases are accompanied by fever as well.
Experiments with sterile inflammation models led to the hypothesis that the body itself produces
pyrogenic substances. The rabbit fever assay has long been a standard for measuring pyrogenic
activity of materials [2]. Upon injection of various pyrogens the rabbit develops fever in a
reproducible manner. Experiments showed that supernatants from in vitro endotoxin-activated
monocytes induced fever upon injection, pointing to the existence of an endogenous pyrogen [3].
It was further hypothesized that the induction of endogenous pyrogen might be a common
pathway in the generation of fever. The fact that prostaglandin synthetase inhibitors (e.g. aspirin)
depress fever of diverse origins also suggests a common pathway of fever induction. The
endogenous pyrogen as produced by monocytes for example was apparently pyrogenic in minute
quantities [3], but its precise characterization had been impossible until more sensitive and
specific technologies, such as recombinant DNA and monoclonal antibodies were available.
Fever is a manifestation of a systemic inflammatory response that involves multiple organ
systems. The coordinated reaction that follows trauma or infection is often called the acute phase
response. In addition to fever, its components are tachycardia, leukocytosis and changes in
circulating protein levels. Acute phase proteins such as C-reactive protein (CRP) and α1antitrypsin (α1AT) are increased, whereas albumin levels are decreased.
On the basis of partially purified protein extracts the existence of the monokine interleukin-1 (IL1), which is produced by activated monocytes was proposed. IL-1 preparations induced the acute
phase response and mediated fever through the release of prostaglandin-E2 in the hypothalamus
[4,5]. However, partly because of the difficulties in measuring IL-1 in biological fluids with the
IL-1 bioassay, only scant evidence for a correlation between circulating IL-1 and the acute phase
response had been put forward [6].
In 1987, the cDNA was cloned of another monocyte product that seemed to be involved in the
acute phase response. Interleukin-6 (IL-6) is a 184 amino acid protein with a molecular weight of
20 to 30 kD (dependent on glycosylation) and was originally described as interferon-ß2, 26K
protein, B-cell stimulating factor-2, and hybridoma growth factor. In human peripheral blood,
monocytes are the main source of IL-6 [7]. IL-6 appeared to be involved in the acute phase
response since Gauldie et al. [8] found that IL-6 is also identical to hepatocyte-stimulating factor,
a major inducer of acute phase protein production in cultured liver cells. We have observed that
recombinant IL-6 (rIL-6) is active in the thymocyte co-stimulation assay, the classical assay for
IL-1. Furthermore, rIL-6 was found to be strongly pyrogenic in rabbits [9].
38
Chapter 4
Table 4.1
Serum IL-6 levels and the development of the acute phase response in patients
with severe burns.
temperature
N
IL-6
CRP
Mean (SE)
α1AT
On admission
(pmol/l)
8.6(5.4)
(mg/l) (IU/ml)
34(17) 81(6)
(oC)
37.8(0.1)
6
One day post-burn 6
2 months post-burn 9
5
Controls
3.9(1.5)
1.3(0.4)
0.4(0.2)
96(21) 116(9)
67(17) 220(9)
<19
80(10)
38.8(0.1)
37.6(0.1)
37.0(0.1)
In this study we investigated the involvement of IL-6 in the acute phase response in humans. In
patients with severe burns and in healthy controls we applied a sensitive IL-6 bio-assay to detect
the presence of increased circulating IL-6 levels. Patients with burns display an immediate severe
inflammatory reaction, including early high fever without apparent infection, suggestive of high
endogenous pyrogen production. Since the time of the injury is well defined, this allows optimal
identification of time-dependent changes of inflammatory parameters.
PATIENTS AND METHODS
In patients admitted to the Burns Unit of the Martini Hospital Groningen serum and plasma
samples were taken, that were frozen directly after centrifugation.
For IL-6 a very sensitive bioassay, that allows detection of as little as 0.1 pmol/l IL-6 in biological
fluids, was used [7,10]. The assay is based on the measurement of 3H-thymidine incorporation in
an IL-6-dependent B-cell line. IL-6 levels were measured in serum with this B9 bioassay as
described previously [7]. In this assay an activity of 1 U/ml corresponds to 1 pg/ml or 5.10-14 mol/l of
IL-6. The samples were preincubated at 56o C for 30 minutes, and tested in triplicate and at multiple
dilutions. Specificity was checked by inhibition with polyclonal goat antibodies raised against
recombinant human IL-6. CRP and α1-antitrypsin (α1-protease inhibitor) were detected in serum by
laser nephelometry (Behring Nephelometer Analyzer, Behringwerke AG, Marburg, Germany).
Samples from healthy controls were run in parallel in all immunological assays. IL-6 values were
log-transformed to approximate a normal distribution. Pearson's regression analysis was used to
calculate coefficients of correlation between two parameters. To determine the significance of
correlations or differences between means, Student's t-test was used. A two-sided P-value of less than
0.05 was considered to represent a significant difference. Means are expressed with standard errors.
RESULTS
13 patients with a mean percentage of burned body surface (TBSA) of 31% were included.
When we monitored circulating IL-6, CRP and α1-antitrypsin we found that within hours of the
injury, IL-6 had risen from 2 to 100 times the normal level, whereas CRP and α1-antitrypsin rose
39
Serum levels of interleukin-6 and acute phase responses
more gradually (Table 4.1). On the other hand, while the acute phase proteins were still raised,
temperature and IL-6 were already decreasing. IL-6 was correlated with temperature (R=0.61,
p<0.0001), and with CRP (R=0.44, p<0.001). Antiserum raised against rIL-6 inhibited the activity
of serum in the IL-6 assay, which proves the assay’s specificity for IL-6.
DISCUSSION
In this study it was shown that IL-6 is strongly elevated in burns patients, and that IL-6 levels
were correlated with manifestations of the acute phase response such as fever and increased CRPlevels. The time-lag between the peak in IL-6 and CRP can be explained by the fact that IL-6 is
directly produced by stimulated monocytes and activated endothelial cells, whereas acute phase
proteins are synthesized in the liver after induction by mediators. The time-lag between the CRP-
40
Temperature
39
38
37
36
Interleukin-6 mol/l
-13
10
1.0E-13
35
-12
10
1.0E-12
-11
10
1.0E-11
-10
10
1.0E-10
Figure 4.1. Relation between body temperature and IL-6 levels in 13 patients with
severe burns. IL-6 was measured in triplicate at three different dilutions; temperatures
are daily means of four measurements. Normal IL-6 range is < 5 10-13 mol/l (10
U/ml).
peak and α1AT-peak is not unexpected since CRP is known to be a faster responding acute phase
protein than α1AT [11]. Our study is obviously limited by the relatively small number of patients
that were included. Although the overall correlation between IL-6 and fever was relatively strong
with an R of 0.61, the factor time was not taken into account, as the limited set of patients did not
allow correlation of IL-6 with fever on fixed times for all patients. But even a correlation of IL-6
with fever at a fixed time after admission as such is no proof of causality. A causal role of IL-6 in
fever in man could only be proved by administration of IL-6. At the time of the study no rhIL-6
for administration to humans was available.
40
Chapter 4
Induction of acute phase response by IL-6 and other cytokines
Several purified recombinant cytokines have been shown to induce both fever and other
components of the acute phase response. Although IL-1 was the first substance to be identified as
an endogenous pyrogen, IL-1, IL-6 and TNFα now all appear to be major endogenous pyrogens.
Studies have shown that both in animals and humans injected with IL-1α [12], IL-1β [13], IL-6
[14,15] and TNFα [16,17], fever is induced. In many models of systemic inflammation IL-6 is
induced by IL-1 and TNFα, with a marked synergistic effect of IL-1 and TNFα on IL-6
production [2]. In 1994 [15] it was shown that administration of IL-6 is able to induce all aspects
of the acute phase response in man. Later it was also observed that IL-6 also induces
procalcitonin, a protein with even faster and more pronounced kinetics than CRP [18]. Although
the pleiotropic clinical manifestations after IL-6 administration are clearly inflammatory, in
comparison with the more potent inflammatory cytokines TNFα or IL-1 some authors have
characterized IL-6 as an anti-inflammatory cytokine[19].
IL-6 as a quantitative marker of inflammation
In the field of plasma cytokine measurements, after more than 10 years of extensive
investigations, IL-6 has emerged as one of the most useful cytokines for quantifying acute
inflammatory responses. Several investigations have shown that higher, and also persistently
higher IL-6 levels are associated with poor outcome in sepsis [20,21,22]. In fact, levels of IL6>1000 pg/ml, as assayed by a semi-quantitative point-of-care test, have been used as an inclusion
criterion in anti-TNF-trials [23]. In early triage for trial purposes it will be of interest to see how
early procalcitonin (PCT) levels will compare with IL-6 levels as a criterion for patient selection
with sepsis trials. PCT is also induced by IL-6 and TNFα and rises much faster than CRP [18]. In
addition some investigators have claimed [24,25,26] that PCT-levels can better discriminate septic
from non-septic patients than CRP.
Cytokine intervention studies
Since TNF and IL-1 appear to be more responsible for the adverse effects of the inflammatory
response, these cytokines have been intensively studied as target for therapeutic interventions.
Many experimental models have shown benefit, with clinically the most impressive beneficial
effect of anti-TNF treatment in the Jarisch-Herxheimer syndrome [27]. It should be noted that in
this successful study anti-TNF was given immediately before the expected Jarisch-Herxheimer
reaction. But in clinical sepsis trials significant effects on mortality have not materialized to date
[28]. The prospective studies that did not show a beneficial effect on mortality in sepsis patients
included human recombinant IL-1 receptor antagonist [29,30], antiTNF antibodies [23,31,32] and
TNF receptor fusion protein [33,34]. The disappointing clinical results with inhibition of a single
cytokines has forced investigators to reassess how elevated cytokine levels in sepsis should be
interpreted. Dozens of cytokines and cell types have been identified to have a role in
inflammation [35], with an overwhelming number of potential interactions. It is now accepted that
cytokines should be viewed as part of a network were information is transmitted by parallel and
sequential induction of cytokines and other mediators. Consequentially it is not surprising that the
inhibition of only one mediator will have limited effects [28].
41
Serum levels of interleukin-6 and acute phase responses
Conclusion
IL-6 is an important endogenous pyrogen and inducer of other aspects of the acute phase
response. IL-6 induces these effects directly and together with IL-1 or TNFα. The measurement
of circulating IL-6 continues to provide one of the most valuable tools for studying inflammatory
responses in patients.
REFERENCES
1.
IUPS Thermal commission. Glossary of terms for thermal physiology: second edition. Pflugers
Arch 1987;410:567-587.
2.
Dinarello CA. Cytokines as endogenous pyrogens. J Infect Dis 1999;179:S2:S294-304.
3.
Bodel P, Atkins E.Release of endogenous pyrogen by human monocytes. N Engl J Med
1967;276:1002-1008.
4.
Editorial. Interleukin-1 in defense of the host. Lancet 1985; ii: 536-537.
5.
Dinarello CA. Interleukin-1. Rev Infect Dis 1984;6:51-95.
6.
Bendtzen K, Baek L, Berild D, Hasselbach H, Dinarello CA, Wolff SM. Demonstration of
circulating leukocytic pyrogen/interleukin-1 during fever. N Eng J Med 1984; 310:596.
7.
Aarden LA, de Groot ER, Schaap OL, Lansdorp PM. Production of hybridoma growth factor by
human monocytes. Eur J Immunol 1987;17:1411-1416.
8.
Gauldie J, Richards C, Harnish D, Lansdorp P, Baumann H. Interferon ß2/B-cell stimulatory factor
type 2 shares identity with monocyte-derived hepatocyte-stimulating factor and regulates the major
acute phase response in liver cells. Proc Natl Acad Sci USA 1987;84:7251-7256.
9.
Helle M, Brakenhoff JP, De Groot ER, Aarden LA, Interleukin 6 is involved in interleukin 1induced activities. Eur J Immunol 1988;18:957-959
10.
van Oers MHJ, van der Heyden AAPAM, Aarden LA, A novel interleukin in serum and urine of
renal transplant recipients. Clin Exp Immunol 1988;71:314-319.
11.
Gabay C, Kushner I. Acute-phase proteins and other systemic responses to inflammation. N Engl
J Med 1999;340:448-454.
12.
Smith JW, Urba WJ, Curti BD, et al. The toxic and hematologic effects of interleukin-1 alpha
administered in a phase I trial to patients with advanced malignancies. J Clin Oncol
1992;10:1141-1152.
13.
Crown J, Jacubowski A, Kemeny N, Elwood LJ, Steis RG, Janik JE, Sharfman WH, Miller LL,
Fenton RG, Conlon KC, et al. A phase I trial of recombinant human interleukin 1β alone and in
combination with myelosuppressive doses of 5-fluorouracil in patients with gastrointestinal
cancer. Blood 1999;78:1420-1427.
14.
Dinarello CA, Cannon JG, Mancilla J, Bishai I, Lees J, Coceani F. Interleukin-6 as an
endogenous pyrogen: induction of prostaglandin E2 in brain but not in peripheral blood
mononuclear cells. Brain Res 1991;562:199-206.
15.
van Gameren MM, Willemse PH, Mulder NH, Limburg PC, Groen HJ, Vellenga E, de Vries EG.
Effects of recombinant human interleukin-6 in cancer patients: a phase I-II study. Blood
1994;84:1434-1441.
42
Chapter 4
16.
Dinarello CA, Cannon JG, Wolff SM, Bernheim HA, Beutler B, Cerami A, Figari IS, Palladino
MA Jr, O'Connor JV. Tumor necrosis factor (cachectin) is an endogenous pyrogen and induces
production of interleukin 1. J Exp Med 1986;163:1433-1450.
17.
van der Poll T, Buller HR, ten Cate H, Wortel CH, Bauer KA, van Deventer SJ, Hack CE,
Sauerwein HP, Rosenberg RD, ten Cate JW. Activation of coagulation after administration of
tumor necrosis factor to normal subjects. N Engl J Med 1990;322:1622-1627.
18.
Nijsten MWN, Olinga P, The TH, de Vries EG, Schraffordt Koops H, Groothuis GM, Limburg
PC, ten Duis HJ, Moshage H, Hoekstra HJ, et al. Procalcitonin behaves as a fast responding acute
phase protein in vivo and in vitro. Crit Care Med 2000 ;28:458-461.
19.
van der Poll T, van Deventer SJH. Interleukin-6 in bacterial infection and sepsis: innocent
bystander or essential mediator? In: Vincent JL (Ed) Yearbook of intensive care and emergency
medicine. Springer-Verlag Berlin Heidelberg 1999:43-53.
20.
Waage A, Brandtzaeg P, Halstensen A, Kierulf P, Espevik T. The complex pattern of cytokines
in serum from patients with meningococcal septic shock. Association between interleukin 6,
interleukin 1, and fatal outcome. J Exp Med 1989;169:333-338.
21.
Lowry SF, Calvano SE, van der Poll T. Measurement of inflammatory mediators in clinical
sepsis. In: Sibbald WJ, Vincent JL (eds) Clinical trials for the treatment of sepsis. SpringerVerlag Heidelberg. 1995:86-105.
22.
Hack CE, De Groot ER, Felt-Bersma RJ, Nuijens JH, Strack Van Schijndel RJ, EerenbergBelmer AJ, Thijs LG, Aarden LA. Increased plasma levels of interleukin-6 in sepsis. Blood
1989;74:1704-1710.
23.
Reinhart K, Wiegand-Lohnert C, Grimminger F, Kaul M, Withington S, Treacher D, Eckart J,
Willatts S, Bouza C, Krausch D, Stockenhuber F, Eiselstein J, Daum L, Kempeni J. Assessment
of the safety and efficacy of the monoclonal anti-tumor necrosis factor antibody-fragment, MAK
195F, in patients with sepsis and septic shock. A multicenter, randomized, placebo-controlled,
dose-ranging study. Crit Care Med 1996;24:733-742.
24.
Assicot M, Gendrel D, Carsin H, Raymond J, Guilbaud J, Bohuon C. High serum procalcitonin
concentrations in patients with sepsis and infection. Lancet 1993;341:515-518.
25.
Oberhoffer M, Karzai W, Meier-Hellmann A, Bogel D, Fassbinder J, Reinhart K. Sensitivity and
specificity of various markers of inflammation for the prediction of tumor necrosis factor-alpha
and interleukin-6 in patients with sepsis. Crit Care Med 1999;27:1814-1818.
26.
Reinhart K, Karzai W, Meisner M. Procalcitonin as a marker of the systemic inflammatory
response to infection. Intensive Care Med 2000;26:1193-1200.
27.
Fekade D, Knox K, Hussein K, Melka A, Lalloo DG, Coxon RE, Warrell DA. Prevention of
Jarisch-Herxheimer reactions by treatment with antibodies against tumor necrosis factor alpha. N
Engl J Med 1996;335:311-315.
28.
Abraham E. Why immunomodulatory therapies have not worked in sepsis. Intensive Care Med
1999;25:556-566.
29.
Fisher CJ Jr, Dhainaut JF, Opal SM, Pribble JP, Balk RA, Slotman GJ, Iberti TJ, Rackow EC,
Shapiro MJ, Greenman RL, et al. Recombinant human interleukin 1 receptor antagonist in the
treatment of patients with sepsis syndrome. Results from a randomized, double-blind, placebocontrolled trial. Phase III rhIL-1ra Sepsis Syndrome Study Group. JAMA 1994;271:1836-1843.
43
Serum levels of interleukin-6 and acute phase responses
30.
Opal SM, Fisher CJ Jr, Dhainaut JF, Vincent JL, Brase R, Lowry SF, Sadoff JC, Slotman GJ,
Levy H, Balk RA, Shelly MP, Pribble JP, LaBrecque JF, Lookabaugh J, Donovan H, Dubin H,
Baughman R, Norman J, DeMaria E, Matzel K, Abraham E, Seneff M.Confirmatory interleukin1 receptor antagonist trial in severe sepsis: a phase III, randomized, double-blind, placebocontrolled, multicenter trial. The Interleukin-1 Receptor Antagonist Sepsis Investigator Group.
Crit Care Med 1997;25:1115-1124.
31.
Abraham E, Anzueto A, Gutierrez G, Tessler S, San Pedro G, Wunderink R, Dal Nogare A,
Nasraway S, Berman S, Cooney R, Levy H, Baughman R, Rumbak M, Light RB, Poole L, Allred
R, Constant J, Pennington J, Porter S. Double-blind randomised controlled trial of monoclonal
antibody to human tumor necrosis factor in treatment of septic shock. NORASEPT II Study
Group. Lancet 1998;351:929-933.
32.
Cohen J, Carlet J. INTERSEPT: an international, multicenter, placebo-controlled trial of
monoclonal antibody to human tumor necrosis factor-alpha in patients with sepsis. International
Sepsis Trial Study Group. Crit Care Med 1996;24:1431-1140.
33.
Fisher CJ Jr, Agosti JM, Opal SM, Lowry SF, Balk RA, Sadoff JC, Abraham E, Schein RM,
Benjamin E. Treatment of septic shock with the tumor necrosis factor receptor:Fc fusion protein.
The Soluble TNF Receptor Sepsis Study Group. N Engl J Med 1996;334:1697-1702.
34.
Abraham E, Glauser MP, Butler T, Garbino J, Gelmont D, Laterre PF, Kudsk K, Bruining HA,
Otto C, Tobin E, Zwingelstein C, Lesslauer W, Leighton A. p55 Tumor necrosis factor receptor
fusion protein in the treatment of patients with severe sepsis and septic shock. A randomized
controlled multicenter trial. Ro 45-2081 Study Group. JAMA 1997;277:1531-1538.
35.
44
Hack CE, Aarden LA, Thijs LG. Role of cytokines in sepsis. Adv Immunol 1997;66:101-195.
CHAPTER 5
INTERLEUKIN 6 AND ITS RELATION TO THE HUMORAL IMMUNE
RESPONSE AND CLINICAL PARAMETERS IN BURNED PATIENTS
M.W.N. Nijsten, C.E. Hack, M. Helle, H.J. ten Duis, H.J Klasen L.A. Aarden
Surgery 1991; 109: 761-767
Interleukin-6 and humoral and clinical responses in burned patients
ABSTRACT
The cytokine interleukin-6 (IL-6), which has been shown to be increased in burn patients, is
produced by activated monocytes and endothelial cells and has many in vitro activities, including
stimulation of acute phase protein synthesis in hepatocytes, immunoglobulin-synthesis in Blymphocytes and stimulation of growth of megakaryocytes. In 13 patients with a mean of 31% full
thickness burns, we studied the relation of serum IL-6 to clinical parameters and parameters of the
acute phase response and immunoglobulin production. IL-6 was already elevated within hours after
sustaining the injury, and it remained elevated for several weeks. All components of the acute phase
response were observed: fever, tachycardia, leukocytosis with an associated left shift, elevation of Creactive protein and α1-antitrypsin, and a decrease in albumin levels. In the second week after burn
injury immunoglobulin M-levels peaked, followed by a prolonged elevation of immunoglobulin Glevels. Platelet counts initially decreased and rebounded to supranormal levels after two weeks. IL-6
levels were positively correlated with acute phase responses.
We believe that the production of IL-6 induces the synthesis of acute phase proteins. High IL-6 levels
may also be an etiologic factor in the marked immunoglobulin response observed. Likewise, the
relation between the megakaryocyte promoting activity of IL-6 and the rebound thrombocytosis,
requires further investigation.
INTRODUCTION
Interleukin-6 (IL-6) is a hormone-like protein that plays a central role in inflammation. It is produced
by various cells, such as monocytes, endothelial cells, and fibroblasts. The multifunctional properties
of IL-6 are illustrated by the fact that after DNA analysis, several proteins known for different activities turned out to be identical to IL-6: hepatocyte stimulatory factor (HSF), B-cell stimulatory factor2 (BSF-2), interferon-beta-2 (IFN-ß2), hybridoma growth factor (HGF), and human myeloid
differentiation-inducing protein (MGI-2) [1].
IL-6 causes fever upon injection [2], and it can induce production of acute phase proteins like Creactive protein (CRP), fibrinogen and α1-antitrypsin (α1-proteinase inhibitor) in hepatocytes [3].
Severely burned patients display a strong acute phase response in the first hours to days after burn
injury. They often have a high fever within 24 hours and produce acute phase proteins such as CRP
and α1-antitrypsin. As we have shown before, these patients have markedly increased serum IL-6
levels, which correlate with body temperature [4].
As a B-cell stimulatory factor IL-6 induces immunoglobulin production [5]. Burned patients display
elevated immunoglobulin levels in the weeks following the burn [6]. Immunoglobulin M (IgM) rises
to supranormal levels after 1 week, whereas IgG shows a slower increase to levels above the normal
range after several weeks. Lymphocytes isolated from burn patients have an elevated spontaneous,
polyclonal immunoglobulin production [7].
50
Chapter 5
The genesis of this immunoglobulin response in burn patients is unclear and has been related to stress
hormone levels, the presence of infection, T-cell alloreactivity and antigens from traumatized tissue
[8,9,10].
Recently it has become apparent that IL-6 is also a direct-acting growth factor for megakaryocytes,
and as such plays a key role in thrombopoiesis [11]. Platelet counts have long been known to be
dramatically increased in burn patients, but this phenomenon has received little attention. It is
supposed to be a 'rebound' after initial sequestration of platelets in the burn wound [12], comparable
to other conditions characterized by a temporary depletion of platelets.
The goal of this study was to assess the clinical and laboratory characteristics of inflammation in
burn patients, focusing on IL-6 production. We specifically looked at the time course of IL-6, acute
phase responses, immunoglobulin synthesis and platelet counts.
PATIENTS AND METHODS
Patients
Eight females and five males (mean age of 33 years), were selected for this study (Table 5.1). The
mean percentage of total body surface area covered with full thickness burns was 31%. None of the
patients had physical disorders before the accident. This patient group is identical with the one briefly
described by us before [4]. On admission patients were vaccinated with tetanus toxoid. Surgical
treatment was performed in phases. Each week approximately 10% of the full thickness burn was
excised and covered by an autologous skin graft. In addition, an essential component of our treatment
was selective decontamination of the intestinal tract and early oral feeding to prevent colonization by
potential pathogens [13]. This decontamination regimen consisted of daily doses of oral colistine,
polymyxin B and co-trimoxazole. Blood cultures were taken when bacteremia was suspected.
Systemic antibiotics were only administered in case of proven bacteraemia. Low-dose diclofenac
(VoltarenR, 5 or 10mg), was administered in case of fever greater than 40o C.
Routine parameters
Records of temperature, heart rate and blood pressure were taken, as well as of fluids and medication
administered. The core temperature and the heart rate values used were the daily means of at least
four measurements. Routine daily examinations included total protein, albumin, hemoglobin, platelets, white cell numbers and white cell differentiation count, blood urea nitrogen (BUN), creatinine
and creatinine clearance.
Immunological assays
Serum and EDTA-plasma samples were taken on admission and during the five subsequent days;
thereafter samples were taken weekly until discharge from the burn unit. Samples were stored in
three aliquots at -70o C and were analyzed directly after thawing. IL-6 levels were measured in serum
with the B9 bioassay as described previously [14]. In this assay an activity of 1 U/ml corresponds to 1
pg/ml or 5.10-14 mol/l of IL-6. The samples were preincubated at 56o C for 30 minutes, and tested in
51
Interleukin-6 and humoral and clinical responses in burned patients
triplicate and at multiple dilutions [15]. Specificity was checked by inhibition with polyclonal goat
antibodies raised against recombinant human IL-6. To assess complement activation, C3a, the activation product of the third component of complement, was measured by radioimmunoassay in plasma
[16] and C3dg was measured with a similar radioimmunoassay. C-reactive protein, α1-antitrypsin,
IgG, IgM, complement factor C4 and α2-macroglobulin were measured in serum by laser
nephelometry (Behring Nephelometer Analyzer, Behringwerke AG, Marburg, Germany) using
appropriate antisera. Samples from healthy controls were run in parallel in all immunological assays.
