Mods Score

Sepsis and Organ Dysfunction
Epidemiology and Scoring Systems

Pathophysiology and Therapy
Springer
Milano
Berlin
Heidelberg
New York
Barcelona
Budapest
Hong Kong
London
Paris
Santa Clara
Singapore
Tokyo
A.E. Bane
G. Berlot
A. Gullo (Eds.)
Sepsis and
Organ Dysfunction
Epidemiology and Scoring Systems
Pathophysiology and Therapy
ORGAN FAILURE ACADEMY
Springer
A.E. BAUEM.D.
Department of Surgery, Saint Louis University, Health Sciences Center, St. Louis - USA
G. BERLOT M.D.
Department of Anaesthesia, Intensive Care and Pain Therapy
University of Trieste, Cattinara Hospital, Trieste - Italy
A. GULLO M.D.
Department of Anaesthesia, Intensive Care and Pain Therapy
O.F.A. - ORGAN FAILURE ACADEMY, VIA BATTISTI, 1 - 34125 TRIESTE (ITALY)

Steering Committee
A.E. Baue M.D., Department of Surgery, Saint Louis University
Health Sciences Center, St. Louis - USA
G. Berlot M.D., Department of Anaesthesia, Intensive Care and Pain Therapy
A. Gullo M.D., Department of Anaesthesia, Intensive Care and Pain Therapy
L. Silvestri M.D., Department of Anaesthesia, Intensive Care and Pain Therapy
G. Sganga M.D., Department of Surgery, and C.N.R. Shock Centre
Catholic University of Sacro Cuore, Rome - Italy
© Springer-Verlag Italia, Milano 1998
ISBN 978-88-470-0297-5 ISBN 978-88-470-2271-3 (eBook)

DOl 10.1007/978-88-470-2271-3
Library of Congress Cataloging-in-Publication Data: Applied for
This work is subject to copyright. All rights are reserved, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, re-use of illustrations, recitation,
broadcasting, reproduction on microfilms or in other ways, and storage in data banks. Duplication of
this publication or parts thereof is only permitted under the provisions of the German Copyright Law of
September 9, 1965, in its current version and permission for use must always be obtained from
Springer-Verlag. Violations are liable for prosecution under the German Copyright Law.
The use of general descriptive names, registered names, trademarks, etc., in this publication does not
imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
Product liability: the publishers cannot guarantee the accuracy of any information about dosage and
application contained in this book. In every individual case the user must check such information by
consulting the relevant literature.
SPIN: 10662618
Sepsis and organ dysfunction are different entities. In leU patients these condi-
tions may appear concomitantly more often than would normally be the case for
other pathophysiological and clilJ,ical manifestations. The development of orga-
nized intensive care units around the world over the past decades has increased
the possibility for survival in patients ranging from the newborn to the oldest
critically ill. Researchers and clinicians have developed a whole set of strategies
for the prevention and management of sepsis and organ failure. Advances in
biotechnology became indispensable to monitor organ function and provide
long-term support in multiple organ dysfunction. The prevention of infections -
whenever possible - is a golden standard. Its implementation, however, though
easy in theory, may come up against some concrete difficulties in practice. So,
Systemic Inflammatory Response Syndrome (SIRS) remains a mysterious condi-
tion, while for sepsis it is still very difficult to reach consensus on its definition,
on the procedures used to perform a correct diagnosis, or the administration of
a selective antibiotic, or the timing of re-laparotomy. Besides, the mortality rate
for patients suffering from sepsis and organ dysfunction remains high. The epi-
demiological aspects and the complexities of sepsis and organ dysfunction are a
well-known phenomenon. The use of prognostic indexes may be an important
tool for patient selection. To date, however, it is not yet clear whether the use of
scores is of actual benefit to individual patients or if it improves therapy. Media-
tors are regarded as vital elements for survival. However, these molecules are
often messengers ofillfate in terms of patient outcome. The problem is to under-
stand how they work, so the enigma remains unsolved. Procalcitonin is one of
the latest mediators to have been proposed as an important marker of infection.
Prostaglandins are a more persistent finding in the bloodstream, so modulat-
ing prostaglandin metabolism in sepsis has provided few answers so far and
many questions remain open. The issue of oxygen delivery optimization in sepsis
is of paramount importance. General consensus has now been achieved con-
cerning the role of inotropic and vasoactive drugs to maintain pressure in the
cardiovascular system. The maintenance of organ peifusion is a different matter,
so oxygen delivery and consumption ratios in sepsis remain controversial. In re-
cent years monoclonal antibodies and receptor blocking agents have represent-
ed important potential advances in the management of sepsis, though unfortu-
VI
nately in many international trials the large amounts of negative data outweigh
the encouraging results obtained in some sub-groups of patients.
So, the high cost of research and the difficulties encountered in enrolling pa-
tients have led to what some, termed the Bermuda Triangle for pharmaceutical
industries. Measuring ipH is a valid tool to monitor gastrointestinal peliusion,
and experimental and clinical data on hepatosplanchnic circulation is very en-
couraging. Systemic Digestive Decontamination (SDD) has proven to be effec-
tive in the prevention of Gram-negative infections. On the other hand, however,
the role of bacterial translocation needs to be confirmed in the clinical setting.
Consumption coagulopathy is a complication leading per se to multiple organ
failure; heparin, aprotinin and antifibrinolytic agents often represent the treat-
ment of choice, even though the mortality rate remains high.
In conclusion, it can be stated that there is no magic recipe for the preven-
tion and treatment of sepsis and organ dysfunction. So it can be said that the
future is likely to be characterized by both favourable and unfavourable devel-
opments. Actually, multiple therapeutic agents offer a series of options, but the
answers for the future will come from a better understanding of the mec'ha-
nisms regulating cellular functions, from the discovery of new mediators able
to prevent apoptosis, as well as the ability to correctly explore immunologic
dissonance.
Monitoring organ functional reserve and understanding messages in their
mutual relationships are the next steps ahead. Some of the tasks outlined abJve
represent a real challenge for researchers and clinicians at the dawn of the
XXlst Century.
Arthur E. Baue
Giorgio Berlot
Antonino Gullo
Trieste. 19th November 1997

Table of Contents
EPIDEMIOLOGY AND CLINICAL COURSE OF SEPSIS AND MODS
Epidemiology and Clinical Course of Sepsis

L. ILKKA, AND J. TAKALA ...•.•.............•......................................................•.•..•.........•........•.......... 15
The Complexities of Sepsis and Organ Dysfunction

A.E. BAUE .................................................................................................................................... 23
SCORING SYSTEMS
An Overview to Introduce Prognostic Indexes in MODS

A.E. BAUE.................................................................................................................................... 35
Multiple Organ Dysfunction (MOD) Score

J.C. MARSHALL............................................................................................................................ 45
The Logistic Organ Dysfunction (LOD) System