Analysis of data
IL-6 and C3a values were log-transformed to approximate a normal distribution. Pearson's regression
analysis was used when appropriate to calculate coefficients of correlation between two parameters.
To determine the significance of correlations or differences between means, Student's t-test was used.
A two-sided P-value of less than 0.05 was considered to represent a significant difference. Unless
indicated otherwise, means are expressed with associated standard errors.
RESULTS
Patients
Table 5.1 shows the general characteristics of the 13 patients studied. All patients survived, and only
one episode of proven bacteremia was encountered in this group (in the 6th week post-burn). One
patient suffered from inhalation injury, requiring mechanical ventilation. No patients were in
hypotensive shock during any phase of the study. We decided to analyze the data from all patients
together. Figs. 5.1, 5.2 and 5.3 display means of measurements for 15 parameters.
Table 5.1
General clinical data
Sex
Age
Full thickness TBSA
Partial and full thickness TBSA
Operations
Stay at burn unit
Deaths
Numbers, means
8 females, 5 males
33 (range 16-33) year
31 (range 10-56) percent
38 (range 17-56) percent
4.0 (range 2-6) operations/patient
77 (range 26-187) days
0
TBSA:Total body surface area
Interleukin 6
IL-6 levels were already elevated at the time of admission and continued to be elevated afterward
(Fig. 5.1). Although the levels varied considerably, in general they were 10- to 100-fold elevated
above normal levels. (controls: <5.10-13 Mol/l ). There was no relation between the extent of the
burns and IL-6 levels, but patients with more severe burns had elevated IL-6 for longer periods.
52
Chapter 5
Acute phase proteins
CRP-levels were elevated from the first day post-burn onward, rose a 100-fold and were down to
control values after two months. The levels of α1-antitrypsin displayed a less pronounced and slower
rise, but remained at twice control levels after two months.
Immunoglobulins
The levels of IgM and IgG initially dropped and later rose to supranormal levels. IgM levels showed
a very significant rise in the second week after burn injury. IgG rose later and remained elevated at
twice normal levels for longer periods.
Complement activation
Increased levels of C3a and C3dg indicated activation of complement factor C3. Levels varied
widely and could not be related to the extent of the burns. Mean levels of C3a, as well as C3dg, were
highest at the end of the second week (Fig. 5.2). C4 levels did not significantly change during the
first week and increased steadily during subsequent weeks.
Routine parameters
Temperature was elevated within hours after admission and remained above 38o C for more than 1
month. One patient reached a body temperature of 41o C within 12 hours after the injury. The heart
rate followed the temperature response and was well over 120/min in the first week, and averaged
over 100/min during the entire stay in the burn unit. Leukocyte counts were increased sharply to
28·109 /l on admission, and remained elevated around 12·109/l for several weeks (Fig. 5.3). In the
leukocyte differentiation a sharp elevation of the percentage of rods (young neutrophils) was noted
(rods in controls <3%). Because of the presence of abnormal leukocytes like myelocytes and
metamyelocytes in the differentiation, the stacked bars in Fig. 5.1 do not always add up to 100%.
High hemoglobin during the first 48 hours reflected hemoconcentration as a result of extensive loss
of fluid. Platelet counts were low during the first week and then rapidly rebounded to supranormal
levels, not returning to normal levels until the second month. Total protein remained below normal
levels for many weeks.
Correlations
Correlations between IL-6 and parameters of the acute phase response are shown in Table 5.2. All
measurements in all patients were used for this analysis. Values of concentrations measured were
only correlated when determined in the same sample. IL-6 appeared to be most strongly correlated
with body temperature. IL-6 was negatively correlated with total protein and platelet counts. We
found no correlation between complement levels and IL-6.
53
Interleukin-6 and humoral and clinical responses in burned patients
39.5
2.00
140
1.90
39.0
130
1.80
38.5
120
1.50
1.40
38.0
110
/min
1.60
degrees Celcius
log units / ml
1.70
37.5
100
1.30
37.0
Interleukin-6
Temperature and Heart rate
1.20
90
36.5
1.10
1.00
36.0
0
1
2
3
4
9
14
21
30
45
80
65
0
Rods
200
1
2
3
4
9
14
Mature neutrophils
21
30
45
65
Lymphocytes
Monocytes
100
80
percentage
mg / l
150
100
50
60
40
20
C-reactive protein
0
0
0
1
2
3
4
9
14
21
30
45
65
0
1
2
3
4
9
14
21
30
Figure 5.1. Time course of means of IL-6, temperature (filled circles), heart rate (squares), CRP and leukocyte
differentiation for the entire patient group. Note that the time scale is chosen to accentuate the important phases; it is
neither linear nor logarithmic. IL-6 levels are shown on a logarithmic scale. The percentages in the leukocyte
differentiation do not always add up to 100%, due to the presence of non-standard types of leukocytes.
54
45
65
Chapter 5
300
300
IgG en IgM
250
200
IU / ml
IU / ml
200
150
alpha-1-Antitrypsin
100
100
50
0
0
0
1
2
3
4
9
14
21
30
45
65
0
200
1
2
3
4
9
14
21
30
45
65
100
100
80
100
60
10
40
50
20
C3a and C3dg
0
1
0
1
2
3
4
9
14
21
30
45
65
0
0
1
2
3
4
9
14
21
30
45
Figure 5.2. Time course of means of α1antitrypsin, IgM (filled circles), IgG (squares), complement factors C4, C3a (filled
circles) and C3dg (open circles) for the entire patient group. Note that the time scale is neither linear nor logarithmic. C3a
values are shown on a logarithmic scale.
55
65
C3dg nmol/l
Complement factor C4
C3a nmol/ml
IE / ml
150
Interleukin-6 and humoral and clinical responses in burned patients
40
80
Leukocytes
70
30
60
Total Protein
g/l
10E9 / l
50
20
40
30
20
10
10
0
0
0
1
2
3
4
9
14
21
30
45
65
0
1
2
3
4
9
14
21
30
45
65
600
200
500
160
Hemoglobin
400
g/l
10E9 / l
120
300
80
Platelet Count
200
40
100
0
0
0
1
2
3
4
9
14
21
30
45
65
0
1
2
3
4
9
14
21
30
Figure 5.3. Time course of means of leukocyte count, total protein, hemoglobin and platelet count for the entire patient
group. Note that the time scale is neither linear nor logarithmic.
56
45
65
Chapter 5
Table 5.2
Correlation (R) of interleukin-6 with parameters of the acute phase
response. (n=88 to 91)
Parameter
Parameter
R
P
IL-6
Temperature
+0.61
< 0.0001
Heart rate
+0.55
< 0.0001
Rods
+0.48
< 0.0001
CRP
+0.44
< 0.0001
Leukocytes
+0.36
< 0.0005
IgM
0.00
N.S.
α1-antitrypsin
-0.06
N.S.
Trombocytes
-0.44
< 0.0001
Total protein
-0.48
< 0.0001
DISCUSSION
This study examined the relation of IL-6 to acute phase responses and humoral and hematological
parameters in severely but not critically burned patients. Such patients tend to display maximal
pathophysiological responses, whereas in critically burned patients acute organ failure often strongly
affects many responses. The in vivo responses of several of the parameters shown in Figs. 1,2 and 3
can be interpreted in terms of in vitro activities of IL-6 (Fig. 5.4). The results support the hypothesis
that IL-6 has a causal role in the acute phase reaction. Moreover, the results indicate a possible effect
of IL-6 on changes of immunoglobulin levels and thrombocyte synthesis.
The sustained elevation of IL-6 indicates continuous production by monocytes, endothelial cells or
fibroblasts. Because IL-6 probably has a half-life of less than one hour [17], it must be produced for
prolonged periods. In theory, levels of a low molecular weight protein such as IL-6 (MW 20 to 30
kD) can also increase through decreased glomerular filtration, but the patients displayed only
marginal renal disturbances as reflected by creatinine, creatinine clearance and BUN. These renal
parameters were not correlated with any of the measured proteins, also making a renal effect on
protein levels unlikely. Monocytes are potent producers of IL-6 [14], and they are known to be
activated in burn patients [18]. Endotoxin or complement can induce the production of several
cytokines by monocytes. The absence of correlation between the complement factors we measured
(C3a, C3d and C4) and IL-6, does not rule out a causal relation, since the levels of such short-lived
agents are always difficult to correlate. Endotoxemia can occur in burns patients without a focus of
infection being present. Guo et al. recently showed that in burn patients levels of circulating
endotoxin and IL-6 can be reduced by polymyxin B therapy [19].
IL-6 is only one of many cytokines involved in inflammation, and in the context of acute phase
responses tumor necrosis factor (TNF) and interleukin-1 (IL-1) are important mediators [20]. We
were unable to detect IL-1 with the standard assay, and we did not attempt to measure TNF. Yet it
57
Interleukin-6 and humoral and clinical responses in burned patients
seems logical that both TNF and IL-1 were also produced. So it would be interesting to measure IL-1
and TNF in such patients with sufficiently sensitive assays.
The course of temperature, heart rate, CRP, α1-antitrypsin, leukocytes and the 'left shift' in the
leukocyte differentiation indicate a huge acute phase response. The rapidly developing fever occurs
before an established infection can be present. The correlations between IL-6 and temperature, heart
rate, CRP, leukocytes and percentage of rods are in accordance with a central role of IL-6 in the acute
phase response [21]. Table 5.2 shows that the heart rate is correlated mainly with body temperature
(R=0.67), which suggests that the tachycardia was induced by fever, and indirectly by IL-6. This
tachycardia was more pronounced than that in patients with fever following an uncomplicated viral
infection [22]. The hypermetabolic state and the lowered hemoglobin levels of the burn patients
Figure 5.4. Diagram of the processes possibly related to interleukin-6 (IL-6) production.
probably explain this 'additional' tachycardia. The rise in CRP within one day and the later rise of α1antitrypsin are in accordance with the well-known property of CRP as a fast and strongly responding
acute phase protein. The sharply dropped albumin levels (closely following total protein levels, data
not shown) are in accordance with the down-regulation of albumin synthesis as part of the acute
phase response but are primarily caused by capillary leakage and the massive loss of protein through
the burned skin.
The typical pattern of IgM and IgG (Fig. 5.2) has been observed by other authors [6,9,10]. One
could argue that the increase of the high molecular weight protein IgM (MW 900 kD) was caused by
58
Chapter 5
the infusion of large amounts of protein coupled with the selective loss of other proteins through
tissue or wound leakage. This is unlikely because of the distinct pattern in time of the IgM-peak and
its presence in virtually all patients. Also this pattern was not observed for α2-macroglobulin, another
high molecular weight protein (data not shown). Moreover the protein spectrum in blister fluid is
only marginally determined by molecular weight [23].
C3a en C3dg levels, both indicative of activation of complement factor C3, were maximal in the
second week post-burn and the C4 pattern indicates consumption of C4. The rapid decrease of IgM
in the second week coincides with increases in C3a, which resemble the changes that occur in serum
sickness. Yet, we saw no clinical signs of immune complex deposition and we could not detect
immune-complexes with the C1q binding assay (data not shown).
The patterns of IgG and IgM levels bear a strong resemblance to the textbook primary immune
response when the immune system is challenged by a new antigen. An autoimmune origin of this
response is unlikely, since screening of 15 serum samples from the second week postburn with
indirect immunofluorescence for the presence of autoantibodies was negative. The polyclonal IgG
and IgM response in patients after uncomplicated surgery has been shown to be directed against
'recall' antigens [24]. Moreover, IL-6 has been shown to augment the production of immunoglobulins
by B- cells in vitro and in vivo [5,25].
The sharp rebound in platelet numbers, preceding the slower recovery of hemoglobin levels reflects a
strong stimulation of bone marrow. This rise of platelets is a familiar phenomenon following trauma
and burns [12]. When rats are selectively depleted of platelets by exchange transfusion, they also
display a rebound thrombocytosis [26]. In fact, some authors consider thrombocytosis an integral part
of the acute phase response [27,28]. Sustained thrombocytosis is also often seen in chronic
inflammatory disease states like rheumatoid arthritis [29]. In this disease IL-6 levels in serum and in
particular in synovial fluids, are increased markedly [30]. Recently IL-6 has been demonstrated to be
a crucial factor for thrombopoiesis in vitro and in vivo [11,31]. Although platelet production depends
on the interaction of several cytokines [32], IL-6 may be instrumental in the thrombocytosis in burn
patients. The negative correlation between IL-6 and platelet counts can be explained by the time-lag
between stimulation of precursor cells in the bone marrow and the appearance of platelets in the
peripheral blood.
In summary, this investigation focused on the role of one important cytokine (IL-6) in burn patients.
The high levels of IL-6 we found may be related to three distinct responses: the acute phase response,
immunoglobulin production and thrombocytosis. Although each of these responses is already known
to be influenced by several other factors, IL-6 may play an essential role in each of them. Current
investigations on the effect of IL-6 administered in vivo will help to answer this question.
ACKNOWLEDGEMENT
The authors wish to thank Douwe Buiter for drawing Figure 5.4 and Mereke Schaub for technical
assistance.
59
Interleukin-6 and humoral and clinical responses in burned patients
REFERENCES
1.
Kishimoto T. The biology of interleukin-6. Blood 1989;74:1-10.
2.
Helle M, Brakenhoff JPP, de Groot ER, Aarden LA. Interleukin 6 is involved in interleukin 1induced activities. Eur J Immunol 1988;18:957-959.
3.
Gauldie J, Richards C, Harnish D, Lansdorp P, Bauman H. Interferon ß2/ B-cell stimulatory factor
type 2 shares identity with monocyte-derived hepatocyte-stimulating factor and regulates the major
acute phase protein response in liver cells. Proc Nat Acad Sci USA 1987;84:7251-7255.
4.
Nijsten MWN, de Groot ER, ten Duis HJ, Klasen HJ, Hack CE, Aarden LA. Serum levels of
interleukin-6 and acute phase responses. Lancet 1987;ii:921.
5.
Hirano T, Yasukawa K, Harada H, Taga T, Watanabe Y, Matsuda T, Kashiwamura S, Nakajima K,
Koyama K, Iwamatsu A, Tsunasawa S, Sakiyama F, Matsui H, Takahara Y, Taniguchi T, Kishimoto
T. Complementary DNA for a novel human interleukin (BSF-2) that induces B lymphocytes to
produce immunoglobulin. Nature 1986;324:73-76.
6.
Arturson G, Johansson SGO, Högman CF, Killander J. Changes in immunoglobulin levels in
severely burned patients. Lancet 1969;i:546-548.
7.
Teodorczyk-Injeyan JA, Sparkes BG, Falk RE, Peters WJ. Polyclonal immunoglobulin production in
burned patients: kinetics and correlations with T-cell activity. J Trauma 1986;26:834-839.
8.
Shorr RM, Ershler WB, Gamelli RL. Immunoglobulin production in burned patients. J Trauma
1984;24:319-322.
9.
Hershman MJ, Cheadle WG, George CD, Cost KM, Appel SH, Davidson PF, Polk HC Jr. The
response of immunoglobulins to infection after thermal and non-thermal injury. Am Surg
1988;54:408-411.
10.
Kagan RJ, Bratescu A, Jonasson O, Matsuda T, Teodorescu M. The relationship between the
percentage of circulating B-cells, corticosteroid levels and other immunologic parameters in
thermally injured patients. J Trauma 1989;29:208-213.
11.
Lotem J, Shabo Y, Sachs L. Regulation of megakaryocyte development by interleukin-6. Blood
1989;74:1545-1551.
12.
Eurenius K, Mortensen RF, Meserol PM, Curreri PW. Platelet and megakaryocyte kinetics following
thermal injury. J Lab Clin Med 1972;79:247-257.
13.
Manson WL, Westerveld AW, Klasen HJ, Sauër EW. Selective decontamination of the digestive
tract for infection propylaxis in severely burned patients. Scand J Plast Reconst Surg 1987;21:269272.
14.
Aarden LA, de Groot ER, Schaap OL, Lansdorp PM. Production of hybridoma growth factor by
human monocytes. Eur J Immunol 1987;17:1411-1416.
15.
van Oers MHJ, van der Heyden AAPAM, Aarden LA. Interleukin 6 (IL-6) in serum and urine of
renal transplant recipients. Clin Exp Immunol 1988;71:314-319.
16.
Hack CE, Paardekooper J, Eerenberg AJM, Navis GO, Nijsten MWN, Thijs LG, Nuijens JH. A
modified competitive inhibition radioimmunoassay for the detection of C3a. J Immunol Methods
1988;108:77-84.
60
Chapter 5
17.
van Deventer SJH, Büller HR, ten Cate JW, Aarden LA, Hack CE, Sturk A. Endotoxin-induced
biological effects: the role of cytokines. In: Endotoxins in the pathogenesis of gram-negative
septicemia, Thesis. Amsterdam, 1988.
18.
Moore FD, Davis CF. Monocyte activation after burns and endotoxemia. J Surg Res 1989;46:350354.
19.
Guo Y, Dickerson C, Chrest FJ, Adler WH, Munster AM, Winchurch RA. Increased levels of
circulating interleukin 6 in burn patients. Clin Immunol Immunopathol 1990;54:361-371.
20.
Dinarello CA, Interleukin 1 and its biologically related cytokines. Adv Immunol 1989;44:153-205.
21.
Heinrich PC, Castell JV, Andus T. Interleukin-6 and the acute phase response. Biochem J
1990;265:621-636.
22.
Karjalainen J, Viitasalo M. Fever and cardiac rhythm. Arch Intern Med 1986;146:1169-1171.
23.
Deitch EA, Emmet M. Early protein alteration in blister fluid and serum associated with burn injury.
J Trauma 1986;26:34-39.
24.
Di Padova F, Durig M, Di Padova C, Pozzoli M, Tritapepe R. Spontaneous and polyclonal Ig
secretion by circulating B cells after surgery. Surgery 1988;103:547-552.
25.
Nawata Y, Eugui EM, Lee SW, Allison AC. IL-6 is the principal factor produced by synovia of
patients with rheumatoid arthritis that induces B-lymphocytes to secrete immunoglobulins. Ann N Y
Acad Sci 1989;557:230-238.
26.
Mimura H, Segal GM, Lee MY, Adamson JW. Megakaryocytosis in the rat: response to
trombocytopenia induced by exchange tranfusion. Exp Hematol 1985;13:1048-1054.
27.
Parry MF, Jacobs B, Scully, Neu HC. Thrombocytosis: an acute-phase reactant, not an adverse
reaction to the new beta-lactam antibiotics. Diagn Microbiol Infect Dis 1984;2:229-231.
28.
Williams JE, Cypher JJ, Mosesson MW. Evidence that production of platelet fibrinogen is
synchronous with platelet production in the turpentine-induced acute phase response. J Lab Clin
Med 1985:106;343-348.
29.
Farr M, Scott DL, Constable TJ, Hawker RJ, Hawker CF, Stuart J. Thrombocytosis of active disease.
Ann Rheum Dis 1983;42:545-549.
30.
Waage A, Kaufmann C, Espevik T, Husby G. Interleukin 6 in synovial fluid from patients with
arthritis. Clin Immunol Immunopathol 1989;50:394-398.
31.
Ishibashi T, Kimura H, Shikama Y, Uchida T, Karyone S, Hirano T, Kishimoto T, Takatsuki F,
Akiyama Y. Interleukin 6 is a potent thrombopoeitic factor in vivo in mice. Blood 1989;74:12411244.
32.
Bruno E, Hofman R. Effect of interleukin 6 on in vitro human megakaryocytopoiesis: its interaction
with other cytokines. Exp Hematol 1989;17:1038-1043.
61
Interleukin-6 and humoral and clinical responses in burned patients
62
CHAPTER 6
PROCALCITONIN BEHAVES AS A FAST RESPONDING ACUTE PHASE
PROTEIN IN VIVO AND IN VITRO
M.W.N. Nijsten, P. Olinga, T.H. The, E.G.E. de Vries, H.Schraffordt Koops,
G.M.M. Groothuis, P.C. Limburg, H.J. ten Duis,
H. Moshage, H.J. Hoekstra, J. Bijzet, J.H. Zwaveling
Crit Care Med 2000;28:458-461.
Procalcitonin in vivo and in vitro
ABSTRACT
Objectives: Procalcitonin (PCT) is a 13 kD protein of which plasma levels are strongly increased
in inflammatory states. PCT-levels are claimed to have a more powerful discriminatory value for
bacterial infection than the acute phase proteins serum amyloid A (SAA) or C-reactive protein
(CRP). The source of production and its mechanism of induction are unknown. We investigated
the inducibility of PCT both in vivo and in vitro, and compared PCT’s behavior with SAA and
CRP.
Design: A prospective descriptive patient sample study and a controlled liver tissue culture study.
Setting: A university hospital.
Patients: Cancer patients who were treated with human tumor necrosis factor α (rhTNFα; 5
patients) or interleukin-6 (rhIL-6; 7 patients).
Measurements and Main results: Serial serum samples were collected for analysis of levels of
PCT, SAA and CRP. In the TNFα group frequent sampling was performed on the first day to
allow analysis of initial responses. In a human liver slice model the release of PCT, SAA and CRP
was measured upon induction with rhTNFα and rhIL-6 for 24 hours. We found that in vivo, after
administration of both rhTNFα and rhIL-6, PCT displayed acute phase reactant behavior. After
rhTNFα-administration PCT reached half-maximal levels within 8 hours, 12 hours earlier than
either SAA or CRP. In vitro PCT, SAA and CRP were produced in detectable quantities by liver
tissue. PCT production by liver slices was enhanced after stimulation with rhTNFα or rhIL-6;
SAA and CRP levels were elevated after stimulation with rhTNFα.
Conclusions: We found that PCT and acute phase proteins such as CRP are induced by similar
pathways. The liver appears to be a major source of PCT-production. Thus PCT may be
considered an acute phase protein. PCT’s different kinetics, and not a fundamentally different
afferent pathway may explain its putative diagnostic potential to discriminate bacterial infection
from other causes of inflammation.
INTRODUCTION
Any significant inflammatory process is associated with an acute phase response including the
production of a broad range of effector proteins such as coagulation factors and protease
inhibitors. Acute phase proteins (APP) such as the archetypal C-reactive protein (CRP) are
elevated in a manner proportional to the extent of the inflammatory process. The cytokines tumor
necrosis factor α (TNFα), interleukin-6 (IL-6) and interleukin-1 have been identified as pivotal in
the induction of APPs. Although the theoretical benefits of measuring plasma cytokine levels are
obvious, in clinical practice only APPs such as CRP serve to “quantify” inflammation. The short
half-life (minutes to hours) of cytokines makes relatively infrequent measurements useless certainly if elaborate assay procedures are involved. It has been shown that cytokine levels are
higher in bacteraemia, but clinically it has proven difficult to discriminate causes of inflammation
on the basis of patterns of cytokine or APP responses. In many patients with fever, it is of
paramount importance to discriminate bacterial causes from other causes of inflammation.
64
Chapter 6
It has been claimed that the serum level of the protein procalcitonin (PCT; 116 amino acids; 13
kD) is superior to APPs in differentiating infectious from non-infectious causes of inflammation
[1]. PCT levels are proportional to the inflammatory stimulus and elevated levels are detectable
within 6 hours after the insult and subsequently decrease with a half-life of about one day [2]. The
dynamic range of PCT is very large: in a healthy person PCT-levels are lower than 0.5 µg/L but in
septic patients PCT-levels of up to 1000 µg/L have been measured. In the differentiation of
infected from non-infected pancreatitis PCT was better than CRP [3]. Likewise in one study PCT
appeared better than CRP in detecting bacterial infection in patients with autoimmune disease [4].
On the basis of these results it has been suggested that, in contrast to APPs, PCT is induced
directly by endotoxin without intervening cytokines [2]. Induction of PCT after injection of
endotoxin has been demonstrated in human volunteers [5], but the role cytokines play is unknown.
Many basic questions about PCT are still unanswered. The question which tissue or organ
produces PCT has not been addressed. Mainly based on the fact that PCT is a precursor of
calcitonin, a neuro-endocrine origin has been assumed – but not found. Likewise the mechanism
of induction is unknown. We hypothesized that, analogous to acute phase proteins, PCT is
produced by the liver and induced by TNFα and IL-6 in the absence of endotoxin
In two in vivo studies we investigated whether rhTNFα and rhIL-6 administered to patients can
directly induce PCT. In vitro, the production of PCT by human liver tissue, and its inducibility by
rhTNFα and rhIL-6 were assessed. In vivo as well as in vitro we compared PCT with serum
amyloid A (SAA) and CRP, since SAA and CRP are the fastest responding APPs.
The first study concerned patients treated with hyperthermic isolated limb perfusion with rhTNFα
[6]. Due to leakage of rhTNFα to the systemic circulation these patients display an intense shortlived systemic inflammatory response [7].
The second study concerned cancer patients who received rhIL-6 as a hematopoietic growth factor
[8].
In the in vitro study we used precision-cut liver slices, which are extensively applied in
pharmacology and toxicology [9]. Luster et al. [10] have shown that endotoxin in human liver
slices induced TNFα and IL-6 release. The inflammatory response of the liver is a multi-cellular
phenomenon, involving Kupffer cells, endothelial cells and hepatocytes. In the liver slice all cell
types of the liver are retained. Therefore, liver slices are an attractive model to study integrated
inflammatory responses.