J.R. LE GALL, J. KLAR, AND S. LEMESHOW .................................................................................. 53
Will the Use of Scores Benefit the Individual Patient and Improve Therapy?
R. LEFERING, AND R.J.A. GORIS .................................................................................................. 65
MEDIATORS AND PROSTAGLANDINS
Old and New Mediators in Sepsis. The Enigma Is Not Yet Resolved
M. ANTONELLI, AND M. PASSARIELLO .......................................................................................... 73
Modulating Prostaglandin Metabolism in Sepsis

M.R. PINSKY ..........•.••••...•...............••..•••..•.•....•...........•....••••••••......•.•......•.•....•.•..•....•........•..•......• 81
DOlV02 IN SEPSIS
Oxygen Delivery and Consumption Relationships in Sepsis

The Role of Inotropic and Vasoactive Drugs
O. BOyD ....................................................................................................................................... 91
COAGULATION
Disturbances and Modulators of Coagulation and Fibrinolysis in Sepsis

L.G. THus .................................................................................................................................... 99
VIII
EXPERIMENTAL AND CLINICAL APPROACH
Therapies Directed against TNF-a and IL-t during Sepsis

K. REINHART, AND W. KARzAI...................................................................................................... 115
Effects of Adrenergic Agents on the Hepato-Splanchnic Circulation: An Update

D. DE BACKER, AND J.-L. VINCENT •••..••••••.••.••••.•..•.•••••..•..••.•.•.•..•.•.•..•.•.•..•..•....•..•••...•.•.••.•.•.•....•. 123
When to Operate or to Stop Operating and to Plan a Reoperation

A.E. BAUE .....••••.•.•..•...•••••••••..•...........••....•.•.......•..•..•..•..•..•..•.•.....••••••....•.........•••....•....•....•••......•. 131
Multiple Therapeutic Agents - Are We Making Progress?

A.E. BAUE, AND H. REDL •............................................•...............•.•.•.................•.............•.•......... 145
NEUROMUSCULAR DYSFUNCTION
SIRS and Sepsis-Induced Neuromuscular Dysfunctions

C.F. BOLTON ....••••.•.•.........•..•.....•..•......•....•....•....•.....•..••••..•........•..•...•.••.•.....•.............•....•.•.•......•. 155
PREVENTION AND MANAGEMENT OF SEPSIS AND MODS
Lights and Shadows in Sepsis and Multiple Organ Dysfunction Syndrome (MODS)
G. BERLOT, L. SILVESTRI, F. IsCRA, G. SGANGA, AND A. GULLO .......•.......................................... 167
Index ........................................................................................................................................... 189

Authors Index
Adriaensen H.
Dept. of Anaesthesiology, University Hospital Antwerp, Edegem (Belgium)
Antonelli M.
Dept. of Anaesthesiology and Intensive Care, La Sapienza University, Rome (Italy)
BaueA.E.
Dept. of Surgery, St. Louis University Health Sciences Center and the Veterans Administration Medical
Center, St. Louis (U.S.A.)
Berlot G.
Dept. of Anaesthesiology and Intensive Care, Trieste University School of Medicine, Trieste (Italy)
Bolton C.F.
London Health Sciences Centre, Victoria Campus, University of Western Ontario, London (Canada)
BoydO.
Dept. of Anaesthesia, St. George's Hospital, London (U.K.)
De BaekerD.
Dept. of Intensive Care, Free University of Bruxelles, Erasme Hospital, Bruxelles (Belgium)
Goris R.J.A.
Dept. of Surgery, Nijmegen St. Radboud University Hospital, Nijmegen (The Netherlands)
Gullo A.
llkkaL.
Division of Intensive Care, Dept. of Anaesthesiology and Intensive Care, Kuopio University Hospital,
Kuopio (Finland)
Isera F.
Karzai W.
Dept. of Anaesthesiology and Intensive Care Therapy, Friedrich Schiller University Hospital, lena (Germany)
Klar J.
Dept. of Intensive Care, St. Louis Hospital, Pmis (France)
Le Gall J.R.
Dept. of Intensive Care, St. Louis Hospital, Paris (France)
Lefering R.
Biochemical and Experimental Division, 2nd Dept. of Surgery, KOin University, Koln (Gernlany)
LemeshowS.
Dept. of Intensive Care, St. Louis Hospital, Paris (France)
x
Marshall J.e.
Dept. of Surgery, University of Toronto, Toronto (Canada)
Passariello M.
Dept. of Anaesthesiology and Intensive Care, La Sapienza University, Rome (Italy)
PinskyM.R.
Dept. of Anesthesiology and Critical Care Medicine, University of Pittsburgh, Pittsburgh (U.S.A.)
RedlH.
Ludwig Boltzmann Institute for Experimental and Clinical Traumatology, Vienna (Austria)
ReinhartK.
Dept. of Anaesthesiology and Intensive Care Therapy, Friedrich Schiller University Hospital, Jena (Germany)
SgangaG.
Dept. of Surgery, Catholic University of Sacro Cuore, Rome (Italy)
Silvestri L.
TakalaJ.
Division of Intensive Care, Dept. of Anaesthesiology and Intensive Care, Kuopio University Hospital,
Kuopio (Finland)
Thijs L.G.
Medical Intensive Care Unit, University Hospital, Amsterdam (The Netherlands)
Vincent J.-L.
Dept. of Intensive Care, Free University of Bruxelles, Erasme Hospital, Bruxelles (Belgium)
Abbreviations
AIS, abbreviated injury score HTI, hospital trauma index

AOSFSS, acute organ system failure scoring ICAMt, intracellular adhesion molecule I
system ICU, intensive care unit
APACHE, acute physiology and chronic IL-t, interleukin-l
health evaluation system
IL-Ira, interleukin-l receptor antagonist
ARDS, acute respiratory distress syndrome
IL-6, interleukin-6
ARF, acute renal failure
IL-S, interleukin-8
ASCOT, a severity characterization of trauma
lIS, injury impairment score
ATI, abdominal trauma index
ISS, injury severity score
ATIII, antithrombin III
LOD, logistic organ dysfunction
BI, bum index
LODS, logistic organ dysfunction system
BSS, Baltimore sepsis scale
LSI, limb salvage index
CARE, clinical assessment, research and edu- MAP, mean arterial pressure
cation system
MARS, mixed antagonistic response
CARS, compensatory anti-inflammatory syndrome
response syndrome
MESS, mangled extremity severity score
CCSS, critical care scoring system
MIP, macrophages inflammatory protein
CHAOS, cardiovascular compromise,
homeostasis, apoptosis, organ dysfunc- MISGS, medical illness severity grouping
tion, suppression of immune system system
CIRS, cumulative index rating scale MOD, multiple organ dysfunction
CIS, cell injury score of Hirasawa MODS, multi-organ dysfunction system
CRAMS, circulation, respiration, abdomen, MOF, multiple organ failure

motor, speech MPS, Mannheim peritonitis score
CVP, central venous pressure MSOF, multiple system organ failure
DEC, diethylcarbamazine MTO, major trauma outcome study
DIC, disseminated intravascular coagulation ODIN, organ dysfunction and/or infection
DPE, daily prognostic estimates OIS, organ injury scales
ENAS, European North American study OSF, organ systems failure
EPI, endocrine prognostic index PAF, platelet activating factor
FCI, functional capacity index PAP, plasmin-antiplasmin
GCS, Glasgow coma score PC, protein C
G-CSF, granulocyte colony stimulating factor PNI, prognostic nutritional index
GM-CSF, granulocyte macrophage-colony PODS, probability of death score

stimulating factor PRISM, pediatric risk of mortality
HPI, hospital prognostic index PSI, predictive salvage index
HR, heart rate PT, prothrombin time
XII
PTS, poly trauma score TAT, thrombin-antithrombin