PATIENTS AND METHODS
In all studies pure E.coli-derived rhTNFα (Boehringer, Ingelheim, Germany) or rhIL-6 (Novartis,
Basel, Switzerland) were used. PCT was measured with an immunoluminometric sandwich assay
(LUMItest PCT, Brahms Gmbh, Berlin, Germany; certified normal levels <0.5 µg/L; certified
detection range >0.08 µg/L [2]). SAA and CRP were measured with enzyme linked
immunosorbent assays [11]. (normal SAA < 2.7 mg/L; normal CRP<2.3 mg/L; detection limit for
both assays is 0.001 mg/L).
65
Procalcitonin in vivo and in vitro
100
1000
10
100
1
10
0.1
1
0.01
SAA and CRP (mg/L)
PCT (ug/L)
Patients treated with TNFα or IL-6.
Five patients with nonresectable soft tissue sarcomas of the lower extremity received 90 min of
0.1
0
20
40
Hour
60
80
Figure 6.1. Mean levels (± SEM) of procalcitonin (PCT; filled circles; normal < 0.08 µg/L), serum amyloid
A (SAA; squares; normal < 2.7 mg/L) and C-reactive protein (CRP; diamonds; normal < 2.3 mg/L) in 5
patients in the first 72 hours after the start of perfusion with tumor necrosis factor alpha. Note the two
separate logarithmic scales.
hyperthermic isolated limb perfusion with 4 mg of rhTNFα and melphalan (10 mg/L limb
volume) [6]. Systemic serum samples were analyzed for PCT, SAA and CRP at baseline, at 30
and 90 min after start of rhTNFα-perfusion, as well as at 2, 6, 12, 24, 48 and 72 h after
termination of perfusion [7].
In a study to define toxicity and effects on hematopoiesis and biochemical parameters, patients
with stage III-IV breast cancer or non-small cell lung cancer received rhIL-6 in a constant daily
dosage of 0.5 to 20 µg/kg for 7 successive days (first day intravenously; subsequent 6 days
subcutaneously) [8]. Of a total of 20 patients, 7 patients with the lowest initial SAA and CRP
values were further studied, in order to exclude patients with substantial inflammation before
administration of rhIL-6. These patients received rhIL-6 doses of 0.5, 1, 2.5, 2.5, 5, 5 and 20
µg/kg/day respectively. At the start of rhIL-6 administration patients had a good performance
status. PCT, SAA and CRP were determined at baseline and after 1, 2, 7 and 14 days respectively.
PCT was determined in serum samples that were stored for up to 7 years at -20oC. SAA and CRP
values have been published earlier [8] but a subset of these data are re-used here to facilitate
comparison with PCT.
Liver slices incubated with TNFα or IL-6.
66
Chapter 6
100
1000
10
100
1
10
0.1
1
0.01
SAA and CRP (mg/L)
PCT (ug/L)
Human liver tissue was obtained from redundant material after bipartitioning of livers procured
from multi-organ donors. The human livers were handled as described before [12]. From these
pieces of liver tissue, cores (diameter 8 mm) were made as previously described, and stored in icecold University of Wisconsin organ preservation solution until slicing. Liver slices (200-300 µm
thickness; wet weight 10-14 mg) were prepared with the Krumdieck slicer. The slices were
incubated 24 h in 3.2 mL Williams’ medium E supplemented to 25 mM glucose and 50 µg/mL
gentamicin. The 6-well tissue culture plates were continuously rocked back and forth (90/min) at
37 oC under 95% O2 and 5% CO2. In addition to unstimulated control slices, slices were
stimulated with rhTNFα or rhIL-6 (final concentration 15.6 ng/mL) that was added at the start of
slice incubation. The liver slices were incubated in triplicate, and both induction studies and
controls were performed in slices from three donor livers. PCT, SAA and CRP were measured in
the culture medium after 24 h.
0.1
0
7
Day
14
Figure 6.2. Mean levels (±SEM) of procalcitonin (PCT; filled circles), serum amyloid A (SAA;
squares) and C-reactive protein (CRP; diamonds) in 7 patients in the two weeks following the start of
interleukin-6 administration. The period of interleukin-6 administration is indicated by the hatched bar.
All studies were approved by the medical ethical commission of our hospital. Patients treated with
rhTNFα or rhL-6 gave written informed consent. Consent from the legal authorities and from the
families concerned was solicited for the explantation of organs for transplantation purposes.
To examine the statistical significance of induction of PCT, SAA and CRP in the two patientgroups, areas under the curve were calculated for each patient after log-transformation. Likewise,
in vitro PCT, SAA and CRP results were also log-transformed for statistical analysis. Student’s t-
67
Procalcitonin in vivo and in vitro
test was used to assess the significance of differences. A p-value <0.05 was considered
significant.
RESULTS
Patients treated with TNFα or IL-6.
As early as 3.5 h after the start of the rhTNFα-perfusion patients had already mean PCT-levels of
1.3 µg/L compared to baseline levels of less than 0.1 µg/L. PCT rose to a maximal level of 28
µg/L. Maximal SAA and CRP-levels were 392 and 196 mg/L respectively, compared to baseline
levels of 1.9 and 0.9 mg/L respectively (Fig. 6.1). PCT reached a half-maximal level within 8 h,
whereas SAA and CRP reached half-maximal values after 20 h. Stimulation of PCT, SAA and
CRP by rhTNFα as assessed by the areas under the curve was significant with a p-value <0.0001.
The patients that received rhIL-6 had maximal levels of PCT, SAA and CRP of 7.8 µg/L, 634
mg/L and 196 mg/L respectively, compared to baseline levels of 0.1 µg/L, 6 mg/L and 7 mg/L
respectively (Fig. 6.2). Stimulation of PCT, SAA and CRP by rhIL-6 as assessed by the areas
under the curve was significant with a p-values of 0.02, 0.0004 and 0.0002 respectively. As
comparison of Fig. 6.1 and Fig. 6.2 shows, increased PCT levels after 1 and 2 days were 9 times
higher in the TNFα group than in the IL-6 group.
Liver slices incubated with TNFα or IL-6.
After incubation of the liver slices unstimulated mean (±SEM) production of PCT (0.4±0.2 µg/L),
SAA (0.27±0.06 mg/L) and CRP (0.52±0.09 mg/L) could be detected. Synthesis of PCT was
increased by stimulation with rhTNFα and rhIL-6 to 274 (p=0.008) and 223 percent (p=0.001) of
control levels respectively. rhTNFα induced significant increases in SAA and CRP production to
194 percent (p=0.02) and 147 percent (p=0.03) (Fig. 6.3); increases of SAA and CRP after rhIL-6
were not significant (p=0.13 and p=0.24 respectively).
DISCUSSION
To our knowledge this is the first study to demonstrate PCT induction in vivo after administration
of TNFα or IL-6. In addition this is the first study to show that PCT was produced by liver tissue
in vitro after stimulation with TNFα or IL-6.
In vivo PCT has a unique time response. The early rise of PCT we observed after the
administration of TNFα was as rapid as observed by Dandona et al. after endotoxin administration
[5]. When compared to SAA and CRP the rise of PCT occurred 12 h earlier, with a subsequent
fall in circulating levels that is comparable to the fall of SAA or CRP levels.
In vitro, after stimulation with TNFα the levels of PCT, SAA and CRP were all increased above
control values. The fact that no significant increase in SAA and CRP was found after stimulation
with IL-6, is in accordance with the concept of cytokine-specific APP-activation: in vitro socalled type I APPs such as SAA and CRP are preferentially induced by TNFα or IL-1, whereas
type II APPs such as fibrinogen are preferentially induced by IL-6 [13]. However, for PCTinduction, under the culture conditions used, no such selectivity was observed. When comparing
responses it is important to realise that differences found in PCT release in vivo and in vitro may
still be related to a dose dependency of the induction by TNFα and IL-6.
68
Chapter 6
PCT offers interesting opportunities for further study in the liver slice model, since in vitro PCTlevels were far above the detection threshold of the commercial PCT-assay we used. Since
endotoxin can induce cytokines in liver slices [10], it needs to be verified if endotoxin can induce
PCT in liver slices, possibly by signalling between the different cell types present in the liver
slice. Such an integrated in vitro model may offer possibilities to study intervention in the
inflammatory cascade as it occurs in the human liver.
It is not necessary to postulate that PCT is principally induced by endotoxin. In fact, a recent
study in trauma patients showed that PCT and CRP were related to inflammation caused by injury
and were not related to infection [14]. Early sampling after the onset of inflammation and the
initially discordant responses of PCT and CRP (or SAA) may explain why some studies revealed
a better association of infection with PCT than CRP, whereas a recent study showed that PCT was
% stimulation
400
300
200
100
PCT SAA CRP
PCT SAA CRP
Figure 6.3:
Stimulation of production procalcitonin (PCT), serum amyloid A (SAA) and C-reactive protein (CRP)
by liver slices by tumor necrosis factor α (closed bars) and interleukin-6 (open bars). Mean production
levels after 24 h (±SEM) in 9 stimulated slices are expressed as percentage of mean unstimulated
control levels in 9 slices. Significant increases are denoted by a star.
not a better predictor than CRP for infection [15]. In estimating the sensitivity and specificity of
elevated PCT in regard to serious inflammation it is also important to carefully define normal
values. PCT’s normal upper limit may well be much lower than the 0.5 µg/L value currently used
[2], since others report a PCT-level of only 0.003 µg/L in pooled serum from 13 healthy
individuals [16].
Although neither “acute phase protein” nor “acute phase response” has been uniformly defined
[13], we think PCT should seriously be considered an APP since it increases sharply after
inflammation, it is produced by liver tissue and it can be induced by TNFα or IL-6. The half-life
of elevated PCT-levels of about one day is also characteristic for APPs. As nearly all APPs are
synthesized in hepatocytes, this still needs to be verified for PCT, since we only measured PCT in
69
Procalcitonin in vivo and in vitro
a system containing multiple cell types (i.e. hepatocytes, endothelial cells and Kupffer cells).
Whether PCT qualifies as an APP will ultimately depend on the clarification of its function.
Recent experiments by Nylen and colleagues with a peritonitis model in Syrian hamsters showed
excess mortality after administration of PCT [17]. However, this study is hampered by the use of
human PCT in animals, the use of non-specific antibodies and mortality differences between
control groups [18]. If PCT turns out to have a paracrine or endocrine function it might be a
cytokine, if PCT has effector functions (e.g. protease inhibition) it belongs to the APP-family.
In conclusion, our results as well as those published by others strongly suggest that PCT is an
APP. Irrespective of PCT’s unidentified function, its unique kinetics compared with both acute
phase proteins and with cytokines may establish PCT as a particularly useful parameter of
inflammation in critical care areas.
ACKNOWLEDGEMENTS
Parts of these studies were supported by grants from Organon International NV and Solvay
Pharmaceuticals BV.
REFERENCES
1
Assicot M, Gendrel D, Carsin H, Raymond J, Guilbaud J, Bohuon C. High serum procalcitonin
concentrations in patients with sepsis and infection. Lancet 1993; 341:515-518.
2
Meisner M. Procalcitonin. A new innovative infection parameter. Biochemical and clinical aspects.
Meisner M. Berlin: Brahms Diagnostica 1996. ISBN 3-00-000803-9.
3
Rau B, Steinbach G, Gansauge F, et al. The potential role of procalcitonin and interleukin 8 in the
prediction of infected necrosis in acute pancreatitis. Gut 1997; 41:832-840.
4
Eberhard OK, Haubitz M, Brunkhorst FM, Kliem V, Koch KM, Brunkhorst R. Usefulness of
procalcitonin for differentiation between activity of systemic autoimmune disease (systemic lupus
erythematosus/systemic antineutrophil cytoplasmic antibody-associated vasculitis) and invasive
bacterial infection. Arthritis Rheum 1997; 40:1250-1256.
5
Dandona P, Nix D, Wilson MF, Aljada A, Love J, Assicot M, Bohuon C. Procalcitonin increase
after endotoxin injection in normal subjects. J Clin Endocrinol Metab 1994; 79:1605-1608.
6
Eggermont AM, Schraffordt Koops H, Lienard D, Kroon BB, van Geel AN, Hoekstra HJ, Lejeune
FJ. Isolated limb perfusion with high dose tumor necrosis factor α in combination with interferon γ
and melphalan for irresectable extremity soft tissue sarcomas: multicenter trial. J Clin Oncol 1996;
14:2653-2665.
7
Zwaveling JH, Maring JK, Clarke FL, van Ginkel RJ, Limburg PC, Hoekstra HJ, Koops HS,
Girbes AR. High plasma tumor necrosis factor (TNF)-alpha concentrations and a sepsis-like
syndrome in patients undergoing hyperthermic isolated limb perfusion with recombinant
TNF-alpha, interferon-gamma, and melphalan. Crit Care Med 1996; 24:765-770.
70
Chapter 6
8
van Gameren MM, Willemse PHB, Mulder NH, Limburg PC, Groen HJ, Vellenga E, de Vries EG.
Effects of recombinant human interleukin-6 in cancer patients: a phase I-II study. Blood 1994;
84:1434-1441.
9
Olinga P, Meijer DKF, Slooff MJH, et al. Liver slices in in vitro pharmacotoxicology with special
reference to the use of human liver tissue. Toxicol in Vitro 1998; 12:77-100.
10
Luster MI, Germolec DR, Yoshida T, et al. Endotoxin-induced cytokine gene expression and
excretion in the liver. Hepatology 1994; 19:480-488.
11
Hazenberg BPC, Limburg PC, Bijzet J, et al.: Monoclonal antibody base ELISA for human SAA.
In: Amyloid and Amyloidosis 1990. Eds Natvig JB, Forre O, Husby G, et al. Kluwer Academic
Publishers, Dordrecht/ Boston/ London 1991:898-901.
12
Olinga P, Merema MT, Hof IH, de Jong KP, Slooff MJ, Meijer DK, Groothuis GM. Effect of
human liver source on the functionality of isolated hepatocytes and liver slices. Drug Metab
Dispos 1998; 26:5-11.
13
Moshage H: Cytokines and the hepatic acute phase response. J Pathol 1997; 181:257-266.
14
Mimoz O, Benoist JF, Edouard AR, Assicot M, Bohuon C, Samii K. Procalcitonin and C-reactive
protein during the early posttraumatic systemic inflammatory response syndrome. Intensive Care
Med 1998; 24:185-188.
15
Ugarte H, Silva E, Mercan D, de Mendonca A, Vincent JL. Procalcitonin used as a marker of
infection in the intensive care unit. Crit Care Med 1999; 27:498-504.
16
Snider RH Jr, Nylen ES, Becker KL. Procalcitonin and its component peptides in systemic
inflammation: immunochemical characterisation. J Investig Med 1997; 45:552-560.
17
Nylen ES, Whang KT, Snider RH, Steinwald PM, White JC, Becker KL. Mortality is increased by
procalcitonin and decreased by an antiserum reactive to procalcitonin in experimental sepsis. Crit
Care Med 1998;26:1001-1006.
18
Braithwaite SS. Procalcitonin – marker or mediator? Crit Care Med 1998; 26:977-978.
71
CHAPTER 7
EARLY POST-TRAUMATIC THROMBOCYTOPENIA IS NOT AFFECTED
BY HIGH-DOSE METHYL-PREDNISOLONE
M.W.N. Nijsten, K. Bartelet, M. Bos, R.J. Porte, T.H. The, H.J. ten Duis
Submitted
Early post-traumatic thrombocytopenia and methylprednisolone
ABSTRACT
Background: After trauma and sepsis platelet counts (PC) decrease during the first two days.
Except in massive bleeding,
this is primarily due to inflammation-induced platelet sequestration. Changes in PC are strongly
related with outcome. In contrast to animal experiments, the effect of early steroids on PC in
critically ill patients has not been studied. Spinal cord injury is currently the only widely accepted
indication for the administration of high-dose steroids. We investigated if early
methylprednisolone (MPS) affected platelet sequestration in patients with spinal injury.
Methods: In the MPS-group, patients with vertebral and spinal cord injury received 30 mg/kg of
MPS followed by an infusion of 5.4 mg/kg/hr for the next 23 hrs. Controls were patients with
vertebral fractures who received no steroids. Patients who received any platelet transfusion or
excessive red cell transfusion were excluded. PC and hemoglobin were determined on day 0, 1
and 2.
Results: In 24 patients MPS was administered on average 3.8 hrs after the injury. 41 patients were
included in the control group. Both groups were comparable in age, Injury Severity Score and
blood loss. Between day 0 and day 2 the PC decreased by 41% and 37% in the MPS and control
group respectively. In both groups hemoglobin decreased by 14%.
Conclusion: High-dose MPS started within 4 hours does not inhibit platelet sequestration during
the first 2 days after trauma. This inability of MPS to prevent platelet sequestration may be
related to the failure of MPS to affect mortality in clinical trials.
INTRODUCTION
In major trauma and sepsis, platelet sequestration is intimately linked with systemic inflammation
and possibly with disseminated intravascular coagulation (DIC) [1].As a result of the trauma the
platelet count (PC) decreases during the first days. In severely injured patients the extent of early
thrombocytopenia is strongly associated with the subsequent occurrence of multiple organ
dysfunction syndrome [2]. In a recent study on 1415 patients admitted to the surgical intensive
care unit (ICU) we found that the rate of change in platelet count was associated with mortality
[3]. This phenomenon was observed in all subgroups: abdominal surgery, vascular surgery, liver
transplantation and trauma. In medical ICU patients it has also been observed that the magnitude
of decreases in PC constitutes an important, independent marker for mortality [4].
Experimental trauma and sepsis studies have shown that very early (i.e. before or <60 minutes of
the insult) administration of corticosteroids has a positive effect on mortality in parallel with a
very marked reduction of platelet sequestration [5,6]. On the other hand, well-conducted clinical
intervention studies in critically ill patients have all consistingly failed to improve survival. In
addition to high-dose corticosteroids [7,8,9] these interventions included anti-endotoxin
antibodies, several cytokine inhibitors, platelet activating factor (PAF) antagonist and bradykinin
antagonist [10]. These studies did not report on the effect of the intervention on quantitative
changes in platelet counts.
74
Chapter 7
To our knowledge, acute spinal cord injury is the only widely accepted indication for the
administration of high-dose corticosteroids after acute injury in particular and acute critical illness
in general. The second and third National Acute Spinal Cord Injury Studies [11,12] (NASCIS-2
and -3) reported a significantly better neurologic outcome for patients treated with
methylprednisolone (MPS) within 8 hours of the injury (30 mg/kg bolus followed by 5.4 mg/kg/hr
for 23 hrs or 48 hrs if started more than 3 hrs after the injury). Although the design and
interpretation of the NASCIS studies have been questioned [13], the NASCIS-protocol is widely
applied, not in the least because of the devastating nature of this trauma. The effect of MPS on
platelet counts was not studied by the NASCIS investigators. In the current study we investigated
patients with spinal cord injury who received MPS according to NASCIS guidelines. Patients with
vertebral fractures and otherwise comparable injuries, but without spinal cord injury were used as
controls. The aim of the study was to assess if high-dose corticosteroids inhibit early platelet
sequestration.
Table 7.1
Patient characteristics
Number of patients
Age (+/-SD)
Male/Female
Died
MPS group
24
38+/- 15
18/6
2
Control group
41
39+/-19
35/6
1
Vertebral fractures
Cervical
Thoracic
Lumbal
Sacral
6
13
4
2
9
15
19
4
Mean ISS ±SD (range)
Patients operated on day 0,1, 2
24±11 (9-41)
8
21±8 (4-34)
16
MPS: methylprednisolone; ISS: injury severity score.
PATIENTS AND METHODS
Patient data were gathered from the hospital data base, from a dedicated trauma database that uses
International Classification of Diseases (ICD-9) codes and from patient files. Patients with ICD-9
code 806 (vertebral injury with spinal cord lesion) who received MPS (methylprednisolone
sodium succinate, Pharmacia-Upjohn, Peapack , New Jersey, USA)
according to the NASCIS protocol were included in the MPS group. Patients with ICD-9 code 805
who received no steroids since they had no spinal injury, were included in the control group. To
minimize confounding effects of excessive injury and bleeding necessitating red cell or platelet
transfusion the following patients were excluded: patients who died within one week, patients
who received any platelet transfusion and patients who received more than 4 units of red cells per
day during the first three days. In all patients the injury severity score (ISS) was determined [14].
Platelet count and hemoglobin were determined on day 0, 1 and 2. Statistical differences between
mean values were assessed with the Student t-test.
75
Early post-traumatic thrombocytopenia and methylprednisolone
RESULTS
The MPS group and control group included 24 and 41 patients respectively. Groups were
comparable in age and ISS (Table 7.1). The mean delay between injury and MPS-treatment was
3.8 hrs (range 0.5-7 hrs). Two MPS patients died and one control patient died during hospital stay
(after 9, 13 and 29 days respectively). In the MPS group 9 patients received a total of 23 red cell
units on day 0, 1 and 2; in the control group 13 patients received 38 units over the same period.
For the two groups taken together, the mean±SD platelet count dropped from 215±56 on day 0 to
133±41 ·103/mm3 on day 2 (a 38% decrease; p<0.001). Over the same period overall hemoglobin
dropped from 11.0±2.6 to 9.4±1.9 g/L (a 14% decrease; p<0.001). The difference between the
relative decrease in PC and the relative decrease in hemoglobin was more than a factor 2 and
significant (p<0.001).
Fig. 7.1 displays the changes of platelet count and hemoglobin for the MPS and the control
groups. No differences between the two groups were observed in the decrease of platelet count
12
Hemoglobin
8
Platelet count
150
6
100
4
50
Hemoglobin g/dL
10
200
3
Platelet count 10 /mm
3
250
2
Methylprednisolone
0
0
0
1
2
Days post Trauma
Figure 7.1. Time course (means+/- SE) of platelet counts (filled symbols) and hemoglobin (open symbols) in
patients with vertebral fractures treated with high-dose methylprednisolone (MPS group; circles) and patients
(control group; squares) that were not treated with MPS. No significant differences were observed between the
MPS-group and the control group for the decrease in PC (41 % and 37 % respectively) and the decrease in
hemoglobin (14 % for both groups). Thus early high-dose MPS does not inhibit platelet sequestration. The
period during which the patients received MPS is shown on the time axis.
and hemoglobin. Also within both groups the relative or fractional drop in platelet count was again
significantly higher than the relative drop in hemoglobin (p<0.001 and p<0.001 respectively). To
76
Chapter 7
further minimize confounding effects of transfusion or operation we also analyzed patients who
received no red cell transfusion and who were not operated (11 MPS patients and 22 control
patients). Again, no significant differences between the MPS and the control group were seen; the
relative drop in PC of 34% was significantly higher than the 15% drop in hemoglobin (p<0.001).
DISCUSSION
Platelet counts significantly decreased after admission in the patients studied. The observed
decrease in platelet count is much greater than can be explained on the basis of blood loss alone.
The 38% drop in platelet count in all patients is more than double the fractional decrease in
hemoglobin. When the effect of the spleen as an exchangable pool of platelets is taken into
account, even more platelets must have disappeared from the circulation compared to red cells.
Since in comparable groups the same decrease in platelet count was observed with or without
MPS, MPS clearly was not effective in preventing platelet sequestration.
Both the unparalleled succes of steroids and the difficulties in replacing corticosteroids with more
modern treatments are related to the very broad spectrum of their anti-inflammatory actions. These
effects include inhibition of many cytokines and mediators, reduction in vascular permeability
and inhibition of several leukocyte types [15]. In the patients studied, even if additional or
‘secondary’ inflammatory responses are suppressed by high-dose MPS the initial events
apparently irreversibly induce sustained platelet sequestration. Whatever the nature of any
pharmacologic intervention after trauma, administering it much earlier than 3.8 hours post-trauma
appears will be difficult in the majority of cases. In the NASCIS-2 and -3 studies, where the
participants were obviously aware of the importance of rapid treatment, the delay to MPSadministration was 8.7 and 3 hrs respectively [11,12].
We prefer to use the term platelet sequestration since it includes both adhesion of platelets to the
vascular wall, and aggregation of platelets in clots. Although DIC is not uniformly defined [16],
demonstrating the consumption of platelets as well as coagulation factors is essential for the
diagnosis. To some extent DIC is present in many patients after major trauma [2], but we did not
measure coagulation parameters to assess the extent of possible DIC in our patients. But DIC is
not a prerequisite for platelet sequestration, since thrombocytopenia often occurs in the absence of
DIC [17]. The 'target' organs of platelet sequestration after trauma or sepsis are many: 111Indiumlabelled platelets have been located in the gut, lung, liver and spleen, especially in patients with
poor outcomes [18]. Regardless of the target organ, the endothelial cell is (by definition) the main
cell with which platelets interact. Various inflammatory stimuli can induce endothelial cell
activation [19] that can trigger coagulation and platelet adhesion, e.g. by inducing release of
tissue factor [20]. Correlating the prolonged platelet sequestration observed in vivo with in vitro
studies is not trivial since it is very difficult to realistically reproduce the interaction between
endothelial cells and platelets. In an interesting in vivo experiment [21] endothelial dysfunction
developed after a 1-hr perfusion with endotoxin of the forearm vessels in healthy persons. This
endothelial "stunning" lasted for more than 48 hours. Thus activation of platelets and activation of
endothelial cells may be present in the absence of significant DIC.