ROC, receiver operating characteristic TFPI, tissue factor pathway inhibitor
SAPS, simplified acute physiology score TISS, therapeutic intervention scoring system
SCOUT, surgical complication outcome TNF-0'., tumor necrosis factor alpha
Sh02, hepatic venous oxygen saturation t-PA, tissue-type plasminogen activator
SIRS, systemic inflammatory response TRISS, trauma and injury severity score
syndrome
TS, trauma score
SMART, systemic mediator-associated
u-PA, urokinase plasminogen activator
response test
V0 2, oxygen consumption
SOFA, sepsis organ failure assessment
VS, ventilator score
STAR, staged abdominal repair
abdominostomy WBC, white blood cell
EPIDEMIOLOGY AND CLINICAL COURSE
OF SEPSIS AND MODS
Epidemiology and Clinical Course of Sepsis
L. ILKKA, J. TAKALA
Sepsis can be broadly defined as the systemic inflammatory response to infec-

tion. Sepsis is an important cause of morbidity and mortality in hospitalized
patients. In intensive care patients, sepsis is one of the most common causes of
prolonged intensive care and death. Accordingly, a sizable proportion of inten-
sive care resources is used to treat patients with sepsis and sepsis-related dys-
function of vital organs. Despite this, the epidemiology and clinical course of
sepsis has not been well defined. Lack of uniform definitions and criteria for
sepsis has certainly contributed to this lack of information.
In the past, the term "sepsis" has been used quite loosely. It has covered bac-
teremia, clinical conditions consistent with infection, and verified local or sys-
temic infections with variable sets of biochemical and clinical signs and symp-
toms (Table 1) [1-13]. It is self-evident that the selection of the criteria will have
a major impact on any epidemiological and clinical data. The various definitions
of sepsis have gradually converged to focus on the clinically obvious link
between serious infection, and its systemic manifestations, especially organ dys-
function.
Verification of infection is usually regarded as a fundamental criterion.
Because of the problems associated with positive cultures, especially bac-
teremia, new radiological findings consistent with infections as well as demon-
stration of gross infection foci have also been used. It is unlikely that one single
definition would be practical for all purposes. For daily clinical routine, a low
threshold of suspicion and aggressive search for infection foci is vital. On the
other hand, therapeutic trials often benefit from much more rigorous criteria.
Every clinician is familiar with "suspected sepsis", where cultures are nega-
tive but the clinical signs and symptoms are consistent with the response to
infection. These patients are treated without great problems of definitions. In
contrast, for research purposes this term is problematic. It is too wide, and can
include very heterogeneous patient population, and is heavily influenced by dif-
ferences between clinicians. A systemic response closely resembling or even
identical to sepsis may also be induced by non-infectious factors, including
ischemia, major bums, or pancreatitis. For the purpose of this review, we use the
term sepsis to indicate the systemic response to infection.
16 L. Ilkka, J. Takala
Table 1. Examples of various definitions or criteria for diagnosis of sepsis
Bacteremia [1, 2]
Documented by positive blood or body fluid cultures, or unmistakable evidence of a septic
process [3]
Host response to a microbiological event (induced by the presence of bactelia, viruses, fun-
gi...) [4]
A serious infection and the systemic response to infection [5]
The systemic response to infection. This systemic response is manifested by two or more of the
following conditions as a result of infection: 1) temperature> 38°C or < 36 °C, 2) heart rate>
90 beats/min, 3) respiratory rate> 20 breaths/min or PaC0 2 < 32 torr / < 4.3 kPa, and 4) white
blood cell count> 12000 cells/mm3, < 4000 cells/mm3, or > 10% immature (band) forms [6]
Septicemia = systemic disease associated with the presence and persistence of pathogenic mi-
croorganisms or their toxins in the blood [7]
Clinical evidence suggestive of infection plus signs of a systemic. response to the infection (all of
the following): tachypnea (> 20 breaths/min, or if mechanically ventilated, > 10 Llmin),
tachycardia (> 90 beats/min); hyperthermia or hypothemlia (core or rectal temperature> 38.4
°C or < 35.6 0c) [8]
The presence of various pus-forming or other pathogenic organisms and/or their 'toxins' in the
blood or tissues [9]
A systemic response to infection and inflammation that is usually characterized by a toxic clinical
picture, including fever or hypothermia, tachycardia, tachypnea, and mental obtundation [10]
A clinical response characterized by alterations in one or more of temperature, white blood cell
count, and mentation in association with a hyperdynamic hypermetabolic state [II]
Fever or hypothermia (temperature> 38.3 °C or < 35.6 0C); tachycardia (> 90 beats/min in the ab-
sence of beta-blockade) and tachypnea (respiratory rate> 20 breaths/min or the requirement
of mechanical ventilation); and either hypotension (systolic blood pressure :0:; 90 mmHg or a
sustained drop in systolic pressure;:::: 40 mmHg in the presence of an adequate fluid challenge
and the absence of antihypertensive agents) or two of the following six signs of systemic toxi-
city or peripheral hypoperfusion: unexplained metabolic acidosis (pH:O:; 7.3, base deficit of> 5
mmollL, or an elevated plasma lactate level); at1erial hypoxemia (Pa0 2 :0:; 75 mmHg or Pa0 2 /
FI0 2 < 250); acute renal failure (urinary output of less than 0.5 ml/kg/hour; elevated pro-
thrombin or partial-thromboplastin time or reduction of the platelet count to less than half the
baseline value or less than 100 000 platelets/mm 3 ; sudden decrease in mental acuity; and car-
diac index of more than 4 Llminlm 2 of body surface area with systemic vascular resistance of
less than 800 dyn . sec· cm- 5 [12]
Known or suspected (gram-negative in this study) infection plus at least one of these signs of sep-
sis: 1) fever or hypothermia: > 38.2 °C or < 36.5 °C, 2) tachycardia: > 90 beats/min, 3) tachyp-
nea: > 20 breaths/min plus at least one of these signs of organ dysfunctions: 1) hypoxemia:
Pa02 / FI0 2 :0:; 280, 2) increased serum lactate concentration: above normal value for laborato-
ry, 3) oliguria: < 0.5 mL/kg of body weightlhr, 4) altered mentation: Glasgow Coma Scale <
15, or decrease;:::: 1,5) new coagulopathy: unexplained [13]
For standardizing the terminology, a North American consensus conference

published in 1992 the recommendations for the definitions of sepsis and its
sequelae [6]. Sepsis was defined as systemic inflammatory response in the pres-
ence of infection, with two to four clinical manifestations (Table 2) [6]. The
term "systemic inflammatory response syndrome" (SIRS) was proposed to be
Epidemiology and Clinical Course of Sepsis 17
Table 2. The definitions of ACCP/SCCM consensus conference
Infection Microbial phenomenon characterized by an inflammatory response to the

presence of microorganisms or the invasion of normally sterile host tissue
by those organisms
Bacteremia The presence of viable bacteria in the blood
SIRS The systemic inflammatory response to a variety of several clinical insults
The response is manifested by two or more of the following conditions:
temperature> 38°C or < 36 °C, heart rate> 90 beats/min, respiratory rate
> 20 breaths/min or PaC02 < 32 torr « 4.3 kPa), and WBC > 12 000
cells/mm3 or < 4000 cells/mm3 or > 10% immature (band) forms
Sepsis The systemic response to infection. This systemic response is manifested
by two or more of the following conditions as a result of infection: tem-
perature > 38°C or < 36 °C, heart rate > 90 beats/min, respiratory rate >
20 breaths/min or PaC02 < 32 torr « 4.3 kPa), and WBC > 12 000
cells/mm3 or < 4000 cells/mm3 or > 10% immature (band) forms
Severe sepsis Sepsis associated with organ dysfunction, hypoperfusion, or hypotension.
Hypoperfusion and hypotension abnormalities may include, but are not
limited to, lactic acidosis, oliguria, or acute alteration in mental status
Septic shock Sepsis with hypotension despite adequate fluid resuscitation, along with
presence of perfusion abnormalities that may include, but are not limited
to, lactic acidosis, oliguria, or acute alteration in mental status. Patients
who are on inotropic or vasopressor agents may not be hypotensive at the
time that perfusion abnormalities are measured
Hypotension A systolic blood pressure < 90 mmHg or a reduction> 40 mmHg from
baseline in the absence of other causes of hypotension
Multiple organ Presence of altered organ function in an acutely ill patient such that home-
dysfunction syndrome ostasis cannot be maintained without intervention
used in cases of an evident systemic response in the absence of a verified infec-

tion. As in the definition of sepsis, the criteria of SIRS include alterations in two
to four parameters: temperature, heart rate, respiratory rate and leucocyte level.
In later epidemiological studies using these consensus conference definitions
it has been clearly shown that SIRS is a too sensitive and nonspecific tool for
research and practice purposes [4, 14, 15]. The vast majority of leU patients
and even hospitalized patients fullfil the criteria of SIRS in some phases of the
hospitalization. The fullfilment of these criteria alone appears to be without
clinical relevance and have no value for guiding the decisions of treatment or
predicting outcome of patients.
Still, some SIRS patients are progressing further to sepsis or even septic
shock states. The problem is how to select the right patients from large popula-
tion. It is apparent that these rather arbitrary criteria need to be considered with
other criteria, for example for including patients to prospective clinical or epi-
demiologic studies.
18 L. llkka, J. Takala
Epidemiology
The epidemiology and especially the clinical course of sepsis are not well
known. The number of prospective epidemiologic studies with sufficient patient
samples is sparse. Two larger European prospective epidemiologic studies of
SIRS and sepsis-related states using ACCP/SCCM-definitions were published in
1995. Rangel-Frausto et al. studied 3700 patients admitted to three critical care
units and three wards [14]. SIRS seemed common about 20% patients pro-
gressed to septic shock. The rates of positive blood cultures, end-organ dysfunc-
tions, and mortality were related to the severity of the systemic inflammatory
response. In the Italian sepsis study with 1100 patients SIRS was again a usual
phenomenon at the time of admission [15]. As well as in the former publication,
in this study also the mortality seemed to be related to the severity of sepsis, but
no prognostic differences were observed between SIRS patients and patients
without SIRS.
Two larger prospective studies evaluated the prevalence and incidence of
ICU-acquired infections and sepsis-related states. A I-day cross-sectional preva-
lence study of about 10,000 patients showed nearly half of the patients being
infected during their ICU stay [16]. Another multicenter study focusing on
severe sepsis showed that this state occurred in nearly 10% ofICU admissions,
and that only three of four patients presenting clinical sepsis had documented
infection [17].
It is well known and established that ICU patients are at increased risk for
nosocomial infections. Less than 10% of hospitalized patients are treated in
ICUs, but patients needing intensive care have about one fourth of all nosocomi-
al infections [18]. Hospitalized patients have prevalence of nosocomial infection
between 5 and 17% [16]. The prevalence of infections among intensive care
patients has been reported to vary from 15 to 40% [19]. The risk of acquiring
infection increases with the stay in the ICU and with the use of invasive devices.
The type of pathogens responsible for infections in ICU have changed during
last decades [18]. In the 1960s and 1970s Gram-negative pathogens were pre-
dominantly causative. Later Gram-positive bacteria have become increasingly
responsible. Especially the proportion of coagulase-negative staphylococci,
methicillin-resistant Staphylococcus aureus, and enterococci has been increas-
ing worldwide. Gram-positive and Gram-negative bacteria cover infections in
roughly equal proportions. The rate of anaerobic bacteremias have decreased,
with Bacteroides fragilis being the most common [20]. The amount of fungal
infections have been greatly rising, especially caused by Candida species [21].
Candidemias account for under 10% of positive blood cultures, but the mortality
rate in these cases is over 50%.
Also the profile of infection foci has changed during past decades [18]. It
differs between ICUs and general wards. In the former respiratory tract infec-
tions are the most usual, whereas in the latter urinary tract infections predomi-
nate. The rate of blood stream infections have increased 5 to 10 fold compared
to earlier decades, covering now more than 10% of infections in hospitals, and
15% of leu infections. More than 40% of these infections are associated with
catheters. Pneumonia and other respiratory tract infections are the leading foci
in infected leu patients, covering about half of infections. Urinary tract is the
source in about 10-20%. Other foci, such as wound, upper airway, gastrointesti-
nal or central nervous system infections account for about one fourth of infec-
tions [16-18].
Mortality is related to the severity of sepsis, the type of infection and bacteri-
al etiology. The crude mortality of infected patients varies greatly, from 10 to
80%, depending on the type of the leu and definitions of sepsis used [19].
Rangel-Frausto et al. showed the mortality rates to increase in the hierarchy
from SIRS and sepsis to severe sepsis to septic shock [14]. SIRS-patients had
one month mortality of less than 10%. Half of the patients progressing to septic
shock died. This is consistent with reports in the literature.
Pneumonia and bacteremia as a cause of sepsis are increasing mortality
about two to three fold [16, 22]. Late-onset pneumonias with high-risk
pathogens, such as Acinetobacter or Pseudomonas species, increase the risk to
several fold [22]. Multiresistant organisms, for example vancomycin-resistant
enterococci or methicillin-resistant Staphylococcus aureus contribute also to
greatly increased risk of death. Gram-negative organisms as a whole group and
coagulase-negative staphylococci seem to carry a lesser risk than other organ-
isms. Multiple sources of sepsis is a phenomenon clearly associated with poor
prognosis [17].
Clinical course of sepsis