As indicated above, in many trauma and sepsis experiments but not in clinical studies, limiting the
decreases in platelet count has been used as a surrogate goal. In observational studies low early
77
Early post-traumatic thrombocytopenia and methylprednisolone
platelet counts were predictive of poor outcome in patients with sepsis or patients with a ruptured
abdominal aortic aneurysm [22,23]. In prognostic models in critical care, the platelet count has
emerged as an important component. For example the multi-organ dysfunction [24] and the
sequential organ failure assessment scores [25] use the platelet count and not the leukocyte count
as one of their components parameters, which in contrast to the older Acute Physiology and
Health Evaluation (APACHE) score [26]. Recent studies have shown that the rate of change in
the platelet count is even a better marker of outcome. Failure of the platelet count to recover
sufficiently after reaching nadir values was associated with poor outcome in medical [4] and
surgical [3] intensive care patients. In surgical ICU patients we found that the rate of change in
platelet count had the the same predictive power for mortality as APACHE-II scores [3].
However, in the immunomodulatory clinical intervention trials [10] platelet count has not been used
as a major parameter. No intervention effects on the platelet count have been reported - although the
platelet count was measured in many trials as part of semiquantitative hematological or coagulation
scores. It would be of interest to re-analyse existing clinical trial databases and compare quantitative
changes in platelet count with the intervention and with outcome.
Thus many lines of evidence point to a relation of platelet count with outcome in a variety of
seriously ill patients. The pathophysiologic importance of platelet sequestration in regard to
outcome can only be proven by selectively blocking this process. Promising interventions
'downstream' the inflammatory cascade may be possible at clinically feasible times. Inhibitors of
tissue-factor activity may be very effective in limiting DIC and platelet sequestration [16].
Antithrombin-III and especially glycoprotein IIb/IIIa (GP IIb/IIIa) -inhibitors have both shown antiinflammatory effects. The GP IIb/IIIa-inhibitors that selectively inhibit platelet adhesion have
shown dramatic effects on coronary platelet sequestration and mortality in several clinical trials [27].
In conclusion we have shown that high dose steroids started within hours after major injury do not
influence subsequent systemic platetet sequestration. We believe that limiting the decrease in
platelet count as a goal in intervention studies remains useful. How instrumental the role of plaletet
sequestration is in systemic inflammation, needs to be addressed by specific intervention studies.
78
Chapter 7
REFERENCES
1.
Gando S, Kameue T, Nanzaki S, Nakanishi Y. Disseminated intravascular coagulation is a
frequent complication of systemic inflammatory response syndrome. Thromb Haemost
1996;75:224-228.
2.
Gando S, Nanzaki S, Kemmotsu O. Disseminated intravascular coagulation and sustained
systemic inflammatory response syndrome predict organ dysfunctions after trauma: application of
clinical decision analysis. Ann Surg 1999;229:121-127.
3.
Nijsten MWN, ten Duis HJ, Zijlstra JG, Porte RJ, Zwaveling JH, Paling JC, The TH. Blunted rise
in platelet count in critically ill patients is associated with worse outcome. Crit Care Med
2000;28:3843-3846.
4.
Vanderschueren S, de Weerdt A, Malbrain M, Vankerschaever D, Frans E, Wilmer A, Bobbaers
H. Thrombocytopenia and prognosis in intensive care. Crit Care Med 2000;28:1871-1876.
5.
Hardaway RM, Williams CH, Dozier SE. Influence of steroids on hemorrhagic and traumatic
shock. J Trauma 1987;27:667-670.
6.
Yoshikawa T, Murakami M, Furukawa Y, Takemura S, Kondo M. Prevention by
methylprednisolone of disseminated intravascular coagulation induced by sustained infusion of
endotoxin in rats. Haemostasis 1983; 13:268-273.
7.
Sprung CL, Caralis PV, Marcial EH, Pierce M, Gelbard MA, Long WM, Duncan RC, Tendler
MD, Karpf M.
The effects of high-dose corticosteroids in patients with septic shock. A
prospective, controlled study. N Engl J Med 1984;311:1137-1143.
8.
Bone RC, Fisher CJ Jr, Clemmer TP, Slotman GJ, Metz CA, Balk RA. A controlled clinical trial
of high-dose methylprednisolone in the treatment of severe sepsis and septic shock. N Engl J Med
1987;317:653-658
9.
The Veterans Administration Systemic Sepsis Cooperative Study Group. Effect of high-dose
glucocorticoid therapy on mortality in patients with clinical signs of systemic sepsis. N Engl J
Med 1987;317:659-665.
10.
Abraham
E.
Why
immunomodulatory
therapies
have
not
worked
in
sepsis.
Intensive Care Med 1999;25:556-566
11.
Bracken MB, Shepard MJ, Collins WF, et al. A randomized, controlled trial of
methylprednisolone or naloxone in the treatment of acute spinal cord injury: results of the second
National Acute Spinal Cord Injury Study. N Eng J Med 1990;322:1405-1411.
12.
Bracken MB, Shepard MJ, Holford TR, et al. Administration of methylprednisolone for 24 or 48
hours or tirilazad mesylate for 48 hours after acute spinal cord injury: results of the third National
Acute Spinal Cord Injury Randomized Controlled Trial. JAMA 1997;277:1597-1604.
13.
Nesathurai S. Steroids and spinal cord injury: revisiting the NASCIS 2 and 3 trials. J Trauma
1998. 45:1088-1093.
14.
Baker SP, O'Neill B. The injury severity score: un update. J Trauma 1976;16:882-885.
15.
Schleimer RP. An overview of glucocortocoid anti-inflammatory actions. Eur J Clin Pharmacol
1993;45 Suppl 1;S3-7.
16.
Levi M, ten Cate H. Disseminated Intravascular coagulation. N Eng J Med 1999;341:586-592.
79
Early post-traumatic thrombocytopenia and methylprednisolone
17.
Neame PB, Kelton JG, Walker IR, Stewart IO, Nossel HL, Hirsch J. Thrombocytopenia in
septicemia:the role of disseminated intravascular coagulation. Blood 1980;56:88-92.
18.
Sigurdsson GH, Christenson JT, el-Rakshy MB, Sadek S. Intestinal platelet trapping after traumatic
and septic shock. An early sign of sepsis and multiorgan failure in critically ill patients? Crit Care
Med 1992; 20:458-467.
19.
Hunt BJ, Jurd KM. Endothelial cell activation. A central pathophysiological process.
BMJ
1998;316:1328-1329.
20.
Gando S, Nanzaki S, Sasaki S, Kemmotsu O. Significant correlations between tissue factor and
thrombin markers in trauma and septic patients with disseminated intravascular coagulation.
Thromb Haemost 1998;79:1111-1115.
21.
Bhagat K, Moss R, Collier J, Vallance P. Endothelial stunning following brief exposure to
endotoxin: a mechanism to link infection to infarction? Cardiovasc Res 1996;32:822-829.
22.
Bonfiglio MF, Traeger SM, Kier KL, Martin BR, Hulisz DT, Verbeck SR. Thrombocytopenia in
intensive care patients: a comprehensive analysis of risk factors in 314 patients. Ann Pharmacother
1995;29:835-842.
23.
Bradbury AW, Bachoo P, Milne AA, Duncan JL. Platelet count and the outcome of operation for
ruptured abdominal aortic aneurysm. J Vasc Surg 1995;21:484-491.
24.
Marshall JC, Cook DJ, Christou NV, Bernard GR, Sprung CL, Sibbald WJ. Multiple organ
dysfunction score: a reliable descriptor of a complex clinical outcome. Crit Care Med
1995;23:1638-1652.
25.
Vincent JL, de Mendonca A, Cantraine F, et al. Use of the SOFA score to assess the incidence of
organ dysfunction/failure in intensive care units: results of a multicenter prospective study. Crit
Care Med 1998; 26:1793-1800.
26.
Knaus WA, Draper EA, Wanger DP, Zimmerman JE. APACHE-II: a severity of disease
classification system. Crit Care Med 1985;13:818-829.
27.
Cannon CP. Incorporating platelet glycoprotein IIb/IIIa inhibition in critical pathways: unstable
angina/non-ST-segment elevation myocardial infarction. Clin Cardiol 1999;22(Suppl):IV30-36.
80
CHAPTER 8
BLUNTED RISE IN PLATELET COUNT IN CRITICALLY ILL PATIENTS
IS ASSOCIATED WITH WORSE OUTCOME
M.W.N. Nijsten, H.J. ten Duis, J.G. Zijlstra,
R.J. Porte, J.H. Zwaveling, J.C. Paling, T.H. The
Crit Care Med 2000;28:3843-3846.
Blunted rise in platelet counts in SICU patients
ABSTRACT
Objective: Low platelet counts (PC) are known to be associated with disease severity in critically ill
patients, but the relevance of time-dependent changes of PC has not been investigated. We tested the
hypothesis that a low rate of change of platelet counts (PC) after admission to the intensive care unit
(ICU) is associated with mortality.
Design: Retrospective study.
Setting: A 12-bed surgical ICU of a university hospital.
Patients: All adult patients admitted for at least 4 days to the ICU over a 7-year period.
Methods On admission Acute Physiology and Chronic Health Evaluation (APACHE-II)-scores
were calculated. PC and leukocyte counts were analyzed from admission to day 10. The daily rise of
the platelet count (∆PC/∆t) from day 2 to day 10 was calculated. 30-day mortality as well as hospital
mortality were determined.
Results: 1415 admissions were studied. Median PC (interquartile range) initially decreased and
subsequently increased, with a higher PC in 1203 survivors than in 212 non-survivors from day 2
onward (302 (181- 438) vs. 129 (62- 228) ⋅109/L at day 10; p<0.001). After stratification of patients
per type of surgery, within each group PC was also higher in survivors. Mean ∆PC/∆t was more than
5 times higher in survivors compared to non-survivors: 30±46 vs. 6±28 ⋅109/L/day (p<0.001). The
area under the receiving-operating-characteristic curve of ∆PC/∆t for 30-day survival was 0.743,
compared to 0.728 for APACHE-II. Leukocyte counts showed marginal differences between nonsurvivors and survivors.
Conclusion: A blunted or absent rise in PC in critically ill patients is associated with increased
mortality. ∆PC/∆t is a readily available and cheap parameter to improve assessment of critically ill
patients.
INTRODUCTION
Increased platelet production is a normal response after inflammatory insults such as trauma or
infection. When patients recover after a significant inflammatory event, thrombocytosis is typically
observed after approximately one week, with platelet counts leveling off to normal levels afterwards.
This phenomenon can be seen as a late part of the acute phase response and has been shown to be
mediated by cytokines, especially interleukin-6 [1,2]. In fact, in patients with chronic diseases such
as rheumatoid arthritis or inflammatory bowel disease, thrombocytosis is a marker of persistent
inflammation.
On the high end of the spectrum of disease severity - e.g. sepsis or major bleeding - low platelet
counts are correlated with disease severity. Significant initial thrombocytopenia (platelet count
<50⋅109/L) is known to be predictive of poor outcome in patients with sepsis or with a ruptured
abdominal aortic aneurysm [3,4]. Once such patients have survived the first days following the acute
event a complex interplay of factors will determine the ultimate outcome. Many parameters
considered either pro-inflammatory or anti-inflammatory have been studied in the quest to predict
and to influence outcome. Clinically useful measurement of parameters such as cytokines is difficult
and associated with considerable costs.
Although many studies have recognized the value of baseline platelet counts in predicting outcome,
82
Chapter 8
the meaning of subsequent time-dependent changes in platelet counts that do not excessively
deviate from so-called normal values has not been studied. We have often noted that critically ill
patients do not display thrombocytosis, although such a response might be expected because of the
strong inflammatory stimuli that are present. To assess the clinical relevance of this phenomenon we
systematically looked at platelet counts after admission to the intensive care unit (ICU), In addition
leukocyte counts were also studied since leukocyte counts are frequently measured as an indicator of
inflammation.
We hypothesized that a lower rise in platelet counts in the days following ICU-admission would be
associated with adverse outcome. Since the objective was to observe sequential changes in the
platelet count in patients that were especially at risk for a complicated course, we only studied
patients that stayed at least 4 days in the unit.
Survivors
Non-survivors
N
1203
212
First admissions
1109
202
Readmissions
94
10
Male/Female
777 / 426
133 / 79
Age
55±19
57±19
APACHE-II
18.0±6.7
23.6±6.9
Trauma
259
27
Vascular surgery
137
32
Abdominal surgery
185
38
Liver transplantation
186
21
Miscelleneous
436
94
Groups
Table 8.1 Patients studied
Note that patients are counted on admission basis. APACHE-II scores are significantly
different between the two groups (p<0.001)
PATIENTS AND METHODS
All patients 15 years and older who were admitted between 1992 and 1998 to the surgical ICU of a
tertiary teaching hospital for 4 days or more were retrospectively studied. Patients had to stay (and
thus survive) at least 4 days in the unit in order to be included. Based on type of surgery, the
following groups were defined: trauma, vascular surgery, abdominal surgery, liver transplantation
and miscellaneous. All platelet count (normal range 150 to 350 ⋅109/L) and leukocyte count
(normal range 4 to 10 ⋅109/L) measurements performed from ICU-admission to 10 days after ICU
admission were analyzed. During ICU-stay, these measurements were performed daily by Coulter
83
Blunted rise in platelet counts in SICU patients
500
500
400
300
Platelet count
Platelet count
400
200
100
300
200
100
0
0
0
0
2
2
4
4
6
6
Days after ICU-admission
8
10
8
10
Days after ICU-admission
9
/L ; interquartile
range) forrange)
survivors
circles) and
Median
platelet
counts
(10counts
Figure
8.2a.
after (open
ICU-admission
fornonsurvivors (open
Overall
median
platelet
(109/L ; interquartile
Figure
8.1.
survivors
(closed
circles)
in
trauma
patients.
The
difference
is
significant
(p<0.001).
circles) and patients that died within 30 days of ICU-admission (non-survivors; closed circles). Non-survivors
do not show the rise in platelet counts observed in survivors (p<0.001).
Counter (Beckman Coulter, Fullerton, California, United States). Acute Physiology and Chronic
Health Evaluation (APACHE-II; [5]) scores were calculated on admission. Survival at 30 days after
admission to the ICU and hospital mortality were determined.
Statistics
Readmissions to the ICU were also included, provided those occurred more than 30 days after a
previous ICU-admission. Thus a patient who was admitted previously once and died during a later
readmission was counted once as a survivor, and once as a non-survivor. When multiple
determinations were made for a single patient on the same day, these values were averaged to one
value for that day before daily group means were calculated. To quantify the daily increase of the
platelet count linear regression estimates were calculated for each individual patient, as well as for
patient groups. On the basis of the known bimodal change of platelet counts, the platelet count
(PC) was assumed to change in a linear fashion from day 2 to day 10, as described by
PC=a⋅t + b.
For each patient the constant a (slope) was calculated, denoted by ∆PC/∆t hereafter.
To assess the relation between the APACHE-II score and ∆PC/∆t with 30-day survival and hospital
survival respectively, receiving-operating-characteristic (ROC) curves were constructed,
84
Chapter 8
with the area under the curve as a measure of discriminatory ability [6]. Data are expressed as
medians with interquartile ranges or means±SD. Significances of differences were assessed with a
Student’s t-test, and corrected according to Bonferroni in case of multiple comparisons.
RESULTS
During the 7-year study period a total of 3286 different patients aged 15-96 years, had 3940
admissions to our surgical ICU. Of these patients 383 died within 30 days. When only patients
who were admitted for more than 4 days were included, 1311 patients (1415 admissions) remained
for further analysis. In this group 212 (15% of admissions) patients died within 30 days after ICUadmission (Table 8.1). In 277 admissions the patients died during hospital stay. A total of 17364
platelet counts and 13695 leukocyte counts were analyzed. The median platelet count of the
entire patient group initially dropped to a nadir of 113 (64- 192) ⋅109/L on the second day after
admission, and subsequently increased to 277 (157- 424) ⋅109/L after 10 days.
When survivors and non-survivors were considered separately (Fig. 8.1) the platelet count on
admission showed no difference. From day 2 to day 10 the platelet count in survivors was
significantly higher than in non-survivors resulting in a platelet count at day 10 of 302 (181428) ⋅109/L and 129 (62- 228) ⋅109/L respectively. Although the mean platelet count differed
markedly between patients groups, within all groups (i.e. trauma, vascular surgery, abdominal
surgery, liver transplantation and miscellaneous) non-survivors consistently had lower rise in
500
Platelet count
400
300
200
100
0
0
2
4
6
8
10
Days after ICU-admission
Figure 8.2b. Median platelet counts (109/L ; interquartile range) for survivors (open circles) and nonsurvivors (closed circles) in vascular surgery patients. The difference is significant (p<0.001).
platelet counts as measured by mean ∆PC/∆t (Fig. 8.2a-d). These
differences were highly
85
Blunted rise in platelet counts in SICU patients
500
Platelet count
400
300
200
100
0
0
2
4
6
8
10
Days after ICU-admission
Figure 8.2c. Median platelet counts (109/L ; interquartile range) for survivors (open circles) and
non-survivors (closed circles) in abdominal surgery patients. This difference is significant
(p<0.001).
significant (p<0.001) for all groups with the exception of the liver transplantation group. For
readmitted patients the time-dependent changes of the platelet count did not differ from patients
who were first admitted.
As opposed to the platelet count, overall and subgroup leukocyte counts in survivors were slightly
lower compared to non-survivors, with values of 13.4 (10.2- 18.0) ⋅109/L and 15.5 (11.421.2) ⋅109/L at day 10 respectively (p=0.03). Patients with low APACHE-II-scores (<18) showed a
somewhat higher platelet count than patients with high APACHE-II scores (>=18) with values of
308 (126-446) and 260 (153- 416) ⋅109/L at day 10 (p=0.03).
In 1290 admissions the daily change in platelet counts (∆PC/∆t) between day 2 and 10 could be
calculated: 30-day survivors had a value of 30±46 ⋅109/L/day whereas non-survivors had a value
of 6±28 ⋅109/L/day (p<0.001). With respect to 30-day survival the area under the ROC was
0.743 for ∆PC/∆t (Fig. 8.3) and 0.728 for APACHE-II. The areas under the ROC for ∆PC/∆t and
APACHE-II when calculated in respect to hospital survival where 0.736 and 0.708 respectively.
86
Chapter 8
DISCUSSION
The aim of this study was to analyze the time-response of the platelet count in relation to outcome
in critically ill patients. In patients who uneventfully recover from an inflammatory insult the
platelet count displays a bimodal response with an initial decrease below baseline values for the
first days, followed by an increase above the normal range after one week. In our study, the
surviving trauma patients clearly showed such a response (Fig. 8.2a). Trauma patients sustained
their injury and subsequent definitive surgery in most cases within 24 hours before ICU admission.
Combined with the fact that nearly all of these patients were previously healthy and relatively
young - as opposed to most patients in the other patient groups – this may explain why the bimodal
time course of platelet counts was most marked in trauma patients.
In the other patient groups lower overall platelet counts were observed, especially in liver
transplants (Fig. 8.2d). In this latter group, splenomegaly is a major determinant of low platelet
counts. Nevertheless, in all patients groups we saw the same diverging pattern between those who
eventually died and those who survived. Platelet counts in survivors were twice the value of nonsurvivors at day 10. When the rate of change ∆PC/∆t from day 2 onward was examined, this value
was 5-fold higher in the survivors compared to non-survivors. If the change in platelet counts is
considered from day 0 onward instead of from day 2 onward, the difference between survivors and
non-survivors would be even more even more striking. But the reason for calculating ∆PC/∆t from
day 2 onward, and not from day 0 onward, was to appropriately fit the bimodal time course of
platelet counts.
500
Platelet count
400
300
200
100
0
0
2
4
6
8
10
Days after ICU-admission
Figure 8.2d. Median platelet counts (109/L ; interquartile range) for survivors (open circles) and
non-survivors (closed circles) in liver transplantation patients.
The constancy and reproducibility of the plaletet count in normal individuals has already been
87
Blunted rise in platelet counts in SICU patients
established by Brecher [7] in 1953 and later by Ross [8] who showed a remarkable intraindividual stability of the platelet count over a 9-month period. Although for a normal population
the platelet count may vary by 200 ⋅109/L, from 150 to 350 ⋅109/L, in any normal individual the
range is only approximately 60 ⋅109/L. The use of ∆PC/∆t removes baseline inter-individual
differences. That ∆PC/∆t shows greater differences between survivors and non-survivors than the
platelet count itself, underscores that Brecher’s observation on normal individuals is very relevant in
patients as well.
The relation between a blunted rise in platelet counts and outcome is not restricted to surgical ICU
patients, as shown for 206 patients at the medical ICU [9]. In this study which excluded patients with
entities known to directly effect platelet counts (i.e. AIDS, leukemia, chemotherapy, immune
thrombocytopenia and hemolytic uremic syndrome) survivors also showed significantly higher
platelet counts from ICU-day 3 onward. In a recent study on the international EURICUS-II database
compiling data from 1636 patients from 61 centers with both medical and surgical ICU patients, we
also found that ∆PC/∆t discriminates survivors from non-survivors [10].
That the platelet count can be a good parameter to follow-up critically ill patients is also reflected
in the new emphasis on this parameter in recently introduced scoring systems for critically ill
patients. The sequential organ failure assessment (SOFA) score [11] uses the platelet count and
not the leukocyte count as one of its 6 components. Likewise, the multi-organ dysfunction score
[12] uses the platelet count and not the leukocyte count as one of 7 components. The older, and
still widely used APACHE-II uses the leukocyte count, but not the platelet count. It is obviously
not our intention to promote ∆PC/∆t as yet another prognostic scoring system, especially since a
single parameter can never be the foundation of a prognostic score for such complex patients as
there are at the ICU. But the fact that ∆PC/∆t is as good as APACHE-II in predicting mortality is
remarkable and makes it a candidate component of scoring systems. It should not be difficult to
verify our findings on the existing databases that are the foundation of the above mentioned
scores [11,12]. Since these scores already use platelet count as a component, they could be
reanalyzed with the inclusion of ∆PC/∆t.
88
Chapter 8
The blunted rise of the platelet counts in the non-survivors must be the result of consumption that is
relatively higher than production. Although it is difficult to discriminate between decreased platelet
synthesis and increased consumption, it is probable that especially increased consumption of
platelets is present. If bone marrow insufficiency or regulatory abnormalities are assumed to cause
decreased platelet synthesis, this phenomenon is not reflected by decreased leukocyte counts in
1.0
0.8
Sensitivity
30
0.6
0.4
0
0.2
0.0
0.0
0.2
0.4
0.6
0.8
1.0
1 - Specificity
Figure 8.3. Receiver-operating characteristic curve that describes the relation between of various cut-off
levels of the daily change in the platelet count (∆PC/∆t) and the sensitivity (true positive fraction) and 1specificity (false positive fraction) in predicting death within 30 days. The area under the ROC-curve is 0.743.
The straight diagonal indicates the ROC of a parameter that has no predictive value at all. The upper arrow
indicates that values of ∆PC/∆t ≤30 ⋅109/L/day have a sensitivity of 0.89 for predicting mortality, with a
specificity of 0.42. A cut-off of ∆PC/∆t ≤0 corresponds to a sensitivity of 0.38 and a specificity of 0.90.
our patient set, since non-survivors even had higher leukocyte counts than survivors.
Consumption of platelets can be caused by disseminated intravascular coagulation (DIC) and
sequestration in organs. DIC is frequently present in critically ill patients: it was found in 29 of 35
ICU-patients with systemic inflammatory response syndrome [13]. In the acute respiratory distress
syndrome, the trapping of platelets in the lung has long been known [14]. Sigurdson et al.
investigated platelet sequestration in ICU patients with 111Indium-labelled platelets at the bedside
[15]. Patients with poor outcomes showed trapping of platelets in the gut as well as in the lung, liver
and spleen. This phenomenon was observed 1 to 4 days before clinical sepsis or multiorgan failure.
After liver transplantation, about 50% of the circulating platelets are sequestrated in the transplanted
liver immediately after reperfusion [16]. It has been shown that platelet sequestration in the liver
89
Blunted rise in platelet counts in SICU patients
graft is associated with increased reperfusion damage. Persistent thrombocytopenia after
transplantation was found to be associated with decreased survival [17], which is in agreement with
the observations in the current study.
Naturally the question arises if interventions aimed at preventing platelet consumption would
affect outcome. Again, reanalysis of existing data sets of intervention studies with regard to
∆PC/∆t would be of great interest. Such analyses could answer: a) if ∆PC/∆t is related to outcome
and b) if the intervention affected ∆PC/∆t. Thus, in patients that are not critically ill increased
platelet counts often indicate ongoing inflammation and disease. But in critically ill patients the
opposite is the case since a blunted rise in platelet counts - i.e. a low ∆PC/∆t - has an unfavorable
prognosis. The presence of a ‘normal’ platelet count
after ICU-admission should not be
automatically interpreted as desirable. Therefore the platelet count, routinely determined at low
cost, should not only be used to detect thrombocytopenia or thrombocytosis. It should also be
used to actively follow its time-dependent changes.
REFERENCES
1.
Lotem J, Shabo Y, Sachs L. Regulation of megakaryocyte development by interleukin-6. Blood
1989;74:1545-1551.
2.
Nijsten MWN, Hack CE, Helle M, et al. Interleukin-6 and its relation to the humoral immune
response and clinical parameters in burned patients. Surgery 1991;109:761-767.
3.
Bonfiglio MF, Traeger SM, Kier KL, et al. Thrombocytopenia in intensive care patients: a
comprehensive analysis of risk factors in 314 patients. Ann Pharmacother 1995;29:835-842.
4.
Bradbury AW, Bachoo P, Milne AA, et al. Platelet count and the outcome of operation for ruptured
abdominal aortic aneurysm. J Vasc Surg 1995;21:484-491.
5.
Knaus WA, Draper EA, Wanger DP, et al. APACHE-II: a severity of disease classification system.
Crit Care Med 1985;13:818-829.
6.
Zweig MH, Campbell G. Receiver-operating characteristic (ROC) plots: a fundamental evaluation
tool in clinical medicine. Clin Chem 1993;39:561-577.
7.
Brecher GB, Schneiderman M, Cronkite EP. The reproducibility and constancy of the platelet count.
Am J Clin Pathol 1953;23:15-26.
8.
Ross DW, Ayscue LH, Watson J, et al. Stability of hematologic parameters in healthy subjects.
Intraindividual versus interindividual variation. Am J Clin Pathol 1988;90:262-267.
9.
Nijsten MWN, Tulleken JE, vd Werf TS, et al. Absence of secondary thrombocytosis is associated
with prolonged ICU-stay and decreased survival [abstract]. Intensive Care Med 1997;23 Suppl
1:59.
10.
Nijsten MWN, Nap R, Reis Miranda D. Rate of change in platelet count is associated with
mortality in EURICUS-II dataset [abstract]. Intensive Care Med 1999;25 Suppl 1:26.
11.
Vincent JL, de Mendonca A, Cantraine F, et al. Use of the SOFA score to assess the incidence of
organ dysfunction/failure in intensive care units: results of a multicenter prospective study. Crit
Care Med 1998; 26:1793-1800.
12.
Marshall JC, Cook DJ, Christou NV, et al. Multiple organ dysfunction score: a reliable descriptor of
a complex clinical outcome. Crit Care Med 1995;23:1638-1652.
90
Chapter 8
13.
Gando S, Kameue T, Nanzaki S, et al. Disseminated intravascular coagulation is a frequent
complication of systemic inflammatory response syndrome. Thromb Haemost 1996;75:224-228.
14.
Heffner JE, Sahn SA, Repine JE. The role of platelets in the adult respiratory distress syndrome.
Culprits or bystanders? Am Rev Respir Dis 1987;135:482-492.
15.
Sigurdsson GH, Christenson JT, el-Rakshy MB, et al. Intestinal platelet rapping after traumatic
and septic shock. An early sign of sepsis and multiorgan failure in critically ill patients? Crit Care
Med 1992;20:458-467.
16.
Porte RJ, Blauw E, Knot EAR, et al. Role of the donor liver in the origin of platelet disorders and
hyperfibrinolysis in liver transplantation. J Hepatol 1994;21:592-600.
17.
McCaughan GW, Herkes R, Powers B, et al. Thrombocytopenia post liver transplantation.
Correlations with pre-operative platelet count, blood transfusion requirements, allograft function
and outcome. J Hepatol 1992;16:16-22.
91
Blunted rise in platelet counts in SICU patients
92
CHAPTER 9
PRIMARY AND SECONDARY CHANGES IN PLATELET COUNT
AND OUTCOME IN
MEDICAL AND SURGICAL ICU PATIENTS
M.W.N. Nijsten, R.E. Nap , H.J. ten Duis, R.J. Porte, D. Reis Miranda
Submitted
Primary and secondary changes in platelet count and outcome in ICU patients
ABSTRACT
Objective: Thrombocytopenia is related to adverse outcome in critically ill patients. In many
patients admitted to the intensive care (ICU), the platelet count (PC) shows an initial decrease
followed by a rebound of varying amplitude. Opposed to primary changes, the relation of
secondary changes of PC with outcome has not been systematically studied. In this study we have
analyzed changes in a large, heterogeneous set of ICU patients. Based on the changes of PC with
time we established previously, PC-changes were evaluated with a simple model that incorporates
the biphasic changes in PC.
Design: Analysis of the prospectively collected EURICUS-II database.
Setting: 53 ICUs in 11 European countries.
Patients: All patients for whom PC were available.
Measurements and results: 5206 Patients were classified as medical (55%), unscheduled surgery
(24%) or scheduled surgery (21%). During ICU-stay daily PC were recorded. The rate of change
of PC between day 0 and 2 was expressed as ∆PC/∆t 0→2, PC on day 2 as PC2 and the rate of
change in PC between day 2 and 10 as ∆PC/∆t 2→10. In non-survivors ∆PC/∆t2→10 was only 1±26
versus 13±31 ·109/l/d for survivors (p<0.001) Differences in early PC as reflected by PC2
(167±112 vs. 188±102 ·109/l; p<0.001) and ∆PC/∆t0→2 (-16±54 vs. -13±41 ·109/l/d; NS) were less
pronounced.
Conclusions:
A low increase or even a decrease in PC after day 2 is associated with poor outcome both in
medical and in surgical ICU patients. In individual patients, clinicians should bear in mind the
typical time course of the PC after ICU admission in order to recognize abnormal patterns. In
patient groups, taking ∆PC/∆t2→10 into account may aid in improving outcome assessment.
INTRODUCTION
Both production and consumption of platelets are increased in diseases that involve systemic
inflammation. Initial changes in PC in patients admitted to the ICU mostly concern decreases in
PC. The magnitude of thrombocytopenia after ICU-admission has proven to be of important
negative prognostic value in trauma [1], ruptured aortic aneurysm [2], malaria [3], meningitis [4],
and sepsis patients [5], as well as in patients in general that were admitted to the ICU [6,7,8,9].
The studies that investigated PC and outcome all focused on changes in PC to subnormal values,
and mostly early after ICU-admission.
Whereas systemic inflammation is initially accompanied by a drop in PC, in uncomplicated
patients a rebound increase in PC is observed. For example in a trauma patient who does not need
prolonged ICU-support, the PC will usually start to rise 2 to 3 days after the injury and reach
94
Chapter 9
supranormal levels after two weeks [10]. Increased production of platelets induced by interleukin6, thrombopoietin and other cytokines [11] is responsible for this secondary thrombocytosis.
We have noted that in critically ill patients who have an ongoing inflammatory response this
rebound thrombocytosis is absent in many cases. This observation led us to the hypothesis that
more severe inflammation and thus adverse outcome, may be related to an absent or blunted
thrombocytotic response. In a recent study in patients admitted to a surgical ICU we found that
non-survivors have a lower rate of change in PC after day 2 [10].
In the present study we attempted to systematically analyze both medical and surgical patients,
early and late changes in PC, and their relation with outcome. A large multicenter database that
had been prospectively collected, was used for this purpose. We hypothesized that the
information provided in the secondary changes of the platelet count would be distinct from the
information carried by initial changes in the platelet count. A simple mathematical model based
on the time course of PC as identified earlier [10] was applied. This model (Fig. 9.1) assumes that
PC decreases linearly during the first two days after ICU-admission, reaches a nadir value after
two days, and changes linearly between day 2 and day 10.
500
PC
0
0
2
10
ICU day
Figure 9.1. Three-parameter model for time dependent changes in platelet count (PC) that was used in this
study. For each patient, the rate of change in the platelet count, ∆PC/∆t was calculated by linear regression
between day 0 and day 2 (∆PC/∆t0→2) and between day 2 and day 10 (∆PC/∆t2→10). Nadir PC was assumed to
occur at day 2 (PC2). In this example ∆PC/∆t0→2 equals the slope of the descending line on the left (-110
·109/l/d), PC2 is the PC at day 2 (135 ·109/l) and ∆PC/∆t2→10 equals the slope of the ascending line (+34
·109/l/d).
95
Primary and secondary changes in platelet count and outcome in ICU patients
PATIENTS AND METHODS
EURICUS-II Database
The EURICUS-II dataset contains data of 53 ICUs (medical and surgical) in 11 European
countries [12]. Patient data were recorded from admission to the ICU until discharge from the
Table 9.1
Patient characteristics
N
Age
Years
SAPS-II
∆PC/∆t 0→2
PC2
∆PC/∆t 2→10
Survivors
Non-survivors
P
4034
1172
59±21
66±18
<0.001
31±15
49±20
<0.001
9
-13± 41
-16±54
0.086
9
188±102
168±112
<0.001
9
13±31
1±26
<0.001
10 /l/d
10 /l
10 /l/d
Values ± SD
ICU. On admission patients were divided into one of three categories: unscheduled surgery,
scheduled surgery or medical as has been described before [13]. In addition, age, the simplified
acute physiology score (SAPS-II; [13]) on admission, daily PC (normal range 150-350·109/l)
between day 0 (admission to the ICU) and day 10, were collected. Patients were classified into
non-survivors and survivors on the basis of hospital mortality.
Model of time dependent changes in PC
The platelet count was assumed to change linearly until day 2 (mostly a decrease), and to change
linearly (increase or decrease) afterwards. For each individual patient three descriptors of serial
changes in platelet count were extracted with a special program that calculated individual
regression coefficients (Fig. 9.1):
∆PC/∆t0→2
PC2
∆PC/∆t 2→10
109/l/d
109/l
109/l/d
the early rate of change of the PC, between day 0 and day 2
platelet count on day 2
the late rate of change of the PC, between day 2 and 10
Thus for each patient the course of the PC was approximated by two linear regression
coefficients reflecting the rates of change (∆PC/∆t0→2 and ∆PC/∆t2→10) and the PC on day 2 (PC2)
which was assumed to approximate nadir PC at any time. To verify the assumption that PC2
estimates the nadir PC, PC2 was compared with nadir PC.
96
Chapter 9
Statistical analysis.
Means of all (derived) parameters were compared between non-survivors and survivors for the
overall patient group, as well as for the three admission groups. For the various analyses which
were performed, our approach was to use the largest available subsets of patients possible for each
specific analysis. Thus the total number of patients used varies in the different analyses. This
variation is mainly dependent upon length of stay, since the EURICUS-II study only recorded
data during ICU-stay. Unless indicated otherwise standard deviations are used. The two-sided
Student's t-test was used for comparison of continuous parameters, without assuming equal
variances. The Bonferroni-correction was used in case of multiple testing.
RESULTS
Overall patient characteristics
20508 PC values collected between day 0 and day 10 were available for 5206 patient admissions.
Of these admissions, 55% were medical, 24% unscheduled surgery and 21% scheduled surgery.
Mean ICU-length of stay was 5.9±10.5 days with a median (interquartile range) value of 2 (1-6)
days. The overall mortality rate was 23% (Table 9.1), with group mortalities of 36% for medical,
35% for unscheduled surgery and 11% for scheduled surgery respectively.
Platelet Counts
On admission, mean PC in the entire group was 217±115·109/l. At this time only 4% had a
PC<50·109. Fig. 9.2 shows the diverging time course of the PC between non-survivors and
survivors. On day 0 the non-surviving patients had a mean PC of 203±122 that dropped to
167±112 on day 2 and rose to 222±137·109/l on day 10. In the survivors this value at day 10 was
317±166, which is 95·109/l higher than in non-survivors (p<0.001). This pattern of an initial
decrease with a subsequent increase was also observed in the three major subgroups: medical,
unscheduled surgery and scheduled surgery, with mean PC reaching a minimal value on day 2. In
support of the model’s assumption, the nadir PC showed a strong correlation with the PC2, with a
Spearman correlation coefficient of 0.89 (p<0.001).
In non-survivors ∆PC/∆t 2→10 was only 1±26 compared to 13±31·109/l/d for survivors (p<0.001).
The difference in PC2 (167±112 vs. 188±102 ·109/l; p<0.001) was less pronounced. ∆PC/∆t0→2
showed no difference (-16±54 vs. -13±41 ·109/l/d; NS).
97
Primary and secondary changes in platelet count and outcome in ICU patients
500
1.00
400
0.80
300
0.60
200
0.40
100
0.20
0
Fraction of Patients
Platelet count
A value of ∆PC/∆t2→10 below -30·109/l/d was associated with a 40% mortality, whereas a rise of
more than +60·109/l/d was associated with a 3% mortality. Figure 9.3 shows the distributions and
associated mortalities of the three parameters ∆PC/∆t0→2, PC2 and ∆PC/∆t2→10 for the three
admission groups. Fig. 9.3 illustrates that in the majority of patients PC indeed decreases between
day 0 and day 2 in most patients, but drops to levels still over 150·109/l in most patients. Only few
patients have severe thrombocytopenia on day 2 and after day 2 PC rises in most patients. Fig 9.3
Figure 2
0.00
0
2
4
6
8
10
ICU day
Figure 9.2. Mean PC +/- SD in non-survivors (ICU and hospital mortality; filled circles) and survivors
(circles). The mean PC values are significantly different between survivors and non-nonsurvivors. The thick
descending line indicates the fraction of patients in which PC was measured; the 100% at day 0 corresponds to
4862 patients.
also indicates that ∆PC/∆t2→10 has the clearest association with mortality.
Since PC2 and ∆PC/∆t2→10 had the strongest relation with mortality, the interaction of these two
parameters (Table 9.2) and their combined relation with mortality (Fig. 9.4) were analyzed. Here
subranges were chosen to approximate interquartile ranges for the two respective parameters. The
ordinal by ordinal Spearman correlation coefficient was -0.16, revealing that PC2 and ∆PC/∆t2→10
are nearly independent. Fig. 9.4 shows that the relations of low PC2 and a low ∆PC/∆t2→10 with
mortality are additive. For example the 108 patients with PC2<100·109/l and ∆PC/∆t2→10 < 0 have
60% mortality rate in contrast to an 8% mortality rate in the 113 patients with PC2>200·109/l and
∆PC/∆t2→10 > 30·109/l/d.
Children
In the small subgroup of children (age<=15 year; 13 nonsurvivors; 39 survivors) only ∆PC/∆t2→10
showed a difference with -9±40 versus 24±40 ·109/l/d (p=0.016) in non-survivors and survivors
respectively.
98
Chapter 9
Late mortality and early changes in PC
Since changes in PC only shortly before death might disproportionately affect PC2 and
∆PC/∆t2→10, we performed an additional analysis on the predictive value of PC2 and ∆PC/∆t for a
longer period ahead. In the subgroup of patients that had a stay of at least 10 days at the ICU, PC2
and the rate of change between day 2 and day 5 (i.e. ∆PC/∆t2→5) were compared. PC2 was
184±124 and 186±100·109/l respectively (NS) and ∆PC/∆t2→5 was 0±30 and 10±32·109/l/d
respectively (p<0.001).
Table 9.2
Number of patients (percentage) according to PC2 and ∆PC/∆t 2→10
∆PC/∆t 2→10 (109/l/d)
PC2 (109/l)
<0
0 ~ 10
10 ~ 30
>30
Totals
<=100
106 (5)
100 (5)
134 (7)
98 (5)
438 (23)
100 ~ 150
107 (6)
99 (5)
127 (7)
87 (5)
420 (22)
151 ~ 200
136 (7)
91 (5)
117 (6)
38 (2)
382 (20)
> 200
299 (16)
123 (6)
153 (8)
113 (6)
688 (36)
Totals
648 (34)
413(21)
531(28)
336 (17)
1928 (100)
DISCUSSION
The relationship between PC and mortality in critically ill patients was analyzed in this study. In a
large group of medical and surgical ICU patients the prognostic importance of PC extends well
beyond initial changes. With regard to PC, day 2 after ICU admission can be viewed as a turning
point (Fig 9.1). The extent of the decrease in PC during the first two days was not related with
mortality. Many investigators observed a relation with early nadir PC and mortality. In this study
we also found that the PC at day 2, which adequately reflected nadir PC, is related with mortality.
But the daily rate of change in PC after day 2 showed the strongest relation with mortality. This
phenomenon was especially marked in unscheduled surgery patients (Fig. 9.3), where these
mortalities were 53% and 0% for low (i.e. <0) and high (>30·109/l/d) values of ∆PC/∆t 2→10
respectively.
Vanderschueren [6] reported that in 329 predominantly medical ICU patients that nadir PC was
associated with mortality. Unfortunately, the authors do no report the time interval between
measuring a nadir PC and the patients' death. Stephan observed in 147 surgical ICU patients [7]
that those patients who had a PC below 100 at any time had a mortality rate of 38%, compared
99
Primary and secondary changes in platelet count and outcome in ICU patients
with a 20% mortality rate for PC >100·109/l. In the latter study thrombocytopenia occurred
1.8±0.5 days after admission.
In a previous study of 1415 admissions to a surgical ICU [10] we also observed an association
with mortality of a low rate of change in PC after day 2. The current study generalizes these
results to a larger, more heterogeneous patient group that includes medical patients as well as
children. That changes in PC between day 2 and day 5 were also associated with the mortality that
occurred after day 10, underscores that the phenomenon we observed is not solely the result of
changes occurring directly before death.
Limitations of this study
The primary purpose of EURICUS-II was to investigate the role of collaborative practice at the
ICU, and to find parameters to measure improvements in nurse-doctor interaction [12]. Thus we
retrospectively analyzed data that were prospectively collected for other purposes. The database
contains very heterogeneous patients without detailed information on their condition before ICUadmission, making analysis of PC behavior in specific syndromes or diseases impossible.
Especially in the medical patients, PC before ICU-admission would have been useful. Preadmission PC would obviously allow a better estimation of ∆PC since in many cases PC will have
decreased already before ICU-admission. Many relevant physiological parameters were not
recorded, such as leukocyte counts or quantitative indicators of systemic inflammation (e.g. Creactive protein). We did not study the mechanisms that lead to decreased PC. Nevertheless, as
pointed out below, we think other studies provide compelling arguments to assume that
inflammation-induced platelet sequestration plays a major role.
Pathophysiology of decreasing platelet counts
Decreases in PC are by definition the result of platelet consumption that is higher than platelet
production. We assume that increased platelet consumption is the main contributor to (relatively)
low PC in ICU patients both during the primary and secondary phases of ICU-stay. Acute PCdecreases must result from increased consumption of platelets since platelets have a normal lifespan of 8 to 11 days [14]. The fact that only 4% had a PC below 50·109/l excludes a significant
impact of patients with acquired bone marrow failure on our results, since such patients usually
have a PC well below 50·109/l. In a study on thrombocytopenia in the surgical ICU 8 out of the 9
patients with sepsis and thrombocytopenia who underwent bone marrow examination during later
phases of their ICU-stay, displayed a normocellular or hypercellular marrow with a normal
number of megakaryocytes [7]. The same observation was made in 15 children with
thrombocytopenia after sepsis without signs of disseminated intravascular coagulation (DIC) [15].
In our previous study in surgical ICU patients [10], non-survivors had higher leukocyte counts
than survivors, which also argues against marrow failure. The most direct observation of
increased platelet sequestration in trauma and sepsis patients in the ICU was seen after
administration of 111Indium-labelled platelets. Increased sequestration was found in several organs
including the lung, liver and gut, especially in patients with adverse outcomes [16].
100
Chapter 9
∆PC/∆t 0→2
∆PC/∆t 2→10
PC2
Number of patients
2000
1500
1000
500
0
<-30
-30~0
0~30
<-30
-30~0
0~30
30~60
>60
0~20
21~50
51~100 101~150
>150
<-30
-30~0
0~30
30~60
>60
0~20
21~50
51~100 101~150
>150
<-30
-30~0
0~30
30~60
>60
100%
Mortality
80%
60%
40%
20%
0%
30~60
>60
Scheduled Surgery
Unscheduled surgery
Medical
Figure 9.3. Distribution of patients (upper panels) and mortality (lower panels) according to ∆PC/∆t0→2
(left panels; 3843 patients) , PC2 (Middle panels ; 2645 patients) and ∆PC/∆t2→10 (Right panels; 2049
patients). Bar shades denote admission categories: scheduled surgery: gray bars; unscheduled surgery:
black bars and medical: white bars.
The upper left panel shows that ∆PC/∆t0→2 is negative in most patients, indicating an initial decrease in
PC. The upper middle panel shows that a day 2, most patients still have a PC>150 ·109/l, and only 35
patients (1%) have a PC< 20 ·109/l. The right upper panel indicates that after day 2, most patients display
have an increase in PC.
The lower middle and lower right panels indicate that decreased values of PC2 and ∆PC/∆t2→10 are
associated with increased mortality. In the unscheduled surgery group with ∆PC/∆t2→10 <-30·109/l/d 19
patients had 53% mortality; while 37 patients with . ∆PC/∆t2→10 > 60·109/l/d had zero mortality.
40%
Mortality
60%
20%
<0
10 ~ 30
0%
<100
100-150
150-200
>200
Figure 9.4. Analysis of combined relations of PC2 and ∆PC/∆t2→10 with mortality. Subranges for PC2 and
∆PC/∆t2→10 are different from Fig. 3., as they were chosen to best correspond to interquartile ranges (see
Table 9.2 for distribution of patient numbers). Mortality is displayed on the vertical axis; the horizontal
across axis shows four PC2-categories; the depth axis shows four ∆PC/∆t2→10-categories. This twodimensional frequency distribution illustrates that ∆PC/∆t2→10 has a stronger association with mortality
than PC2.
101
Primary and secondary changes in platelet count and outcome in ICU patients
The term platelet sequestration is broad enough to include the distinct process of adhesion to
endothelium and aggregation in clots. Apart from the effect of systemic inflammation that is
present in the majority of ICU-patients [17], blood loss [18] and intravascular coagulation [19]
are frequent causes of acute decreases PC. Although sometimes acting in concert [20], these
processes are separate and in most sepsis patients with thrombocytopenia significant diffuse
intravascular coagulation (DIC) is not present [21]. Endothelial cell activation as part of systemic
inflammation can trigger platelet adhesion by itself through the expression of a variety of
adhesion molecules [22]. Platelet sequestration is also observed without DIC in several important
models of systemic inflammation. In experimental models of TNF-administration [23] or malaria
[24] mortality was related to the extent of thrombocytopenia, and not DIC.
Experimentally and clinically it is difficult to measure the contribution of endothelial activation to
platelet consumption. It is also not known if sequestration of platelets is directly causal to
mortality, but the inflammatory potential of adherent or aggregating platelets is well known. In
the adult respiratory distress syndrome [25] or ischemia-reperfusion syndromes [26] such as
observed after liver transplantation [27] platelets are powerful and important mediators of tissue
injury.
The pervasiveness of platelet sequestration in the critically ill hints that platelet sequestration may
be a pathogenic process many organs.
Theoretical advantages of ∆PC/∆t2→10
One reason that ∆PC/∆t2→10 may discriminate better than absolute PC is that inter-individual
variation is reduced. Intra-individual variation in PC amounts is only 30% of the inter-individual
variation in healthy persons [28,29].
Major methodological criteria which a physiologic measure used in a scoring system should fulfill
are reproducibility, responsiveness and validity [30].
Responsiveness means that a measure detects clinically meaningful changes in the process of
interest, and that changes in the measure correspond to clinically significant changes. With regard
to this criterion it is contradictory to transform a responsive, continuous parameter into a
discontinuous parameter with only a limited number of values. Moreover scoring systems such as
SOFA (sequential organ failure score; [31]) or MODS (multiple organ dysfunction score; [30])
only start to score the PC as abnormal if it is subnormal. For example a PC>150 has 0 SOFA
points, and will still have 0 points when the PC drops 250 on day 2 to 160 on day 10, a drop that
we find is associated with increased mortality. Although the practical advantage is obvious,
namely calculating a total score without a computer, important information is lost in this process.
Yet, the SOFA-authors [31], like the MODS-authors [30] have subsequently observed that
changes in their organ scores (i.e. ∆SOFA or ∆MODS) are also powerful predictors of outcome in
addition to the initial scores themselves. The added value of ∆SOFA or ∆MODS is intuitively
logical since these incremental increases in organ dysfunction portend by definition additional,
more recent information. The special relevance of these incremental scores has been interpreted as
reflecting de novo events that arise at the ICU, and may thus be amenable to therapeutic
intervention. We believe that the same phenomenon is apparently true for ∆PC/∆t2→10.
102
Chapter 9
Validity is the extent to which a parameter is a meaningful representation of the entire spectrum of
the process of interest. Although the authors who incorporated PC into their scoring systems
considered PC a useful and valid parameter, it is in fact unclear for which pathophysiological
process it is valid. The SOFA authors call PC a "coagulation" parameter while MODS authors
call it an "hematological" parameter. Coagulation and hematology are two rather broad and
different processes. In fact it may turn out that PC and especially ∆PC/∆t reflect endothelial
activation in many instances, as argued above. Thus we think that although PC and ∆PC/∆t are
useful parameters, the process they reflected is not established.
Although it is obviously futile for application as a solitary parameter, we think ∆PC/∆t2→10 could
be considered for inclusion in future scoring systems.
Clinical and research implications
In clinical practice, PC2 and ∆PC/∆t2→10 are not difficult to assess, even without making a graph
or performing calculations. Simple observation of the PC over a number of days will easily show
whether it is decreasing, increasing or does not change after day 2. Our results illustrate that
stable platelet counts, even in the so-called normal range (150-350 ·109/l) are not ideal in many
circumstances. In a patient who is recovering from severe injury the PC should be high-normal or
supranormal after 1 week [10]. If the PC fails to rise in the first week, even if it is 200·109/l, this
should alert the clinician to potential complications.
If platelet sequestration is a pathogenic process and ∆PC/∆t2→10 reflects platelet sequestration, this
raises the logical question if inhibition of platelet sequestration will improve outcome. Analysis of
intervention trial data with respect to ∆PC/∆t2→10 may be useful for addressing this question. Like
platelets, the agents protein C [32], antithrombin-III [33], glycoprotein IIb/IIIa (GPIIb/IIIa; [34])
and P-selectin [35] are all involved at intersection of the coagulatory and inflammatory processes.