The clinical progress of sepsis is poorly known. The duration of leu stay and
hospitalization, and incidence of organ failures progressing after sepsis are bet-
ter established. Infections prolong the hospitalization time by several days. Sev-
eral matched case-control studies assessing the extra length of hospitalization
due to infection have been published [22]. Dependent on the type of infection,
the excess length of stay has varied among leu patients between one and two
weeks. In a study of bacteremic leu patients the median length of hospitaliza-
tion was 40 days compared to 26 days with matched control patients without
infections [22]. In sepsis of other foci the excess lenght of stay has been report-
ed to be around one week.
In the study of Rangel-Frausto et al. 56% of sepsis patients were septic
immediately on the day of admission. Similarly 42% of patients in severe sepsis
and 29% of septic shock patients were in their categories already during the day
of admission. The median intervals from one category to another were variable.
The interval times with culture-negative patients were similar to that of patients
with confirmed infection. The 28-day mortality in this study was 9%, but an
20 L. llkka, J. Takala
additional proportion of patients died within six months of discharge from ICU.
In the study of 1052 patients in severe sepsis, the median length of ICU stay
was over one week [17]. It greatly differed from survivors to nonsurvivors, 34
days and 4 days, respectively. One third of patients stayed in the ICU more than
two weeks.
End-organ dysfunctions are common in sepsis. In the past, multiple organ
failure (MOF) was thought to be caused by uncontrolled or undiagnosed infec-
tion [11]. Later it has become clear that organ dysfunctions can develop and
progress in the absence of uncontrolled infection, and the treatment of the infec-
tion may fail to prevent the development of MOE In the study of Rangel-Fraus-
to et al. end-organ failure rates increased with the number of SIRS criteria [14].
The attack rates of acute respiratory distress syndrome (ARDS), disseminated
intravascular coagulation (DIC), acute renal failure (ARF), and shock were quite
similar in SIRS and sepsis patients (about 4, 16, 13 and 21%, respectively).
Patients progressing to severe sepsis and septic shock had increasing rates of
these dysfunctions in each stage (respectively 6, 18,20, and 25% in severe sep-
sis, and 18, 38,45, and naturally 100% in septic shock) (Table 3).
Table 3. Attack rates for end-organ dysfunction
Syndrome
No. of ARDS DIe ARF Shock
patients (%) (%) (%) (%)
SIRS with 2 criteria 2 8 9 11
SIRS with 3 criteria 3 15 13 21
SIRS with 4 criteria Total in SIRS 2527 6 19 19 27
Positive culture sepsis 649 6* 16 * 19 * 20 *
Negative culture sepsis 892 3 20 5 27
Severe sepsis with positive cultures 467 8 18 23 * 28 *
Severe sepsis with negative cultures 527 4 17 16 22
Septic shock with positive cultures 110 18 38 51 100
Septic shock with negative cultures 84 18 38 38 100
• = p value < 0.05 between culture positive and culture negative stages.
(Modified from [14])
In summary, sepsis is an important reason for morbidity and mortality in

ICU patients. Despite the vast amount of resources used for its treatment and in
the search of new therapies, the epidemiology and clinical course of sepsis
remain poorly documented. In order to better understand the pathophysiology
and the develop of new therapies, a more defined description of this syndrome
and its clinical course is needed.
References
1. Wiles JB, Cerra FB, Siegel JH, Border JR (1980) The systemic septic response: does the
organism matter? Crit Care Med 8:55-60
2. Parker MM, Shelhamer JH, Natanson C et al (1987) Serial cardiovascular variables in sur-
vivors and nonsurvivors of human septic shock: Heart rate as an early predictor of prognosis.
Crit Care Med 15:923-929
3. Shoemaker WC, Appel PL, Kram HB et al (1993) Sequence of physiologic patterns in surgical
septic shock. Crit Care Med 21:1876-1889
4. Vincent J-L (1997) Dear SIRS, I'm sorry to say that I don't like you ... Crit Care Med 25:
372-374
5. Parrillo IE (1993) Pathogenetic mechanisms of septic shock. New Engl J Med 328:1471-1477
6. American College of Chest Physicians I Society of Critical Care Medicine Consensus Confer-
ence (1992) Definitions for sepsis and organ failure and guidelines for the use of innovative
therapies in sepsis. Crit Care Med 20:864-874
7. Increase in national hospital discharge survey rates for septicemia - United States 1979-1987.
Morb Mort Weekly Rep 39:31-34 .
8. Bone RC (1991) The pathogenesis of sepsis. Ann Intern Med 115:457-469
9. Bone RC (1991) Sepsis, the sepsis syndrome, multi-organ failure. A plea for comparable defi-
nitions. Ann Intern Med 114:332-333
10. Rackow EC, Astiz ME, Weil MH (1988) Cellular oxygen metabolism during sepsis and shock.
The relationship of oxygen consumption to oxygen delivery. JAMA 259: 1989-1993
11. Marshall JC, Sweeney (1990) Microbial infection and the septic response in critical surgical
illness. Sepsis, not infection, determines outcome. Arch Surg 125: 17 -25
12. Ziegler EJ, Fisher CJ, Sprung CL et a1 (1991) Treatment of Gram-negative bacteremia and
septic shock with HA-1A human monoclonal antibody against endotoxin. A randomized, dou-
ble-blinded, placebo-controlled trial. New Engl J Med 324:429-436
13. Bone RC, Balk RA, Fein AM et al (1995) A second large controlled clinical study of E5, a
monoclonal antibody to endotoxin: Results of a prospective, multicenter, randomized, con-
trolled trial. Crit Care Med 23:994-1006
14. Rangel-Frausto MS, Pittet D, Costigan M et al (1995) The natural history of the systemic
inflammatory response (SIRS). A prospective study. JAMA 273: 117-123
15. Salvo I, de Cian W, Musicco M et al (1995) The Italian sepsis study: Preliminary results on the
incidence and evolution of SIRS, sepsis, severe sepsis and septic shock. Int Care Med 21:
S244-249
16. Vincent J-L, Bihari DJ, Suter PM et al (1995) The prevalence of nosocomial infection-in inten-
sive care units in Europe. Results of the European prevalence of infection in intensive care
(EPIC) study. JAMA 274:639-644
17. Brun-Buisson C, Doyon F, Carlet J et al (1995) Incidence, risk factors, and outcome of severe
sepsis and septic shock in adults. A multicenter prospective study in intensive care units.
JAMA 274:968-974
18. Wildmer A (1994) Infection control and prevention strategies in the ICU. Int Care Med 20:
S7-S11
19. Pittet D, Brun-Buisson C (1996) Nosocomial infections and the intensivist. Curr Opin Crit
Care 2:345-346
20. Goldstein EJC (1996) Anaerobic bacteremia. Clin Inf Dis 23:S97 -S 101
21. Pfaller M, Wenzel R (1992) Impact of the changing epidemiology of fungal infections in the
1990s. Eur J Clin Microbiol Infect Dis 11:287-291
22. Girou E, Brun-Buisson C (1996) Morbidity, mortality, and the cost of nosocomial infections in
critical care. Curr Opin Crit Care 2:347-351
fhe Complexities of Sepsis and Organ Dysfunction
A.E. BADE
Inflammation in itself is not to be considered as a