Thus these agents (i.e. activated protein C, antithrombin-III) or their inhibitors (i.e. GPIIb/IIIainhibitors or P-selectin inhibitors) are prime candidates for an examination of their impact on
∆PC/∆t2→10.
In conclusion, serial platelet counts carry important information in critically ill patients. Taking
the biphasic changes of PC into account helps to assess what is normal and what is abnormal.
Scoring systems might be improved by accommodating the information contained in changes in
platelet counts. Although some may take the process of platelet sequestration for granted, it is
obvious that a fundamental process must be involved as it occurs in the large variety of critically
ill patients. Intervention studies may show whether platelet sequestration is a process that is
causal to mortality or whether it is only another of the many markers that have been associated
with poor outcome in the ICU.
ACKNOWLEDGEMENT
This study was supported by a grant from the European Commission (BMH4-CT96-0817).
103
Primary and secondary changes in platelet count and outcome in ICU patients
REFERENCES
1.
Sauaia A, Moore FA, Moore EE, Norris JM, Lezotte DC, Hamman RF. Multiple organ failure can
be predicted as early as 12 hours after injury. J Trauma 1998;45:291-301.
2.
Bradbury AW, Bachoo P, Milne AA, Duncan JL. Platelet count and the outcome of operation for
ruptured abdominal aortic aneurysm. J Vasc Surg 1995;21:484-491.
3.
Horstmann RD, Dietrich M. Haemostatic alterations in malaria correlate to parasitaemia. Blut
1985;51:329-335.
4.
van Deuren M, Neeleman C, van 't Hek LGFM, van der Meer JWM. A normal platelet count at
admission in acute meningococcal disease does not exclude a fulminant course. Intensive Care Med
1998;24:157-161.
5.
Peduzzi P, Shatney C, Sheagren J, Sprung C. Predictors of bacteremia and gram-negative
bacteremia in patients with sepsis. The Veterans Affairs Systemic Sepsis Cooperative Study
Group. Arch Intern Med 1992;152:529-535.
6.
Vanderschueren S, de Weerdt A, Malbrain M, Vankersschaever D, Frans E, Wilmer A, Bobbaers
H.Thrombocytopenia and the prognosis in intensive care. Crit Care Med 2000; 28:1871-1876.
7.
Stephan F, Hollande J, Richard O, Cheffi A, Maier-Redelsperger M, Flahault A.
Thrombocytopenia in a surgical ICU. Chest 1999;115:1363-1370.
8.
Baughman RP, Lower EE, Flessa HC, Tollerud DJ. Thrombocytopenia in the intensive care
unit.Chest 1993;104:1243-1247.
9.
Lee KH, Hui KP, Tan WC. Thrombocytopenia in sepsis: a predictor of mortality in the intensive
care unit. Singapore Med J 1993;34:245-246.
10.
Nijsten MWN, ten Duis HJ, Zijlstra JG, Porte RJ, Zwaveling JH, Paling JC, The TH. Blunted rise in
platelet count in critically ill patients is associated with worse outcome. Crit Care Med 2000;28:38433846.
11.
Uppenkamp M, Makarova E, Petrasch S, Brittinger G. Thrombopoietin serum concentration in
patients with reactive and myeloproliferative thrombocytosis. Ann Hematol 1998;77:217-223.
12.
The effect of harmonising and standardising the nursing tasks on the Intensive Care Units of the
European Community – EURICUS II. Biomedical and health research programme 1994-1998.
Summaries for research projects supported under BIOMED 2, volume II;494-495.
13.
Le Gall Jr, Lemeshow S, Saulnier F . A new Simplified Acute Physiology Score (SAPS-II) based on a
European/ North American multicenter study. JAMA 1993;270:2957-2963.
14.
Heyns AD, Badenhorst PN, Lotter MG, Pieters H, Wessels P . Kinetics and mobilization from the
spleen of indium-111-labeled platelets during platelet apheresis. Transfusion 1985;25:215-218
15.
Mitterstieler G, Kurz R, Haas H, Desoye G. Thrombopoiesis and blood coagulation in pediatric
patients with septicemia. Padiatr Padol 1981;16:257-265.
16.
Sigurdsson GH, Christenson JT, el-Rakshy MB, Sadek S. Intestinal platelet trapping after traumatic
and septic shock. An early sign of sepsis and multiorgan failure in critically ill patients? Crit Care
Med 1992;20:458-467.
17.
Pittet D, Rangel-Frausto S, Li N, Tarara D, Costigan M, Rempe L, Jebson P, Wenzel RP .
Systemic inflammatory response syndrome, sepsis, severe sepsis and septic shock: incidence,
morbidities and outcomes in surgical ICU patients. Intensive Care Med 1995;21:302-329.
104
Chapter 9
18.
Harvey MP, Greenfield TP, Sugrue ME, Rosenfeld D . Massive blood transfusion in a tertiary
referral hospital. Clinical outcomes and haemostatic complications. Med J Aust 1995;163:356-359.
19.
Levi M, ten Cate H . Disseminated Intravascular coagulation. N Eng J Med 1999;341: 586-592.
20.
Gando S, Kameue T, Nanzaki S, Nakanishi Y. Disseminated intravascular coagulation is a frequent
complication of systemic inflammatory response syndrome. Thromb Haemost 1996;75:224-228.
21.
Neame PB, Kelton JG, Walker. Stewart IO, Nossel HL, Hirsch J. Thrombocytopenia in septicemia:
the role of intravascular coagulation. Blood 1980;56:88-92.
22.
Hunt BJ, Jurd KM. Endothelial cell activation. A central pathophysiological process.
BMJ
1998;316:1328-1329.
23.
Senaldi G, Piguet PF. Mortality and platelet depletion occur independently of fibrinogen
consumption in murine models of tumor necrosis factor-mediated systemic inflammatory
responses. Cytokine 1998;10:382-389.
24.
Piguet PF, Da Laperrousaz C, Vesin C, Tacchini-Cottier F, Senaldi G, Grau GE. Delayed mortality
and attenuated thrombocytopenia associated with severe malaria in urokinase- and urokinase
receptor-deficient mice. Infect Immun 2000;68:3822-3829.
25.
Heffner JE, Sahn SA, Repine JE. The role of platelets in the adult respiratory distress syndrome.
Culprits or bystanders? Am Rev Respir Dis 1987;135:482-492.
26.
Massberg S, Enders G, Leiderer R, Eisenmenger S, Vestweber D, Krombach F, Messmer K.
Platelet-endothelial cell interactions during ischemia/reperfusion: the role of P-selectin. Blood
1998;92:507-515.
27.
Sindram D, Porte RJ, Hoffman MR, Bentley RC, Clavien PA. Platelets induce sinusoidal
endothelial cell apoptosis upon reperfusion of the cold ischemic rat liver. Gastroenterology
2000;118:183-191.
28.
Buckley MF, James JW, Brown DE, Whyte GS, Dean MG, Chesterman CN, Donald JA. A novel
approach to the assessment of variations in the human platelet count. Thromb Haemost
2000;83:480-484.
29.
Ross DW, Ayscue LH, Watson J, et al: Stability of hematologic parameters in healthy subjects.
Intraindividual versus interindividual variation. Am J Clin Pathol 1988;90:262-267.
30.
Marshall JC, Cook DJ, Christou NV, Bernard GR, Sprung CL, Sibbald WJ. Multiple organ
dysfunction score: a reliable descriptor of a complex clinical outcome. Crit Care Med
1995;23:1638-1652.
31.
Vincent JL, de Mendonca A, Cantraine F, et al. Use of the SOFA score to assess the incidence of
organ dysfunction/failure in intensive care units: results of a multicenter prospective study. Crit Care
Med 1998; 26:1793-1800.
32.
Esmon CT, Fukudome K, Mather T, Bode W, Regan LM, Stearns-Kurosawa DJ, Kurosawa S.
Inflammation, sepsis, and coagulation. Haematologica 1999;84:254-259.
33.
Fourrier F, Jourdain M, Tournoys A. Clinical trial results with antithrombin III in sepsis Crit Care
Med 2000;28(9 Suppl):S38-43
34.
Cannon CP. Incorporating platelet glycoprotein IIb/IIIa inhibition in critical pathways: unstable
angina/non-ST-segment elevation myocardial infarction. Clin Cardiol 1999;22(Suppl):IV30-36
35.
Frenette PS, Moyna C, Hartwell DW, Lowe JB, Hynes RO, Wagner DD. Platelet-endothelial
interactions in inflamed mesenteric venules. Blood 1998;91:1318-1324.
105
Primary and secondary changes in platelet count and outcome in ICU patients
106
CHAPTER 10
SUMMARY, DISCUSSION AND LINES OF FURTHER
INVESTIGATION
Summary, discussion and lines of further investigation
SUMMARY
Physiological parameters and circulating markers in humans who display systemic
inflammation were the main focus of this thesis. The patients investigated in this thesis varied
from trauma patients with or without fat embolism (FES), patients with burns, patients
receiving cytokine treatment to critically ill patients.
Fat embolism syndrome
In Chapter 2 we hypothesized that the nature of a femoral fracture, timing of the operation
and early inflammatory responses would affect the risk for subsequent development of FES.
We studied the incidence of FES in patients with an isolated fracture of the femoral shaft
admitted to our hospital in the period 1968-1985. In a detailed analysis of the factors
associated with the occurence of FES an elevated temperature was the only factor that
discriminated FES patients from other patients on day 0, suggesting that early acute phase
responses are intimately connected with the onset of FES.
Chapter 3 addressed the hypothesis that an open foramen ovale is necessary for large fat
globules gain entrance the systemic arterial circulation. We demonstrated that the presence of
a patent foramen ovale with a right-to-left shunt is not relevant in FES, thus indicating that the
fat globules are able to reach the systemic circulation through the lungs.
Interleukin-6 and acute phase responses
In chapter 4, it was hypothesized that IL-6 is an endogenous pyrogen as well as an inducer of
acute phase responses. Thanks to the availability at that time of the sensitive and specific B9.9
assay for IL-6, we were the first to show that circulating IL-6 is increased after inflammation
in humans. IL-6 was already sharply increased when burns patients presented at the hospital.
IL-6 was correlated with fever, and the IL-6 peak preceded increases in acute phase proteins.
In chapter 5, all major subsequent parts of the acute phase response were looked for and
actually observed in patients with severe burns: fever, tachycardia, leukocytosis, left shift in
the leukocyte differentiation, increased CRP and increased α1-proteinase inhibitor. As late
parts of the inflammatory response thrombocytosis, a distinct and short-lasted IgM-peak and
finally increased IgG levels were observed. There was no relation between the extent of the
burns and IL-6 levels, but patients with more severe burns had elevated IL-6 for longer periods.
The early parts of the acute phase response were postively correlated with increased IL-6
levels, compatible with a causal role of IL-6 in the induction of these responses.
108
Chapter 10
Procalcitonin
Chapter 6 addressed that hypothesis that endotoxin is not a sine qua non for the induction of
PCT. We found that in vitro, PCT is produced by human liver slices in significant amounts
and that PCT-production could be stimulated with IL-6 or TNFα. In vivo, it was also found
that IL-6 and TNFα induce PCT. Thus was concluded that cytokines without the presence of
endotoxin can induce PCT. In the patients treated with TNFα, the very pronounced acute
phase response that was induced also allowed the comparison of response times of PCT and
CRP. PCT reached half-maximal levels within 8 hours, compared to 20 hours for CRP.
Primary changes in platelet count - thrombocytopenia
Early decreases in platelet count occur in nearly all trauma patients and in other critically ill
patients. It is also known that the magnitude of the early drop in platelet count is related with
outcome.
Chapter 7 investigated early platelet sequestration after major trauma, and the effect of highsteroids on platelet counts. Changes of the platelet count and hemoglobin in the first two days
after moderate injury were measured retrospectively. In two groups of matched patients, one
group received a very high dose of methylprednisolone starting 4 hours after the injury. Over
48 hours, the platelet count dropped by 39% whereas hemoglobin only dropped by 14%. By
taking transfusions and changes in hemoglobin into account, it was concluded that the posttraumatic thrombocytopenia is mainly caused by sequestration of platelets, not blood loss.
Methylprednisolone had no effect on platelet-sequestration. Methylprednisolone is a very
powerful and pleiotropic inhibitor of inflammation that can prevent platelet sequestration if
administered before an inflammatory stimulus. However, inflammation-induced platelet
sequestration apparently becomes refractory to steroid treatment within hours.
Secondary changes in platelet count - (blunted) thrombocytosis
Based on extensive evidence that links systemic inflammation with the sequestration of
platelets we hypothesized that secondary changes in platelet count would be also be related
with outcome.
In chapter 8 this conjecture was verified in patients in our own ICU. Surgical ICU patients
were analyzed for temporal changes in the platelet count as well as the leukocyte count. Both
in the large (N=1415) overall group and in all subgroups (trauma, abdominal surgery,
vascular surgery and liver transplantation) the same phenomena were observed: Platelet
counts decrease to nadir values at 2 to 3 days after ICU admission and failure of the platelet
count to recover to normal or supranormal levels is associated with mortality. Leukocyte
counts were not associated with mortality. The parameter ∆PC/∆t between day 2 and day 10
was a better predictor of mortality then the APACHE-II score at ICU-admission.
In chapter 9 we tested the assumption that the observations for surgical ICU patients in
chapter 8 are also true for ICU patients in general. In addition we verified if secondary
changes in PC (i.e. after ICU day 2) have a stronger correlation with outcome than initial
109
Summary, discussion and lines of further investigation
changes in PC. Thus the relation between temporal changes in PC and mortality were verified
in a larger mixed patient set from the second European ICU Study (EURICUS-II; N=5206).
In this population as well, low ∆PC/∆t values were associated with mortality. Changes after
day 2 indeed had a stronger relation with outcome than initial changes in platelet counts.
DISCUSSION AND FUTURE DEVELOPMENTS
Although we formally only studied patients and not healthy persons, trauma patients can be
effectively considered as previously healthy persons who subsequently sustained an injury.
This is the justification for the title of this thesis "Acute systemic inflammation in health and
disease".
Except for chapters 8 and 9 where a substantial number of patients with underlying chronic
disease were also included, the other chapters concern patients who were basically previously
healthy.
Table 10.1. (See cover)
Sequence of inflammatory parameter changes studied in this thesis
Time after
inflammatory
stimulus
Hour
Hours
Day
Days
Week
Weeks
Parameter
TNFαÇ, IL-6 Ç
fever, tachycardia
platelet count È
leukocyte count Ç
PCT Ç
CRP Ç, SAA Ç
α1-antiproteinaseÇ
albumin È
platelet count Ç/=
IgM Ç
IgG Ç
The main purpose of this thesis was to define the time course of systemic inflammatory parameters and thus help address hypotheses concerning the role of these parameters in the
inflammatory sequence. The implications of the findings are discussed in more detail below.
Table 10.1 summarizes the time-scale and the associated order in which parameter changes
were observed. The final paragraph of this chapter discusses what should be the properties of
an ‘ideal’ inflammatory marker and why this is not necessarily the same as a marker of
disease severity. In this context, procalcitonin and the platelet count are considered as markers
of inflammation and severity respectively.
Inflammation in health and disease - Physiological and pathophysiological response
110
Chapter 10
Classifying responses as either physiological or pathophysiological depends to a certain
degree on an arbitrary cut-off. Nobody would probably call wound healing an unhealthy
response. On the other hand, septic shock due to an exaggerated release of cytokines after
endotoxin stimulation is definitely not a beneficial response. Thus calling a response
physiological depends on the extent of the primary stimulus and the induced responses and
whether the host is overwhelmed or not. We consider the trauma patients previously healthy
and assume that their primary responses are physiological. The severity of illness varied in the
different studies: the patients studied in chapters 2, 3, 6 and 7 were mainly admitted to the
ward, patients in chapters 4 and 5 to the burn unit, while patients studied in chapters 8 and 9
were admitted to the ICU. The systemic inflammatory responses in the moderately injured
patients who stayed at the ward could still reasonably be called physiological, since these
responses still are beneficial and do not appear to worsen the patient's condition. In intensive
care patients on the other hand, frequently disproportional host responses are observed that
can be interpreted as pathophysiological.
The fat embolism syndrome: a peculiar case of inflammation
The first two chapters concerned the fat embolism syndrome (FES). A femoral shaft fracture
is the typical trauma associated with FES. FES is the secondary systemic result (generalized
petechiae, cerebral and respiratory disturbances) of a primary local trauma. Since FES
develops in far fewer patients than the number of patients in which release of bone marrow fat
is observed, additional pathogenic mechanisms must be present. In chapter 2 we found that
local factors, in particular a closed femoral fracture coupled with late operative stabilization
of this fracture, was associated FES. Thus we interpreted that femoral fractures where the
surrounding soft tissues (skin and muscle) were not decompressed, led to FES. A conceivable
mechanical explanation is that increased pressure around the fracture enhances the entrance of
fat into the systemic circulation. The other major observation in chapter 2 is the association of
FES with an early onset of fever.
Nowadays with modern early fracture stabilization and supportive volume therapy, FES is
rarely seen as an isolated syndrome [1]. But regardless of the decreasing incidence, FES
remains a specific entity with an unknown pathogenesis. Causal treatment of FES is still not
possible, while positive end-expiratory pressure ventilation is the mainstay of supportive
therapy in cases of pulmonary insufficiency. Many prophylactic schemes have been tested,
but none have shown real benefit. As we and others have observed, early operative
stabilization of fractures lowers the incidence of FES. The higher incidence of FES [2] when
multiple femoral fractures are present [3] implies that in multiple injured patients FES is often
present, but can sometimes not be recognized due to the concomitant injuries to the brain or
lungs.
Several years after the our publication that dismissed patent foramen ovale as relevant in FES
(chapter 3), a report based on one patient with presumed FES and a patent foramen ovale
appeared in the New England Journal of Medicine [4] The authors claimed that patent
foramen ovale is important in the pathogenesis of FES. As pointed out in chapter 3 we
111
Summary, discussion and lines of further investigation
disagree since the single patient described in this report did not conform to important
diagnostic criteria of classical FES. Later, Forteza [5] and colleagues showed the presence of
cerebral fat emboli with transcranial doppler in 5 patients with long bone fractures. 4 of these
5 patients did not have a patent foramen ovale. It should be noted that, in contrast to FES, the
presence of persisting foramen ovale is an important risk factor in ‘macro-embolic’ entities
such as cerebral thrombo-embolism or air-embolism.[6,7].
The association of fat embolism with early fever, and thus with early systemic inflammation,
that we found has also been observed by others [8]. Elevated levels of CRP have been
implicated as a cause, since CRP has fat agglutinating potential under certain circumstances
[9].
Some remaining questions regarding FES:
⋅ Fat and marrow embolization can be imaged by TEE nearly always during orthopedic
surgery [10]. Why does it not harm the vast majority of patients?
⋅ Neutral fat which is relatively non-toxic can be converted to much more reactive fatty
acids by lipase or lipoprotein lipase. These fatty acids can be toxic to the pulmonary
vasculature [11]. But it is still not clear if this biochemical process plays a significant role
in FES.
⋅ How do large (up to 100 µm) fat globules manage to pass the glomerulus [16], a filter that
otherwise prevents the leakage of much smaller macro-molecules or lipoproteins?
After decades of experiments, a suitable animal model still needs to be developed. The socalled models of FES typically involved rather extreme stimuli, rapidly resultant in the
animals death. Typical models executed were: severe experimental shock [13], acute right
heart failure [14] after massive fat infusion or acute oleic acid pulmonary injury models [15].
Rapidly developing shock is part of these models, but shock is not part of the classic FES.
Also, none of these models reproduces the intriguing interval free of symptoms before the
FES becomes manifest.
Full-blown FES with petechiae and respiratory distress and cerebral disturbances has become
less frequent, but subclinical manifestations with permanent damage may frequently occur.
When a successful animal model of fat embolism is developed the mechanisms by which fat
emboli are pathogenic may be understood more clearly. Such models could lead to directed
therapy that may limit subclinical damage by fat emboli that may be more pervasive and
irreversible [17] than the acute clinical condition suggests.
112
Chapter 10
IL-6
Since our first studies on IL-6, an extraordinary amount of research has been performed on
IL-6. Together with many other cytokines, such as TNFα and IL-1, IL-6 is produced shortly
after the local tissue damage [18] or systemic endotoxin stimulation occurs. Amongst others,
monocytes and endothelial cells can produce large amounts of IL-6 [19,20]. After the
discovery of many other cytokines that have many functions, IL-6 still stands out for its very
broad range of actions. Platelet production is stimulated by IL-6 [21,22] and thrombopoietin
[23]. The sharp IgM-peak during the second week and later the more sustained elevation in
IgG, reflect immunoglobulin stimulation by the cytokines, especially by IL-6 [24]. IL-6
appears to be involved in the pathogenesis of multiple myeloma [25] and Castleman's disease
[26]. In fact, sustained inflammation is in general associated with hypergammaglobulinemia
and under these conditions high IgG levels contribute to the elevated ESR [27] as is shown by
determining the ‘defibrinated’ ESR. Although obsolete now, measuring the ESR after fibrin
removal was an elegant way of quantifying chronic inflammation [28].
Unlike in patients with mechanical injury, in sepsis cytokines often do more harm than good.
It has now been extensively shown that shock in sepsis is mediated by sometimes extremely
high levels of cytokines such as TNFα, IL-6 and IL-1 [29], due to massive activation by
endotoxin. Thus after limited trauma the inflammatory often is in a compensated state, but in
sepsis it is overwhelmed, resulting in a decompensated state. At that point cytokines have lost
their beneficial function. It is only logical to assume that the inflammatory system was
'designed' by evolution for local purposes, i.e. containment of infection and restoration after
trauma. The exquisite sensitivity of some myeloid cell types to endotoxin is beneficial under
local circumstances. But when local defenses are overwhelmed, the massive induction of
cytokines serves no purpose, and will be lethal unless rapid medical intervention is
performed. Depending on the sort of disease, on the stage of the disease and on the
investigator, some cytokines have been classified as inflammatory or anti-inflammatory.
Although IL-6 levels are especially increased in sepsis, and higher levels associated with poor
outcome, some have nevertheless called IL-6 anti-inflammatory [30]. The observation that
elevated cytokine levels are correlated with increased mortality has generated an array of
sepsis intervention trials. In addition to high-dose corticosteroids, agents aimed at blocking or
inhibiting the effects of endotoxin, TNFα, IL-1-receptor, platelet-activating factor receptor,
bradykinin, prostacyclin and thromboxane-A2 were tested in randomized controlled trials
[31]. Although it may appear conceptually helpful to divide cytokines into pro-inflammatory
and anti-inflammatory, such a division is artificial, because many cytokines have multiple
“functions” and because many functions are performed by multiple cytokines. Thus it appears
more appropriate to view cytokines not as sharply defined hormones, but as connections in a
signalling network in which considerable overlap exists. Overlap and parallel pathways may
partly explain the failure of nearly all sepsis intervention trials to affect mortality. These trials
have been aimed at blocking a single pathway [31]. At the effector level it possibly easier to
divide proteins into pro-inflammatory and anti-inflammatory. For example, the enzyme
elastase that is released by neutrophils could be considered pro-inflammatory. Many acute
113
Summary, discussion and lines of further investigation
phase reactants appear to be anti-inflammatory, like for example α1-proteinase inhibitor, a
serine protease inhibitor that inactivates elastase by binding it. The high molar circulating
concentration of α1-proteinase inhibitor compared to elastase keeps the proteolytic activity of
elastase restricted to a limited space in the direct neighborhood of the activated neutrophils
[32].
Procalcitonin
Two recent studies [32,33] found that in septic patients (35 and 24 patients respectively) PCT
was superior to IL-6, CRP, soluble CD14 and TNFα in early detection of non-survivors (13
and 8 patients respectively).
PCT levels may be an attractive tool for clinical monitoring, but the origin of increased PCTsynthesis may be even more important pathophysiologically. Endocrine cells in the lung have
been initially proposed as a source of increased PCT synthesis after pulmonary injury like
inhalation burns [35] or pneumonitis [36] although others found no relation of PCT elevation
with the presence or absence burn injury with or without inhalation [37]. Despite the
intensive search for a source of PCT, at the time of this writing no significant in vitro
production equaling or surpassing the release of PCT that we found in liver slices [38], has
been reported, regardless of cell type or tissue used. In support of our finding that the liver
produces PCT, hepatic vein levels of PCT were higher than central venous levels in
extracorporeal bypass patients [39].
The in vitro PCT-concentrations that were found in the supernatant after stimulation (≈1
µg/L) were considerable, both when compared to in vivo PCT concentrations, and when
compared to in vitro CRP and SAA concentrations (<1 mg/L). The measured liver synthesis
rate of PCT could account for observed levels to in vivo. We could not find reports of PCTmeasurements of supernatants of hepatocytes cultured with various cytokines, and cytokine-
Table 10.2.
Comparison of PCT with cytokines and acute phase proteins
Function
Source
Molecular weight
Mass concentration
Induction time to half
maximal levels
Half life
Cytokines
Procalcitonin
Acute Phase Proteins
Effector. (Innate
bacteriostatic,coagulation,
protease inhibition)
Signalling
?
Leukocytes, endothelial
cells
<50 000
0.1-1 ng/l
Liver ?
Cell type ?
13 000
1-100 ug/l
> 50 000
10 mg/L – 10 g/l
minutes- hours
8 hrs
>20 hrs
minutes - hours
day
day - days
Hepatocytes
stimulated HepG2 cells produced no PCT (unpublished data). Such measurements may
114
Chapter 10
clarify the role of the liver and which cell type is involved in PCT synthesis, which is
particularly important as this organ is the sole producer of most acute phase proteins.