disease ... and in disease, where it can alter
the diseased mode of action, it likewise
leads to a cure; but where it cannot accomplish
that solitary purpose ... it does mischief
John Hunter: A Treatise on the Blood,
Inflammation, and Gunshot Wounds
London, 1794 [1]
I am pleased to join Professor Gullo in welcoming all of you to the Sixth Organ
Failure Academy Meeting here in Trieste, Italy, and to the program today, "An
update on the pathophysiology of and potential therapy for sepsis and organ
dysfunction".
Much has happened since the Organ Failure Academy was established by
Professor Gullo. We have learned many things and gradually patient care has
improved. The expectation that there would be magic bullets to solve many of
our problems has not been fulfilled [2]; thus, we must meet today comparing
notes on what we know and what we must learn. The power of molecular biolo-
gy to provide information and solve problems has yet to meet its full potential.
We look forward to hearing from all of you and to have important questions
raised by those of you in the audience.
There are many complexities concerning sepsis and organ dysfunction and
there is also considerable confusion about them. There are six complexities
which I will comment upon in this brief introduction. Some of my questions and
the complexities related to them will be reviewed by many authors in the pre-
sent book. These six are shown in Table 1. I will review each of these in some
detail.
Review of complexities
Sepsis
Although we have a general idea as to what it is included with sepsis, the word
means many different things to different individuals. Is it due to infection or is it
24 A.E. Baue
Table 1. Complexities (and confusion) about...
Sepsis - What is it? And how do you treat it?

Terminology - Acronymania - Does this help us?
Inflammation - Is it an organized sequential system which can be blocked, stimulated or
modulated?
Pro- and Anti-inflammatory mediators - Can we time their activities so as to intervene?
There are many biologic conundrums (puzzles) or paradoxes
The causes of organ dysfunction and failure are complex
not? What is the organism? How do we define bacteremia and how is it different
from septicemia? What is the sepsis syndrome? Roger Bone went to great
lengths to try to get us to agree on terminology about the sepsis syndrome,
believing that it was either due to infection or to an inflammatory reaction; how-
ever, there was never general agreement about this [3]. Severe sepsis is another
matter and then, of course, there is septic shock. If there is an inflammatory
process, what is it from? What causes it and how does it produce its problems?
Tissue injury, whether due to trauma or to a planned operation or therapeutic
event, requires inflammation in order to heal. Can there be too much inflamma-
tion? Can overwhelming inflammation be harmful and produce, not only toxici-
ty, but actual death? In other words, do we self destruct if the inflammatory
process and the injury is overwhelming [4]?
Finally, how do we treat all of these? Can we get away from specific diseases
and specific infections with specific cultures of abscesses and of the blood to the
concept of treating general phenomena such as SIRS and MODS? This is a big
and difficult question.
Complexity of terminology
There seems to be continuous development of new terminology and we must
question whether this has helped us. I mentioned earlier the attempt of Roger
Bone to get everyone to agree on the sepsis syndrome but this did not happen.
Then the concept of SIRS and MODS was developed even though MOF seemed
perfectly acceptable earlier [5], The definitions of SIRS and MODS and MOF
are known to all of you and to the reader; however, these terms or expressions
are not treatable. The patients having these expressions of different diseases can
be treated for the disease but not the manifestations alone. These are constructs,
or definitions of being sick; thus, there is no therapy for SIRS or MODS or
MOF other than to try to prevent SIRS from becoming MODS and that from
becoming MOP. How would you treat SIRS? Therapy of each aspect of SIRS is
shown in Table 2. Obviously this is ridiculous. Vincent raised serious questions
about the concept of SIRS in an editorial entitled "Dear SIRS, I'm sorry to say
that I don't like you" [6]. In this article, he states his belief that SIRS is too sen-
The Complexities of Sepsis and Organ Dysfunction 25
sitive and does not help us understand the pathophysiology and does not help in
clinical trials or in practice. He concludes by stating: "Dear SIRS, I'm afraid we
don't need you". Now Bone has made a new proposal and that is the use of
CARS, MARS and CHAOS [7]. This suggests to me a form of acronymania.
Bone's proposal for CARS, MARS and CHAOS is shown in Table 3. Perhaps
Bone stands for biologically occult but natural explanations*.
Table 2. Early treatment of SIRS
Temperature
a) If low, warm patient with warming blankets, extracorporeal warming circuit, warm i. v. fluids,
irrigate peritoneal cavity with warm saline
b) If high, use external cooling, cool bath, rectal aspirin, etc
Rapid heart rate
Slow with calcium channel blockade, rapid acting digitalis, etc
Rapid breathing
i.v. morphine, other sedation, intubation, paralysis, and controlled ventilation
White blood count
a) For a low WBC - give G-CSF
b) For a high WBC - consider chemotherapeutic agents to control bone marrow
Table 3. MOF, MODS, SIRS and now ...
CARS Compensatory, antiinflammatory response syndrome

MARS Mixed antiinflammatory response syndrome
CHAOS Cardiovascular compromise (shock)
Homeostasis
Apoptosis
Organ dysfunction
Suppression of the immune response
BONE Biologically occult but natural explanations
Of what therapeutic benefit are these terms?
What will Bone/they think of next?
My question is: Of what therapeutic benefit are these terms? What will they
think of next? How much longer must we invent new terminology which does
not help with therapy and perhaps does not even help with our understanding of
these disease processes?
Finally, I would define acronymania as shown in Table 4. The likelihood of
being able to lump together a lot of diverse infectious and injurious processes
* We are all sorry to learn of the untimely death of Roger Bone in the spring of 1997 after a long
illness. We will miss his leadership.
26 A.E. Baue
under some generalized description is not going to allow us to treat patients

more effectively. I have made a plea for breaking down all of these things into
specific disease processes so that we can treat a disease rather than a construct
or an acronym.
Table 4. Acronymania
Basic antiinflammatory response (BARS)

Compensating antiinflammatory response (CARS)
Functional antiinflammatory response (PARS)
Mixed antiinflammatory response (MARS)
Whole antiinflammatory response (WARS)
All of these may lead to CHAOS
Inflammation
"Natural forces are the healers of disease". Hippocrates, 460 B.C. Epidemics VI, VI
All of the problems of injury, operation, infection, sepsis, or inflammatory dis-
eases require an inflammatory process to overcome the illness, if possible. John
Hunter described the reactions to injury as "inducing both the disposition and
the means of cure" [I]. As mentioned earlier, if the inflammatory process
becomes excessive, it is believed now that this can be self destructive as well.
However, the problem is that sepsis and inflammation may not be an organized
sequential system. Those that have studied it extensively or actually observed it
believe that there are many vagaries and it is not a process that can be under-
stood well because of its variation on the theme. As Lewis Thomas stated [8],
"First, I would like to construct the straw man that I shall need to demolish
before getting on. Nobody really believes it, but let us pretend that it is the gen-
eral belief that inflammation really exists as an entity among biologic mecha-
nisms, that represents an orderly sequence of time and coordinated events,
staged to occur in such a way that the host is protected against a foreign adver-
sary and able to minimize damage to his own tissues, kill off the adversary and
finally tidy up the place and make whatever repairs are necessary. This is my
straw man. I begin by saying that there really is no such mechanism.
I suspect that the host is caught up in mistaken, inappropriate and unques-
tionably self-destructive mechanisms by the very multiplicity of defenses avail-
able to him which do not seem to have been designed to operate in net coordina-
tion with each other. The end result is not defense; it is an agitated, committee-
directed harum-scarum effort to make war, with results that are markedly like
those sometimes observed in human affairs when war-making institutions pre-
tend to be engaged in defense".
The question then is - If inflammation is necessary for survival, can it be
controlled if it is excessive? Can it be blocked, stimulated or modulated and are
multiple agents required to do that? So far, single agents have been quite unsuc-
cessful.
Mediators
We now know about any number of pro inflammatory mediators. We more
recently have learned about antiinflammatory mediators. There are a multitude
of mediators in both systems. There is natural control of the inflammatory
process. There are multiple enzyme cascades which serve biologic functions. As
we learn more about them and, as more mediators are discovered, we learn that
it is a very complex system that defies sequential change and modulation. I have
called this a modem "Horror Autotoxicus" [4], a term borrowed from Ehrlich
and Morgenroth who predicted in 1905 that autoimmune reactions would be a
"Horror Autotoxicus" [9]. Can we control nature? Can we control a necessary
process that becomes excessive? When does it become excessive and how? How
do we get it back to baseline without getting the patient or the animal into more
difficulty?
Biologic conundrums - Puzzles or paradoxes?

There are a number of circumstances that we now know about where there are
puzzles or paradoxes or conundrums in the information about various mediators.
Too much of a certain mediator is bad but too little is a disaster. One can prepare
a long list of such factors where a substance, on one hand, is helpful but, if
given in another circumstance or in another organ system, creates a disaster
[10]. Can we hope to understand these conundrums and deal with them? Cer-
tainly we can strive toward that. However, again, the complexity of these reac-
tions is most impressive. Can we then work in between, for example, can we
control tumor necrosis factor? If an animal is given an injection of endotoxin or
gram-negative bacteria, or if a small dose of endotoxin is given to a normal
human volunteer, there are deleterious consequences which are mediated in
good part by endotoxin which activates tumor necrosis factor (TNF) [11, 12]. In
that circumstance, if one is to receive a bolus of endotoxin or bacteria, a mono-
clonal antibody to TNF would be protective so long as it was given before the
insult [13]. How does one know that one is about to receive a bolus infusion or
injection of endotoxin? In contrast, if a monoclonal antibody is given to an ani-
mal and then the animal is given a clinically relevant form of peritonitis, the
mortality is much higher in those animals that received the monoclonal antibody
to TNF [14]. How can we have it both ways? If, in one circumstance, TNF is
necessary but, in another, it is a disaster, how can we decide beforehand what
the circumstance of that patient is? There are many such biologic conundrums,
which must be better understood, some of which are shown in Table 5.
28 A.E. Baue
Causes of organ dysfunctionlfailure

There are many causes or reasons why remote organs away from the site of the
injury or away from the site of the insult or infection may become dysfunctional
or fail (Table 6). Obviously, with trauma, there may be pulmonary contusions
resulting in pulmonary failure or dysfunction, and the same with the myocardi-
um, the gastrointestinal tract, bones, muscles, metabolism and other factors.
However, a number of problems produce remote organ dysfunction or failure
when they occur. First among these is ischemia and dysoxia. Where there has
been a prolonged period of ischemia, such as with a low cardiac output, organ
damage occurs. Ischemiaireperfusion will also do this when there has been a
period of ischemia such as ischemia of the lower extremities during repair of an
abdominal aortic aneurysm. With reperfusion, metabolites and other substances
or mediators from the area of ischemia return to the central circulation and dam-
age the lungs and perhaps other organs. There is white cell activation with
injury where oxidants are produced which may damage adjacent cells, particu-
larly endothelial cells, and proteases such as elastase are activated which may
damage tissues. There may be failure of antioxidants where they are used up or
are not available. This also can be a cause of remote organ dysfunction.
In our early description of multiple organ failure after injury we described a
one hit phenomenon or insult where a patient with a severe injury developed
multiple organ failure initially in the first few days, or a second hit when a
patient was resuscitated and seemed to be getting along and then later got into
difficulty [15]. The first hit was injury and the second often seemed to be related
to infection. This two-hit phenomenon has been confirmed by Moore et a1. [16].
Then we learned about overwhelming inflammation as a cause of organ failure
in what I have called a modem horror autotoxicus [4] after the original descrip-
tion from Ehrlich and Morgenroth [9].
Lipopolysaccharide or endotoxin from gram-negative organisms can also
wreak havoc with an individual as can gram-positive exotoxins and other infec-
tious organisms. Bacterial translocation may be a cause of difficulty, although
this has not been documented in all types of human illness. Cytokine activation
can be a problem. Inadequate resuscitation certainly occurs frequently. Finally,
the hyperdynamic, hypermetabolic high oxygen consumption requirements of
sepsis may be a problem [17]. Can we treat that by appropriate action to
increase it? We willleam more about that this morning.
Table 5. Biologic conundmms
Anti-ll-6 mab given to endotoxemic mice yields increased IL-6

11-6 deficient mice with liver failure have defective hepatocyte reg
Interferon gamma given to animals after injury decreased lethal abdominal sepsis
Interferon gamma given to animals without injury increased lethality of bacteria
TNF enhances antibiotic efficacy after hemorrhage shock
LPS pretreatment protects lungs from hepatic ischemia/reperfusion
Germ-free animals after hemorrhagic shock have increased survival but increased inflammatory
mediators
Interferon gamma primes only a sUbpopulation of PMNS (leukocytes)
Abscess model Anorexia, weight loss and hepatic-acute phase response is due in part to
IL binding to type I receptor. TNF binding is not required
Recombinant tissue decreases abscesses in rats with peritonitis but increases bacteremia and
pathway inhibitor death
A leukocyte CD 18 mab increases endotoxemia and CV injury with septic shock
Intestinal permeability but there is no relation to septic complications
is increased by trauma
and shock
Polymyxin decreased but there was no decrease in sepsis scores, IL-6 levels or mortality
plasma endotoxin
High TNF levels associated with survival in patients with abdominal septic shock
Immune-enhancing exaggerated immune response and increased ARDS
patients
TNFmab protects against endotoxin infusion but increases mortality with CLP
Macrophage-depleted increased systemic bacterial translocation decreased systemic toxic re-
mice sponse decreased mortality
TNF deficient c3H/HcJ had increased survival from hemorrhagic shock
mice but
anti-TNF mab ---7 abolished increased TNF but did not increase survival
PGE2 helps the liver
contributes to immunosuppression
Inhibition of CD 11/18 with intrabronchial E. coli improved lung function but increased mor-
(mab) - integrin receptor tality
on neutrophils
Prophylactic G-CSF may improve or worsen organ failure
upregulates immune
response increased
neutrophil function
Macrophage deficient mice ---7 decreased TNF, IL-l, GCSF and endotoxin induced bacterial trans-
(absent CSF-I) location but
morbidity and mortality same as in normal animals
LPS Induces PGE 2 which decreases macrophage TNF production
NSAIDS have beneficial effects on endotoxemia but increase TNF production
TGF-B initiates and terminates tissue repair but sustained production increases
fibrogenesis - tissue fibrosis
PGE2 produces immunosuppression
but
rats - 30% body bum Ibuprofen decreases certain aspects of immune function
but
rats - CLP Ibuprofen decreases survival
30 A.E. Baue
Table 6. Cause of organ dysfunction/failure
Ischemia - Dysoxia (Hypoxia)