PCT - cytokine or acute phase protein
When Kushner’s broad definition of an acute phase protein is applied [40], PCT fits the
definition since it is increased by more than 25% in inflammation. But the more important
question is whether PCT behaves as other ‘established’ acute phase proteins like CRP,
fibrinogen or α1-proteinase inhibitor. Whereas CRP has been established as important in
aspecific antibacterial defenses, the potential function of PCT is still unknown. Although, as
the name implies, PCT is a precursor of calcitonin, circulating PCT levels appear to be largely
unrelated to calcitonin levels. This is impressively illustrated by the fact that PCT levels can
increase orders of a magnitude, with unchanged calcitonin levels. The relatively large
quantities of PCT that were produced in vitro by human liver slices, when extrapolated to the
in vivo situation, could account for most PCT-levels observed in septic patients. Since the
time of publication, no other convincing source of PCT has been reported. But the question
remains which cell type within the liver (e.g. hepatocyte, endothelial cell or Kupffer cell) is
the source of PCT. The characteristics of PCT as indicated in the table are intermediate
between cytokines and acute phase proteins.
In experiments in Syrian hamsters, Nylen [41] found that PCT exacerbates mortality in
experimental sepsis. The methodology of this study has been criticized [42] on several points.
The investigators used non-recombinant human PCT in hamsters, and it was inhibited by goat
anti-calcitonin antibodies. Thus the effects observed may not necessarily be due to PCT.
Analogous to the debate if some cytokines are pro-inflammatory or anti-inflammatory, the
fact that very high levels of PCT are associated with poor outcome does of course not
automatically imply that PCT is 'bad' or pro-inflammatory. Related to calcitonin's actions, it
has been suggested that PCT could play a role in osteoclast activity [43]. Maybe PCT is
neither a cytokine, nor an acute phase protein. PCT could also simply have no function, or it
may be a useless by-product.
Clinical value of determining cytokines, acute phase proteins and PCT
In clinical practice CRP has generally substituted the ESR as a marker of inflammation,
although for some acute diagnostic problems and especially chronic diseases clinicians still
use the ESR [44]. The direct measurement of cytokines, such as IL-6 has limited practical
applicability. Due to the short half-life of cytokines (minutes to hours) relatively infrequent
point measurements are only of limited value in assessing the area under the curve.
Nevertheless it has been shown that high IL-6 levels are related to mortality in sepsis patients,
and IL-6-levels >1000 pg/ml have been used as a criterion to enroll patients in the anti-TNF
intervention trials.
115
Summary, discussion and lines of further investigation
The published evidence indicates that the overall response of parameters of the acute phase
response appears to be one-dimensional in the sense that the extent of the overall
inflammatory stimulus determines the height of both cytokine levels and APP’s. A local
infection will hardly result in practically measurable cytokine levels and may give a discrete
elevation of CRP. A major burn will result in elevated cytokine levels and a large increase in
CRP and other acute phase proteins. Sepsis can result in extreme increases of cytokines and
maximal levels of acute phase proteins. According to this view daily CRP-determinations will
inform the clinician on the extent of ongoing inflammation. As reported in chapter 6, PCT
achieves half-maximal levels 12 hours earlier than CRP. This can be a very important clinical
advantage of PCT since in acute situations the clinician is informed 12 hours earlier on the
extent of the systemic inflammation present.
Whatever may be the ultimate role of PCT in addition or even instead of CRP, it appears of
little interest to pursue a role for SAA as an inflammatory marker [45]. SAA dynamics may
be somewhat more pronounced than CRP-dynamics, but both SAA and CRP display a time to
half-maximal levels of 20 hours, as opposed to only 8 hours for PCT [38]. Thus measuring
SAA levels appears to present little additional information when PCT and CRP levels are
known.
Open questions on the role of PCT
⋅ PCT shows an extraordinary rise after severe inflammation. This rise may be more
pronounced than that of any known acute phase protein. Baseline levels of PCT appear to
be below 0.01 µg/L [46,47]. To fully define its dynamic range, reliable baseline levels of
PCT in healthy persons should be determined.
⋅ Does most PCT originate in the liver? Is the endothelial cell the source of PCT, because
inducing PCT in hepatocytes and myeloid cells has not succeeded?
⋅ Can endotoxin reproducibly induce PCT in liver slices? If so, this might represent an
unique in vitro inflammatory model. If cytokine-induction is a prerequisite for PCTsynthesis, this would be a two stage system (stage 1: endotoxin induces cytokines; stage 2:
cytokines induce PCT).
Platelet counts
Causes of secondary (relative) thrombocytopenia
We found that patients who do not survive after admission to a surgical ICU have a blunted or
absent secondary increase in platelet counts [48]. Data on the development of platelet counts
in medical ICU patients in our hospital indicated that these patients also show a relation
between platelet count and outcome [49]. Bleeding had only a minimal overall contribution to
this phenomenon in both groups. Bone marrow synthesis in these patients appears to be more
than adequate [50]. Probably very few of the surgical ICU patients (chapter 8) and few of the
116
Chapter 10
EURICUS-patients (chapter 9) had a compromised bone marrow before ICU-admission.
Besides, leukocyte counts in the non-survivors were at least as high as in survivors [48], also
not indicative of marrow failure. Thus drops in platelet count, or inappropriately low platelet
counts, very probably result from sequestration. We interpret this sequestration of platelets as
a reflection of ongoing disease activity, i.e. SIRS or infection. The prognostic importance of
platelet counts led to its inclusion (with the exclusion of the leukocyte count in some cases) in
ICU severity scores such as the SOFA-score [51].
Two main causes of platelet sequestration can be distinguished: aggregation and adhesion.
Aggregation occurs in the process of (disseminated) coagulation. The prime target of platelet
adhesion is activated endothelium or subendothelium, a process that can occur independent of
coagulation. Although we believe strong arguments exist to assume that platelets adhere to
activated endothelium that is associated with systemic inflammation, as also expanded on
elsewhere in this thesis, some alternative hypotheses exist.
Alternative explanations for the disappearance of platelets
Gando [52,53] contends in an number of studies that DIC with platelet consumption is nearly
always present in several critically ill patient groups. But others found only in 40% of the
thrombocytopenic surgical ICU patients proof of coexisting DIC. In a liver transplant model
extensive adherence of solitary platelets to apoptotic sinusoidal endothelial cells in the
absence of (occlusive) clots was detected and even visualized [54]. Tissue factor (TF), the
generally accepted starting point of the (extrinsic) coagulatory cascade has been detected in
elevated levels in trauma and sepsis [55] as further evidence of disseminated activation of
coagulation. But TF-levels as well as changes in other coagulation parameters are relatively
modest in many patients compared to the extensive disappearance of platelets from the
circulation. For example platelet consumption often occurs in the absence of fibrinogen
depletion [56]. Others have observed so-called hemophagocytosis in bone marrow in very
small, strongly selected patients sets [57], while another study has implicated that
autoimmune phenomena such as platelet associated immunoglobulins have been detected in
critically ill patients [58]. The importance of the qualitative observation of hemophagocytosis
is not established, and quantifying platelet immunoglobulins is methodologically difficult.
Patients who underwent a liver transplantation present a special case, since in these patients
thrombocytopenia is not only strongly coupled to graft function [59] but also to an increased
splenic volume [60].
Early changes in PC and methylprednisolone
Steroids, including methylprednisolone, have been shown in many experimental models to
strongly inhibit endotoxin- or trauma-induced inflammation and limit or prevent the
thrombocytopenia that is seen in these models. In healthy persons high-dose steroids do not
affect the platelet count [61]. But evidently it fails to affect the early deposition of platelets in
trauma patients when given after within hours the trauma. Steroids have failed to affect
outcome in sepsis intervention trials. Unfortunately other sepsis-intervention studies have also
117
Summary, discussion and lines of further investigation
not resulted in decisive benefits for other immunomodulatory agents. The fact that these
studies also showed no impact on platelet counts or subscores based on the platelet count
underscores the apparently important relation of platelet counts and survival.
∆PC/∆t as a marker
Recently a study was published that related the incidence of thrombocytopenia in two medical
ICUs with ICU mortality [62]. After admission, nadir platelet counts below 150 109/L or a
decrease of more than 50% was associated with higher death rates with an odds ratio of 6.0
(CI 3.0-12). The occurrence of thrombocytopenia had more predictive power than either
APACHE-II, SAPS II and MODS scores. The study did not address the relevance of a blunted
rise in platelet counts. In chapter 9 we comprehensively studied time-dependent changes in
PC. In was found that ∆PC/∆t after day 2 had a stronger relation with outcome than early
changes in PC or nadir PC, thus pointing to the importance of day 2 as a ‘turning point’ with
regard to platelet counts
Open questions on time-dependent changes in platelet counts.
⋅ Based on periodic recordings of the ESR (or CRP) and platelet count [63], one could try
to deduce the ‘true individual normal’ platelet count in patients with chronic inflammatory
diseases. The platelet count that is measured at times when ESR and CRP are low could
be assumed to be normal for that particular individual. The individual ∆ platelet count
could subsequently be analyzed in relation with elevated ESR and CRP levels.
⋅ Where do the platelets go to, when platelet consumption is observed during SIRS or
sepsis? In one study that imaged labeled platelets in critically ill patients [68], the
intestinal organs showed the highest platelet activity. It would be interesting to know if
the homing behavior of the platelets is associated with clinical organ damage, for example
acute renal failure or ARDS.
⋅ What are the clinically relevant receptors and ligands in platelet - endothelial cell
interaction in critically ill patients (in the absence of DIC)? In cardiovascular medicine it
has recently been proven that systemic inflammation is an important independent risk
factor for ischemic events [69,70], underscoring the clinical relevance of the relation
between inflammation and coagulation. The great success of glycoprotein IIb/IIIa receptor
(GP-IIb/IIIa) inhibitors in the CCU is [71] might in turn also be related with the antiinflammatory effects of inhibiting GPIIb/IIIa. The P-selectin receptor, that is expressed
by activated endothelial cells, has an important role both in leukocyte adhesion and in
fibrinogen-independent platelet-adhesion [69]. Sindram has suggested that P-selectin is
important in mediating reperfusion injury in transplanted livers [54].
⋅ Is the platelet endothelial interaction (adhesion) always irreversible or is it under some
conditions still reversible? Maybe marginated platelets, analogous to leukocytes, can
reenter the circulation. Both in in vitro and in vivo [73] experiments this phenomenon has
been observed. The spleen is an example of a microenvironment where platelets are
reversibly segregated from the circulation. If platelet re-entry is relevant, part of the
118
Chapter 10
effects of potential agents that selectively block platelet- endothelial interaction might be
reflected by increases in platelet counts.
Arbitrary units
The ideal inflammatory marker
After discussion of the merits of the several parameters that were studied as markers of
inflammation and disease severity, it is useful to define the "ideal” inflammatory marker as a
thought experiment. What should be its properties?
It must have very low values in healthy persons. It increases rapidly once an inflammatory
state develops. Its increase is proportional to the extent of the underlying inflammatory
process - for example proportional to the total amount of damaged tissue or the total quantity
1
10
Hours
100
Figure 10.1. Comparison of levels of ideal inflammatory marker (solid line) with
levels of cytokine (dashed line) after an acute inflammatory event. An ideal
marker rises rapidly after inflammation in a manner proportional to the
inflammatory stimulus. When the inflammation disappears, the marker decreases
with a half-life of one day. The cytokine rises rapidly but also decreases rapidly.
of cytokines produced. After the marker has rapidly increased to this proportional level, it
predictably decreases. The half-time should be around 1 day - a convenient unit of time in
clinical practice. The marker can be determined for little cost in serum or plasma and is not
sensitive to in vitro changes during procurement or storage.
119
Summary, discussion and lines of further investigation
Indicators of inflammation and severity - Procalcitonin and the platelet count.
On the basis of the formulated profile of the ideal inflammatory marker, procalcitonin
probably comes closest to these requirements. Disadvantages of PCT as an inflammatory
marker may be blunted responses in neutropenia [74] and interindividual differences that are
larger than for CRP [75]. The combination of CRP and PCT may turn out to be a particularly
effective indicator of the inflammatory state and the possibility of bacterial infection.
It is important to note that a marker of inflammation is not the same as a marker of severity. A
good marker of severity should be associated with outcome (i.e. mortality, hospital stay),
whereas a marker of inflammation reflects the amplitude of the inflammatory response as
such, regardless of the impact this inflammatory response has on ultimate outcome. Wellknown examples of markers of severity are the severity scores: Acute physiological and
chronic health evaluation (APACHE-II) [76], injury severity score (ISS) [77] or multiple
organ dysfunction score (MODS) [78].
Changes in platelet counts (nadir platelet count and ∆PC/∆t) are also strongly related to
outcome. The fact that ∆PC/∆t after ICU-day 2 is as good a predictor of outcome as the
APACHE-II score in surgical ICU patients [48] emphasizes that platelets are deeply related to
outcome. Correlating platelet count with PCT and mortality may show to what extent the
platelet count reflects inflammation and organ dysfunction respectively.
Arbitrary units
Interactions of acute systemic inflammation with other physiological systems.
Many interactions between the inflammatory system and other (patho)physiological systems
have been uncovered. Except where platelets and coagulation are concerned, these systems
were not subject of this thesis. To illustrate that inflammation has connections with many
1
10
Hours
100
Figure 10.2. Comparison of CRP (dashed line) with ideal inflammatory marker
(solid line).
120
Arbitrary units
Chapter 10
1
10
Hours
100
Figure 10.3. Comparison of levels of PCT (dashed line) with an ideal
inflammatory marker (solid line). PCT rises considerably faster than CRP, but still
has a relatively long half-life.
other important physiological processes some examples:
− Inflammation ↔ specific immune response:
Example: induction of antigen specific immunoglobulin and T-cell responses
− Inflammation ↔ stress system
Examples: Astronauts produce increased levels of IL-6 during launch and landing [79];
exercise induces cytokines [80] and acute phase proteins [81]
− Inflammation ↔ central nervous system
Example: inflammation and IL-6 induce sleep [82]; sleep induces IL-6 [83]
− Inflammation ↔ metabolism
Example: down-regulation of albumin and induction of cachexia by inflammation
[84,85,86]
Systemic inflammation in health and disease
To summarize, in this thesis several aspects of acute systemic inflammation were studied.
FES is a special form of systemic inflammation in which fat globuli are involved. Why this
response occurs only in a small subset of patients with long bone fractures and why it does
not occur in the vast majority of apparently similar patients has only partly been clarified.
SIRS with it’s associated increases in cytokines, PCT and acute phase proteins, is a universal
response in patients who are critically ill, whether due to trauma, surgery or infection.
121
Summary, discussion and lines of further investigation
Sequestration of platelets accompanies SIRS initially in virtually all patients, and secondarily
in many patients with complicated courses. For a number of reasons, PCT and the platelet
count stand out for their diagnostic and prognostic value compared to many other parameters.
Although we do only partly understand why this is the case, further investigation into these
parameters must lead to phenomena central to the outcome of critically ill patients.
REFERENCES
1.
Bulger EM, Smith DG, Maier RV, Jurkovich GJ. Fat embolism syndrome. A 10-year review.
Arch Surg 1997;132:435-439.
2.
Pinney SJ, Keating JF, Meek RN. Fat embolism syndrome in isolated femoral fractures: does
timing of nailing influence incidence? Injury 1998;29(2):131-133.
3.
ten Duis. Fat embolism syndrome. Injury 1997;28:77-85.
4.
Pell AC, Hughes D, Keating J, Christie J, Busuttil A, Sutherland GR. Brief report:
fulminating fat embolism syndrome caused by paradoxical embolism through a patent
foramen ovale. N Engl J Med 1993;329:926-999.
5.
Forteza AM, Koch S, Romano JG, Zych G, Bustillo IC, Duncan RC, Babikian VL.
Transcranial doppler detection of fat emboli. Stroke 1999;30:2687-2691
6.
Lechat P, Mas JL, Lascault G, Loron P, Theard M, Klimczac M, Drobinski G, Thomas D,
Grosgogeat Y. Prevalence of patent foramen ovale in patients with stroke. N Engl J Med
1988;318:1148-1152.
7.
Moon RE, Camporesi EM, Kisslo JA. Patent foramen ovale in decompression sickness in
divers. Lancet 1989;i:513-514.
8.
Peltier LF. Fat embolism, an appraisal of the problem. Clin Orthop 1984;187:3.
9.
Hulman G. Fat macroglobule formation from chylomicrons and non-traumatic fat embolism.
Clin Chim Acta 1988;177:173-138.
10.
Christie J, Robinson CM, Pell AC, McBirnie J, Burnett R. Transcardiac echocardiography
during invasive intramedullary procedures. J Bone Joint Surg Br 1995;77:450-455.
11.
Fonte DA, Hausberger FX. Pulmonary free fatty acids in experimental fat embolism. J
Trauma 1971;11:668-672.
12.
Tyrrell DJ, Horne AP, Holme KR, Preuss JM, Page CP. Heparin in inflammation: potential
therapeutic applications beyond anticoagulation. Adv Pharmacol 1999;46:151-208.
13.
Derks CM, Peters RM. The role of shock and fat embolus in leakage from pulmonary
capillaries. Surg Gynecol Obstet 1973;37:945-948.
14.
Tornabene VW, Fortune JB, Wgner PD, Halasz NA. Gas exchange after pulmonary fat
embolism in dogs. J Thorac Cardiovasc Surg 1979;78:589.
15.
Jones JG, Minty BD, Beeley JM, Royston D, Crow J, Grossman RF Pulmonary epithelial
permeability is immediately increased after embolisation with oleic acid but not with neutral
fat. Thorax 1982;37:169-174.
122
Chapter 10
16.
Fabian TC, Hoots AV, Stanford DS, Patterson CR, Mangiante EC. Fat embolism syndrome:
prospective evaluation in 92 fracture patients. Crit Care Med 1990;18:42-64.
17.
Erdem E, Namer IJ, Saribas O, Aras T, Tan E, Bekdik C, Ziledi T. Cerebral fat embolism
studied with MRI and SPECT. Neuroradiology 1993;35:199-201.
18.
Ueo H, Inoue H, Honda M, Uchida I, Nischimura M, Aranaga S, Nakashima H, Akiyoshi T.
Production of interleukin-6 at operative wound sites in surgical patients. J Am Coll Surg
1994;326-32.
19.
Aarden L, Helle M, Boeije L, Pascual-Salcedo D, de Groot E. Differential induction of
interleukin-6 production in monocytes, endothelial cells and smooth muscle cells. Eur
Cytokine Netw 1991;2:115-20.
20.
Krishnaswamy G, Kelley J, Yerra L, Smith JK, Chi DS. Human endothelium as a source
multifunctional cytokines: molecular regulation and possible role in human disease. J
Interferon Cytokine Res 1999;19:91-104.
21.
Hollen CW, Henthorn J, Koziol JA, Burstein SA. Elevated serum IL-6 levels in patients with
reactive thrombocytosis. B J Haematol 1991;79:286-290.
22.
van Gameren MM, Willemse PHB, Mulder NH, et al.: Effects of recombinant human
interleukin-6 in cancer patients: a phase I-II study. Blood 1994; 84:1434-1441.
23.
Kaushansky K. Thrombopoietin. N Engl J Med 1998;339:746-754.
24.
Fattori E, Della Rocca C, Costa P, Giorgio M, Dente B, Pozzi L, Ciliberto G. Development
of progressive kidney damage and myeloma kidney in interleukin-6 transgenic mice. Blood
1994;83:2570-2579
25.
Sati HI, Apperley JF, Greaves M, Lawry J, Gooding R, Russell RG, Croucher PI.
Interleukin-6 is expressed by plasma cells from patients with multiple myeloma and
monoclonal gammopathy of undetermined significance. Br J Haematol 1998;101:287-295
26.
Nishimoto N, Sasai M, Shima Y, Nakagawa M, Matsumoto T, Shirai T, Kishimoto T,
Yoshizaki K. Improvement in Castleman's disease by humanized anti-interleukin-6 receptor
antibody therapy. Blood 2000;95:56-61.
27.
Talstad I, Haugen HF. The relationship between the erythrocyte sedimentation rate (ESR)
and plasma proteins in clinical materials and models. Scand J Clin Lab Invest 1979;39:51924.
28.
Gudmundsson M, Bjelle A. Viscosity of plasma and blood in rheumatoid arthritis.
Br J Rheumatol. 1993;32:774-779.
29.
Hack CE, Aarden LA, Thijs LG. Role of cytokines in sepsis. Adv Immunol 1997;66:101-195.
30.
Xing Z,Gauldie J, Cox G, Baumann H, Jordana M, Lei XFR, Achong MK. IL-6 is an antiinflammatory cytokine required for controlling local or systemic acute inflammatory
responses. J Clin Invest 1998;101:311-320.
31.
Abraham E. Why immunomodulatory therapies have not worked in sepsis. Intensive Care
Med 1999;25:556-566.
32.
Weiss SJ. Tissue destruction by neutrophils. N Engl J Med 1989;320:365-76
33.
Herrmann W, Ecker D, Quast S, Klieden M, Rose S, Marzi I. Comparison of procalcitonin,
sCD14 and interleukin-6 values in septic patients. Clin Chem Lab Med 2000;38:41-6
123
Summary, discussion and lines of further investigation
34.
Schroder J, Staubach KH, Zabel P, Stuber F, Kremer B. Procalcitonin as a marker of severity
of septic shock. Langenbecks Arch Surg 1999;384:33-38.
35.
Nylen ES, Jeng J, Jordan MH, Snider RH, Thompson KA, Lewis MS, O'Neill WJ, Becker
KL. Late pulmonary sequela following burns: persistence of hyperprocalcitonemia using a 157 amino acid N-terminal flanking peptide assay. Respir Med 1995;89:41-46.
36.
Nylen ES, Snider RH Jr, Thompson KA, Rohatgi P, Becker KL. Pneumonitis-associated
hyperprocalcitoninemia. Am J Med Sci 1996;312:12-18.
37.
Carsin H, Assicot M, Feger F, Roy O, Pennacino I, Le Bever H, Ainaud P, Bohuon C.
Evolution and significance of circulating procalcitonin levels compared with IL-6, TNF
alpha and endotoxin levels early after thermal injury. Burns 1997;23:218-224.
38.
Nijsten MW, Olinga P, The TH, de Vries EG, Koops HS, Groothuis GM, Limburg PC, ten
Duis HJ, Moshage H, Hoekstra HJ, Bijzet J, Zwaveling JH. Procalcitonin behaves as a fast
responding acute phase protein in vivo and in vitro. Crit Care Med 2000;28:458-461.
39.
Silomon M, Bach F, Ecker D, Graeter T, Grundmann U, Larsen R. Procalcitonin nach
extrakorporaler
Zirkulation.
Synthese
im
Hepatosplanchnikusgebiet?
Anaesthesist
1999;48:395-398.
40.
Kushner I. The phenomenon of the acute phase response. Ann N Y Acad Sci 1982;389:39-48.
41.
Nylen ES, Whang KT, Snider RH, et al.Mortality is increased by procalcitonin and
decreased by an antiserum reactive to procalcitonin in experimental sepsis. Crit Care Med
1998; 26:1001-1006.
42.
Braithwaite SS. Procalcitonin – marker or mediator? Crit Care Med 1998; 26:977-978.
43.
Braithwaite SS. Procalcitonin: new insights on regulation and origin. Crit Care Med
2000;28:586-587.
44.
Brigden M. The erythrocyte sedimentation rate. Still a helpful test when used judiciously.
Postgrad Med 1998;103:257-274.
45.
Mozes G,Friedman N,Shainkin-Kestenbaum R. Serum amyloid A: an extremely sensitive
marker for intensity of tissue damage in trauma patients and indicator of acute response in
various diseases. J Trauma 1989;29:71-74.
46.
Snider RH Jr, Nylen ES, Becker KL: Procalcitonin and its component peptides in systemic
inflammation: immunochemical characterisation. J Investig Med 1997; 45:552-560.
47.
Dandona P, Nix D, Wilson MF, Aljada A, Love J, Assicot M, Bohuon C. Procalcitonin
increase after endotoxin injection in normal subjects.
J Clin Endocrinol Metab
1994;79:1605-1608.
48.
Nijsten MWN, ten Duis HJ, Zijlstra JG, Porte RJ, et al. Blunted rise in platelet count in
critically ill patients is associated with worse outcome. Crit Care Med 2000;28:3843-3846.
49.
Nijsten MWN, Tulleken JE, vd Werf TS. Absence of secondary thrombocytosis is associated
with prolonged ICU-stay and decreased survival. Intensive Care Med 1997;23 S1:59.
50.
Mitterstieler G, Kurz R, Haas H, Desoye G. Thrombopoiesis and blood coagulation in
pediatric patients with septicemia. Padiatr Padol 1981;16:257-65.
51.
Vincent et al. Use of the SOFA score to assess the incidence of organ dysfunction/failure in
intensive care units: results of a multi-center, prospective study. Working group on "sepsis-
124
Chapter 10
related problems" of the European Society of Intensive Care Medicine. Crit Care Med
1998;26:1793-1800.
52.
Gando S, Nanzaki S, Kemmotsu O. Disseminated intravascular coagulation and sustained
systemic inflammatory response syndrome predict organ dysfunctions after trauma:
application of clinical decision analysis. Ann Surg 1999;229:121-127.
53.
Gando S, Kameue T, Nanzaki S, Nakanishi Y. Disseminated intravascular coagulation is a
frequent complication of systemic inflammatory response syndrome. Thromb Haemost 1996
;75:224-228.
54.
Sindram D, Porte RJ, Hoffman MR, Bentley RC, Clavien PA. Platelets induce sinusoidal
endothelial cell apoptosis upon reperfusion of the cold ischemic rat liver. Gastroenterology
2000;118:183-191.
55.
Gando S, Nanzaki S, Morimoto Y, Kobayashi S, Kemmotsu O. Systemic activation of
tissue-factor dependent coagulation pathway in evolving acute respiratory distress syndrome
in patients with trauma and sepsis. J Trauma 1999;47:719-23.
56.
Neame PB, Kelton JG, Walker. Stewart IO, Nossel HL, Hirsch J (1980) Thrombocytopenia
in septicemia: the role of intravascular coagulation. Blood 56:88-92.
57.
Stephan F, Hollande J, Richard O, Cheffi A, Maier-Redelsperger M, Flahault A
Thrombocytopenia in a surgical ICU. Chest 1999;115:1363-70.
58.
Stephan F, Cheffi MA, Kaplan C, Maillet J, Novara A, Fagon J, Bonnet F. Autoantibodies
against platelet glycoproteins in critically ill patients with thrombocytopenia. Am J Med
2000;108:554-60.
59.
Chatzipetrou MA, Tsaroucha AK, Weppler D, Pappas PA, Kenyon NS et al.
Thrombocytopenia after liver transplantation. Transplantation 1999;67:702-706.
60.
Hoefs JC, Wang FW, Lilien DL, Walker B, Kanel G] A novel, simple method of functional
spleen volume calculation by liver-spleen scan. J Nucl Med 1999;40:1745-1755.
61.
Demir G, Derman U, Berkarda B. Haematological effects of pulse steroid therapy. Int J
Pharmacol Res 1994;14:101-106.
62.
Vanderschueren S, de Weerdt A, Malbrain M, Vankersschaever D, Frans E, Wilmer A,
Bobbaers H. Thrombocytopenia and the prognosis in intensive care. Crit Care Med
2000;28:1871-1876.
63.
Buckley MF, James JW, Brown DE, Whyte GS, Dean MG, Chesterman CN, Donald JA. A
novel approach to the assessment of variations in the human platelet count. Thromb Haemost
2000;83:480-484.
64.
Porte RJ, Blauw E, Knot EAR, et al: Role of the donor liver in the origin of platelet
disorders and hyperfibrinolysis in liver transplantation. J Hepatol 1994;21:592-600.
65.
McCaughan GW, Herkes R, Powers B, et al: Thrombocytopenia post liver transplantation.
Correlations with pre-operative platelet count, blood transfusion requirements, allograft
function and outcome. J Hepatol 1992;16:16-22.
66.
Kas-Deelen AM, de Maar EF, Harmsen MC, Driessen C, van Son WJ, The TH. Uninfected
and cytomegalic endothelial cells in blood during cytomegalovirus infection: effect of acute
rejection. J Infect Dis 2000;181:721-724.
125
Summary, discussion and lines of further investigation
67.
Almeida-Porada GD, Ascensao JL. Cytomegalovirus as a cause of pancytopenia. Leuk
Lymphoma 1996;21:217-223.
68.
Sigurdsson GH, Christenson JT, el-Rakshy MB, et al: Intestinal platelet trapping after
traumatic and septic shock. An early sign of sepsis and multiorgan failure in critically ill
patients? Crit Care Med 1992;20:458-467.
69.
Verheggen PW, de Maat MP, Cats VM, Haverkate F, Zwinderman AH, Kluft C, Bruschke
AV. Inflammatory status as a main determinant of outcome in patients with unstable angina,
independent of coagulation activation and endothelial cell activation. Eur Heart J
1999;20:567-574.
70.
Vallance P,Collier J, Bhagat K. Infection, inflammation, and infarction: does acute
endothelial dysfunction provide a link? Lancet 1997;349:1391-1392.
71.
Cannon CP. Incorporating platelet glycoprotein IIb/IIIa inhibition in critical pathways:
unstable
angina/non-ST-segment
elevation
myocardial
infarction.
Clin
Cardiol
1999;22(Suppl):IV30-36.
72.
Frenette PS, Denis CV, Weiss L, Jurk K, et al. P-Selectin glycoprotein ligand 1 (PSGL-1) is
expressed on platelets and can mediate platelet-endothelial interactions in vivo. J Exp Med
2000;191:1413-1422.
73.
Woltjes J, de Jong JC, ten Duis HJ, Wildevuur CR.The priming of extracorporeal circuits:
the effect on canine blood elements. Transfusion 1979;19:552-557.
74.
de Bont E, Vellenga E, Swaanenburg J, Kamps W. Procalcitonin: a diagnostic marker of
bacterial infection in neutropenic cancer patients with fever? Infection 2000; 28:398-400.
75.
Meisner M. Procalcitonin. A new innovative infection parameter. Biochemical and clinical
aspects. Berlin: Brahms Diagnostica 1996. ISBN 3-00-000803-9
76.
Knaus WA, Draper EA, Wanger DP, Zimmerman JE. APACHE-II: a severity of disease
classification system. Crit Care Med 1985;13:818-829.
77.
Baker SP, O'Neill B. The injury severity score: un update. J Trauma 1976;16:882-885.
78.
Marshall JC, Cook DJ, Christou NV, Bernard GR, Sprung CL, Sibbald WJ. Multiple organ
dysfunction score: a reliable descriptor of a complex clinical outcome. Crit Care Med
1995;23:1638-1652.
79.
Stein TP, Schluter MD. Excretion of IL-6 by astronauts during spaceflight. Am J Physiol
1994;266:448-452.
80.
Pedersen BK, Ostrowski K, Rohde T, Bruunsgaard H. The cytokine response to strenuous
exercise. Can J Physiol Pharmacol 1998;76:505-511.
81.
Shek PN, Shepard RJ. Physical exercise as a model of limited inflammatory response. Can J
Physiol Pharmacol 1998;76:589-597.
82.
Spath-Schwalbe E, Hansen K, Schmidt F, Schrezenmeier H, Marshall L, Burger K, Fehm
HL, Born J. Acute effects of recombinant human interleukin-6 on endocrine and central
nervous sleep functions in healthy men. J Clin Endocrinol Metab 1998;83:1573-1579.
83.
Vgontzas AN, Papanicolaou DA, Bixler EO, Lotsikas A, Zachman K, et al. Circadian
interleukin-6 secretion and quantity and depth of sleep. J Clin Endocrinol Metab
1999;84:2603-2607.
126
Chapter 10
84.
LeGrand EK. Why infection-induced anorexia? The case for enhanced apoptosis of infected
cells. Med Hypotheses 2000;54:597-602.
85.
Haupt W, Holzheimer RG, Riese J, Klein P, Hohenberger W. Association of low
preoperative serum albumin concentrations and the acute phase response. Eur J Surg
1999;165:307-13.
86.
Michie HR. Metabolism of sepsis and multiple organ failure. World J Surg 1996;20:460-464.
127
Summary, discussion and lines of further investigation
128
NEDERLANDSE SAMENVATTING
Acute systemische ontsteking bij gezonden en zieken
Acute systemische ontsteking
In dit proefschrift zijn een aantal systemische uitingen van acute ontsteking onderzocht. Deze
kunnen grotendeels als generalisaties van de locale ontsteking worden beschouwd. Klassiek
worden 5 kenmerken van de locale ontsteking beschreven, te weten calor, tumor, rubor, functio
laesa en dolor. De systemische equivalenten van de eerste vier uitingen zijn respectievelijk koorts,
oedeem, erytheem en orgaanfalen. Het begrip acute fase reactie wordt gebruikt om delen van deze
gecoördineerde respons te beschrijven. Behalve koorts kenmerkt de acute fase reactie zich door
verhoogde spiegels van een scala aan effector eiwitten, de zogenaamde acute fase eiwitten.
Metingen van deze eiwitten, met als prototype het C-reactive protein (CRP), maken het mogelijk
de omvang en duur van de acute fase reactie te kwantificeren. De afgelopen jaren is duidelijk
geworden dat cytokinen een centrale rol spelen bij de inductie van de acute fase reactie.
Behalve sterk gestegen cytokine-spiegels en verhoogde spiegels van acute fase eiwitten bestaan er
meer markers in het bloed die wijzen op de aanwezigheid van een systemische ontstekingsreactie.
Voorbeelden hiervan zijn het leukocytengetal (aantal witte bloedcellen) en het trombocytengetal
(aantal bloedplaatjes).
Om een beter beeld te krijgen van de wijze waarop de diverse responsen met elkaar samenhangen,
is het van belang de volgorde waarin de verschillende markers veranderen, te bepalen. Om deze
volgorde zo duidelijk mogelijk waar te nemen is het zinvol ongevalspatiënten te onderzoeken. Bij
deze patiënten, die tevoren in de regel gezond waren, is het moment waarop de ontstekingsprikkel
ontstaat (het ongeval) duidelijk. Dit in tegenstelling bijvoorbeeld tot een chronisch zieke patiënt
die geleidelijk aan een sepsis ontwikkelt.
Het stereotiepe gedrag van de acute systemische ontsteking bij ongevalspatiënten, heeft als
methodologisch voordeel dat bij relatief kleine patiëntengroepen reeds significante patronen
zichtbaar worden.
Het vetembolie syndroom
De hoofdstukken 2 en 3 betreffen een bijzonder soort ontstekingsreactie, het vetemboliesyndroom (FES). FES ontstaat soms bij patiënten die één of meer lange pijpbeenderen hebben
gebroken. Na een klachtenvrij interval ontwikkelt de patiënt oxygenatie-stoornissen, cerebrale
stoornissen en petechiën. Van deze drie manifestaties is aangetoond dat zij door embolisatie van
beenmergvet veroorzaakt worden. Aangezien bij de meeste patiënten met fracturen, circulerend
beenmergvet kan worden aangetoond, is het onduidelijk welke factoren ervoor verantwoordelijk
zijn dat slechts een minderheid van de patiënten FES ontwikkelt.
In hoofdstuk 2 is gekeken naar de relatie tussen kenmerken van de fractuur en het optreden van
FES. Tevens is gekeken naar de vroege verschijnselen van systemische ontsteking. Het bleek dat
een gesloten, en niet direct geopereerde fractuur (de zogenaamde non-decompressie groep)
geassocieerd was met verhoogde kans op FES. Ook bleken de patiënten met FES reeds vroeg (dag
0) koorts te krijgen, hetgeen suggereert dat systemische ontsteking een rol speelt bij de inductie
van FES. Daar bekend is dat CRP vet kan agglutineren, is als mogelijke verklaring voorgesteld
130
Nederlandse samenvatting
dat verhoogde CRP spiegels de agglutinatie en embolisatie van circulerend beenmergvet
bevorderen.
In hoofdstuk 3 is getoetst of patiënten die een FES hebben doorgemaakt een open foramen ovale
hebben. De aanwezigheid van een open foramen ovale maakt een rechts-links shunt mogelijk
zodat embolieën de grote circulatie kunnen bereiken zonder het longvaatbed te passeren. Voor
cerebrale luchtembolieën en thrombo-embolieën is aangetoond dat een open foramen ovale het
risico hierop inderdaad verhoogt. Bij onderzoek met transoesophageale echocardiografie bij
patiënten die een FES hadden doorgemaakt, werden geen relevante shunts gevonden. Derhalve is
het onwaarschijnlijk dat een open foramen ovale een belangrijke rol speelt bij FES. Blijkbaar zijn
de vetbollen in staat zich te vervormen en zo de longcapillairen te passeren waarmee ze de grote
circulatie te bereiken.
Interleukine-6 en de acute fase reactie
In hoofdstuk 4 en 5 zijn interleukine-6 (IL-6) spiegels, tezamen met een aantal markers van
ontsteking, bij patiënten met brandwonden gemeten. IL-6 bleek gecorreleerd met CRP en vooral
met koorts.
Zoals schematisch op de omslag van dit proefschrift is weergegeven, werd een duidelijke
volgorde van de responsen waargenomen: eerst IL-6↑, daarna koorts, CRP↑, α1-antitrypsine↑,
trombocytengetal↑, IgM↑ en IgG↑.
De mogelijkheid dat IL-6 bij deze reacties een causale rol speelt, wordt besproken op basis van de
destijds (±1988) bekende gegevens. Latere onderzoeken, zoals het toedienen van IL-6 aan
mensen, hebben de veronderstelling bevestigd dat IL-6 al de bovengenoemde onderdelen van de
acute fase reactie kan veroorzaken. Voor de volledigheid moet worden aangetekend dat ook
cytokinen zoals tumor necrosis factor α (TNFα) en IL-1 eveneens een belangrijke rol spelen bij
de inductie van de acute fase reactie.
Procalcitonine
De afgelopen vijf jaar staat het eiwit procalcitonine (PCT; 13kD) als nieuwe marker van
ontsteking toenemend in de belangstelling. Verhoogde spiegels van PCT zijn volgens sommige
auteurs met name geassocieerd met systemische ontsteking als gevolg van bacteriële infectie. Als
dit inderdaad het geval is, zou dat van grote klinische betekenis kunnen zijn bij het differentiëren
tussen bacteriële infectie en andere oorzaken van systemische ontsteking. Sommige auteurs
hebben gesuggereerd dat PCT direct door endotoxine zou worden geïnduceerd en niet door de
tussenkomst van cytokinen, zoals dat voor de acute fase eiwitten het geval is. In hoofdstuk 6
hebben wij getoetst of cytokinen in vitro en in vivo PCT kunnen induceren. In een humaan leverslice model kon met zowel TNFα als met IL-6 significante eiwitproductie van PCT worden
geïnduceerd In twee groepen van patiënten met kanker die respectievelijk TNFα en IL-6 kregen
toegediend, werd eveneens duidelijke PCT-productie gemeten. Bij de TNFα behandelde
patiënten bleek tevens dat PCT na circa 8 uur half-maximale waarden bereikte, tegenover 20 uur
bij CRP. Op basis van deze observaties is geconcludeerd dat endotoxine geen noodzakelijke
131
Acute systemische ontsteking bij gezonden en zieken
factor is voor PCT-inductie. De aanzienlijk snellere stijging van PCT dan van CRP maakt PCT in
ieder geval een aantrekkelijke marker bij patiënten bij wie tijdige informatie over de omvang van
de systemische ontsteking gewenst is. In hoofdstuk 6 en ook in hoofdstuk 10 worden de
eigenschappen van PCT vergeleken met die van cytokinen en acute fase eiwitten, in het licht van
de onbekende functie van PCT. De titel van hoofdstuk 6 stelt dat PCT zich gedraagt als een snel
acuut fase eiwit. Of het ook een acuut fase eiwit is, zal nog moeten blijken.
Het tijdsverloop van het trombocytengetal
De hoofdstukken 7, 8 en 9 onderzoeken de betekenis van het aantal trombocyten in patiënten met
ernstige ontstekingsverschijnselen De begrippen stolling en ontsteking werden vroeger als min of
meer aparte processen beschouwd. Maar het is gaandeweg duidelijk geworden dat het om twee
sterk geïntegreerde systemen gaat. Situaties waaraan het stollingssysteem en het
ontstekingssysteem het hoofd moeten bieden, dienen zich vaak op dezelfde plaats en tijd aan. De
trombocyten blijken ook allerlei eigenschappen te hebben die de functie als stollingsdeeltje te
boven gaan. Er zijn allerlei aanwijzingen (hoofdstuk 1, 7, 8, 9 en 10) dat de sequestratie van
trombocyten de hoofdoorzaak is van dalingen in het trombocytengetal zoals die zowel bij
traumapatiënten als bij intensive care (IC) patiënten wordt gezien. Hoewel met gevoelige
methoden bij veel van deze patiënten de aanwezigheid van enige diffuse intravasale stolling
(DIC) kan worden aangetoond, vindt belangrijke sequestratie van trombocyten in allerlei organen
plaats, zonder dat er sprake is van duidelijke DIC.
Het trombocytengetal is eenvoudig, betrouwbaar en goedkoop te bepalen. Terwijl veel onderzoek
naar (kwalitatieve) eigenschappen van trombocyten verricht is, is het gedrag van het
trombocytengetal relatief weinig bestudeerd. Afhankelijk van de ernst en van de fase waarin een
systemische ontstekingsreactie zich bevindt, kan het trombocytengetal dalen of stijgen. De
respons van het trombocytengetal na stomp trauma (hoofdstuk 2) of na brandwonden (hoofdstuk
5) is al langer bekend. Het trombocytengetal daalt de eerste twee dagen (primaire verandering),
om daarna te stijgen (secundaire verandering), en bij die patiënten die verder een
ongecompliceerd beloop hebben na ongeveer een week het beeld van trombocytose (>350⋅109/l)
te vertonen.
Trombocytengetal - de primaire veranderingen
Hoofdstuk 7 beschrijft verloop van het trombocytengetal gedurende de eerste 2 dagen bij
patiënten die een matig ernstig ongeval hebben gehad waarbij zij de rug hebben gebroken. Het
trombocytengetal daalt gedurende de eerste 48 uur relatief sneller dan het hemoglobine. Deze
daling van het aantal trombocyten is groter dan uit het bloedverlies kan worden verklaard.
Gezien de argumenten dat dalingen in het trombocytengetal met name het gevolg zijn sequestratie
secundair aan systemische ontsteking, is in hoofdstuk 7 gekeken naar het effect van een dosis
methylprednisolon van ± 10 000 mg/24 uur (!), gestart binnen 4 uur na het ongeval, op het
dalende trombocytengetal. Bij vergelijkbare groepen patiënten met wervel-trauma, was er geen
verschil in daling van het trombocytengetal tussen de methylprednisolon-groep en de
132
Nederlandse samenvatting
controlegroep. Dit betekent dat de processen (endotheel-activatie?) die verantwoordelijk zijn voor
trauma-geïnduceerde trombocyten-sequestratie binnen 4 uur refractair zijn voor een van de meest
potente ontstekingsremmende interventies die momenteel beschikbaar zijn.
Trombocytengetal - de secundaire veranderingen
De meeste patiënten (zowel traumapatiënten als IC-patiënten) vertonen eerst een daling van het
trombocytengetal. Zoals reeds aangegeven, stijgt bij patiënten die een ongecompliceerd klinisch
beloop hebben het trombocytengetal tot supranormale waarden na 1 week. Deze reactieve
thrombocytose is onder andere het gevolg van verhoogde IL-6 spiegels, reden waarom rhIL-6
vóór de ontdekking van het thrombopoietine getest is als thrombopoeiticum. Ondanks het feit dat
de meeste patiënten op de IC verhoogde IL-6 spiegels hebben met een adequate aanmaak van
trombocyten, stijgt hun trombocytengetal vaak nauwelijks. Bekend is dat geactiveerd endotheel,
zoals dat aanwezig is bij een systemische ontstekingsreactie, trombocyten kan binden zonder dat
er sprake is van DIC.
Op basis van deze observaties werd de hypothese geformuleerd
stijging van het trombocyten bij ernstig zieke patiënten een
sequestratie bij een persisterende systemische ontsteking. Daar
persisterende ontsteking met mortaliteit is, hebben wij het beloop
survivors en non-survivors vergeleken.
dat een afwezige secundaire
uiting is van voortdurende
er een bekende relatie van
van het trombocytengetal bij
In hoofdstuk 8 is naar de relatie gekeken bij patiënten opgenomen op de chirurgische IC. Voor
iedere patiënt werd tussen IC-dag 2 en 10 met regressie-analyse de dagelijkse verandering in het
trombocytengetal (∆PC/∆t) gekwantificeerd. Bij vergelijking van ∆PC/∆t met de gangbare
APACHE-II score, toonde ∆PC/∆t een groter onderscheid tussen survivors en non-survivors. Bij
de survivors bedroeg ∆PC/∆t 30⋅109/l/dag versus slechts 6⋅109/l/dag bij de non-survivors. Dit
fenomeen was niet beperkt tot een bepaalde populatie IC-patiënten: in iedere subgroep vertoonden
de survivors een hogere ∆PC/∆t.
In de literatuur zijn er veel observaties beschreven over de associatie van het laagste
trombocytengetal met mortaliteit, bij een diversiteit van ernstige acute aandoeningen zoals
polytrauma, gebarsten aneurysma van de aorta, malaria, meningitis en septische shock.
In hoofdstuk 9 is de waarde van de het trombocytengetal geverifieerd in een andere groep IC
patiënten. Het betrof ruim 5000 IC-patiënten, zowel medisch als chirurgisch, die waren
onderzocht in een Europese multicenter studie. Bij deze gevarieerde groep werd de relatie van
primaire en secundaire veranderingen in het trombocytengetal met mortaliteit bestudeerd. Uit de
dagelijkse metingen van het trombocytengetal werden voor iedere patiënt drie parameters
berekend: ∆PC/∆t0→2 de dagelijkse verandering in het trombocytengetal tussen IC-dag 0 en 2;
PC2 het trombocytengetal op dag 2 en ∆PC/∆t2→10 de dagelijkse verandering van het
trombocytengetal tussen dag 2 en 10.
PC2 was uitstekend gecorreleerd met het laagste waargenomen trombocytengetal, een maat die
veel auteurs hebben gerelateerd met mortaliteit. Hoewel PC2 enige associatie toonde met
mortaliteit, vertoonde ∆PC/∆t2→10 de sterkste associatie met mortaliteit. De relaties van PC2 en
133
Acute systemische ontsteking bij gezonden en zieken
∆PC/∆t2→10 met mortaliteit waren additief. De relatie van ∆PC/∆t2→10 met mortaliteit werd ook bij
de onderscheiden subgroepen van medische en chirurgische patiënten, evenals kinderen,
waargenomen.
Ten slotte
In hoofdstuk 10 worden de resultaten van de oudere studies (hoofdstukken 2 t/m 5) aan de actuele
literatuur gespiegeld. Met betrekking tot het vetemboliesyndroom moet worden vastgesteld dat er
geen essentieel nieuwe gezichtspunten zijn ten opzichte van 10 jaar geleden. Een echt
representatief diermodel ontbreekt nog steeds. De klinisch manifeste vorm van FES wordt nog
slechts zelden gezien; de therapie is ondersteunend - niet causaal - met een redelijk goede
prognose. De inmiddels gangbare vroege stabilisatie van fracturen heeft waarschijnlijk een
bijdrage geleverd aan de verminderde incidentie van FES.
Wat betreft de interleukines is er erg veel vooruitgang geboekt. De rol van IL-6, TNF-α en IL-1
bij de inductie van systemische ontsteking is uitgebreid in kaart gebracht. Helaas hebben
interventiestudies gericht op het remmen van deze mediatoren nog geen succes opgeleverd.
Naar de rol van PCT is vooral veel klinisch onderzoek in de vorm van spiegel-bepalingen bij
allerlei patiëntengroepen verricht. Er moet nog veel fundamenteel PCT onderzoek verricht
worden. PCT heeft zowel eigenschappen van een acuut fase eiwit als van een cytokine. De
fysiologische functie van PCT is nog onduidelijk. Hoewel wij hebben aangetoond dat de lever
aanzienlijke hoeveelheden PCT kan synthetiseren is de cellulaire bron nog een raadsel.
De studies naar het trombocytengetal hebben aangetoond dat een simpele, goedkope parameter
die op iedere IC dagelijks bepaald wordt, meer informatie in zich herbergt dan veelal is gedacht.
De hypothese dat het beloop van het trombocytengetal een belangrijke indicator is van
systemische endotheel-activatie kan in de komende jaren hopelijk getoetst worden. Het is
interessant te zien wat de effecten van diverse interventies gericht op het raakvlak van stolling en
ontsteking (e.g. antithrombine-III, geactiveerd proteine C, P-selectine remmers) op het
trombocytengetal zullen zijn.
Hopelijk zullen zulke studies bijdragen aan een beter begrip van de processen die de outcome van
ernstig zieke patiënten bepalen.
134
DANKWOORD
Graag zou ik allen die hebben bijgedragen aan dit proefschrift willen bedanken.
In de eerste plaats gaat mijn dank natuurlijk uit naar mijn promotores.
Henk-Jan ten Duis wil ik danken voor zijn nimmer aflatende enthousiasme en inzet voor vele
gevarieerde onderzoeksonderwerpen, en voor zijn vaste optimisme ten aanzien van een goede
afloop.
Maar ook zonder het initiatief, de voortdurende steun en wijze sturing van Hauw The was dit
proefschrift niet tot stand gekomen.
Het is altijd een bijzonder genoegen om over de vele aspecten van ontsteking in het algemeen en
van de bloedplaatjes in het bijzonder met Robert Porte van gedachten te wisselen.
De leescommissie, bestaande uit Jan Goris, Reinout van Schilfgaarde en Bert Thijs, dank ik voor de
moeite die zij hebben genomen bij de beoordeling van het manuscript.
Hans Makkinga en Bart de Smet, jullie kunnen nu tien jaar nadat jullie ervoor gevraagd zijn,
optreden als paranymf en weer eens naar Groningen afreizen.
Erik Hack, Lucien Aarden, en Bert Feltkamp wil ik hartelijk bedanken voor de geboden
gastvrijheid en steun bij het verrichten van onderzoek op het CLB. De sfeer en de
onderzoeksmentaliteit op de afdeling Autoimmuunziekten van het CLB hebben tijdens mijn korte
verblijf op het CLB een onuitwisbare positieve indruk achtergelaten. Bert, ik neem aan dat de
uitvoering van de omslag in ieder geval je instemming kan vinden.
Ik hoop dat ik nog met een aantal van jullie in de toekomst onderzoek kan verrichten.
Lieve Anja, je geduld is erg lang op de proef gesteld, maar nu is het toch eindelijk af.
135
136