Ischemia/Reperfusion
WBC activation - oxidants/failure of antioxidants, elastase
One hit - two hits - Injury, Infection
Overwhelming Inflammation
The Horror Antitoxicus
LPS
Bacterial translocation
Cytokine activation
Inadequate resuscitation
The hyperdynamic, hypermetabolic high O2 consumption of sepsis
Conclusion
We have made great advances in our understanding of the problems of inflam-
mation, injury and infection. However, in spite of this knowledge, our abilities
to treat patients successfully with these problems have been limited. I have
reviewed the complexities related to this and the areas where we need to learn
more. There is no doubt that, in the future, we will have information which will
help us to more specifically treat patients for problems related to injury, infec-
tion, leu problems and organ failure. As we do this more patients may survive
only to develop complications. Thus, the frequency and mortality of multiple
organ failure may stay the same.
References
1. Hunter J (1823) Treatise on the blood, inflammation, and gunshot wounds. 2nd edn James
Webster, Philadelphia, p 205
2. Baue AE (1997) SIRS, MODS, MOF - Why no magic bullets? Arch Surg 132: 1-5
3. Bone RC (1991) Sepsis, the sepsis syndrome, multi-organ failure: A plea for comparable defi-
nitions. Ann Intern Med 114:332-333
4. Baue AE (1992) The horror autotoxicus and multiple-organ failure. Arch Surg 127:1451-1462
5. Members of the American College of Chest Physicians/Society of Critical Care Medicine
Consensus Conference Committee (1992) American College of Chest Physicians/Society of
Critical Care Medicine consensus conference: definitions of sepsis and organ failure and
guidelines for the use of innovative therapies in sepsis. Crit Care Med 20:864-874
6. Vincent JL (1993) Dear SIRS, I'm sorry to say that I don't like you ... Crit Care Med
25(2):372-374
7. Bone RC (1996) Sir Isaac Newton, sepsis, SIRS, and CARS. Crit Care Med 24(7):1125-1128
8. Thomas L (1970) Adaptive aspects of inflammation, third symposium international inflamma-
tion club. Excerpta Medica 108. Upjohn Co, Kalamazoo
9. Ehrlich P, Morgenroth J (1901) Uber Hfunolysins. Fiinfte Mitheilung. Bed Klin Wochenschr
38:251-257
10. Baue AE (1994) Organ dysfunction (MODS) - Organ failure (MOF) and Therapeutic conun-
drums in injured and septic patients. A Gullo (ed) Organ Failure Academy. Fogliazza, Milan,
2:9-39
11. Elin RJ, WolffRJ (1976) Biology of endotoxin. Annual Rev Med 27:127-141
12. Ollodart RM, Hawthorne I, Attar S (1967) Studies in experimental endotoxemia in man. Am J
Surg 113:599
13. Tracey KJ, Fung Y, Hesse KR et al (1987) AnticachectinlTNF monoclonal antibodies prevent
septic shock during lethal bacteremia. Nature 330:662-664
14. Echtenacher B, Falk W, Mannel DN, Krammer PH (1990) Requirement of endogenous tumor
necrosis factor/cachectin for recovery from experimental peritonitis. J ImmunoI145:3762-3766
15. Faist MD, Baue AE, Dittmer H, Herberer G (1983) Multiple organ failure in trauma patients. J
Trauma 23(9):775-787
16. Sauaia A, Moore FA, Moore EE, Lezotte DC (1996) Early risk factors for postinjury multiple
organ failure. World J Surg 20:392-400
17. Durham RM, Neunaber K, Mazuski JE et al (1996) The use of oxygen consumption and deliv-
ery as endpoints for resuscitation in critically ill patients. J Trauma 41 (1 ):32
I SCORING SYSTEMS I
An Overview to Introduce Prognostic Indexes
in MODS
A.E. BAUE
Predictions are trickly

particularly about the future.
Sam Goldwyn
Hollywood, CA
Our purpose in this symposium is to seek a consensus on scoring or to try to de-

velop a methodology or system of outcome prediction which could be used by
all - a universal scoring system. Such a system could be inter-active and help
evaluate therapeutic interventions. Can we all speak the same language? The
ground work for this meeting was laid in 1995 here in Trieste when Professors
Gullo, Vincent and I reviewed the prognostic indexes and the newly developed
sepsis-related organ failure assessment (SOFA) program [1].
There have been several regional conferences on predicting outcome. One of
these was the 2nd European Consensus Conference in Intensive Care Medicine
in 1993 entitled "Predicting Outcome in ICU patients" [2, 3] *. The questions
raised at this conference were as follows:
1. How should outcome of intensive care be defined and assessed?
2. How accurately can we measure severity of illness in ICU patients?
3. Can the various means of measuring illness severity be used in the ICU to
predict outcome in groups of patients and in individual cases?
4. What are the human and economic costs of treating patients in the ICU who
have a very high risk of death or severe disability?
5. In clinical practice, can the measurement of illness severity be used to pro-
vide appropriate and avoid useless intensive therapy?
Three scoring systems were reviewed: Apache III [4], SAPS [5] and MPM-O
and 24 [6]. As nearly as I can tell, no consensus was reached other than to keep
working. Recommendations included doing research on quality of life after dis-
charge, examination of the decision making process and maintaining medical
control of clinical decisions. Illness severity scores were not recommended for
* The first consensus conference dealt with selective decontamination of the gut
36 A.B. Baue
routine clinical decision making or for triage. The conference concluded that
Severity Scores can be valuable for predicting mortality and for clinical trials.
The current low sensitivity of the scores precludes their use for predicting out-
come in individual patients.
Wright et al. in a recent publication on "Measurement in Surgical Clinical
Research" describe the issues involved in measurement and the development of
an index [7]. First is the definition of the purpose of the index. Feinstein classi-
fied four different objectives or purposes [8]: 1) to evaluate patients at a single
point in time (Status Indexes); 2) measurement of clinical change (Change In-
dexes); 3) prediction of an outcome (Prognostic Indexes); and, 4) description of
clinical change (Clinical Guidelines). Second is the focus or areas of interest of
the index. The focus can be objective outcomes, SUbjective evaluations or gener-
ic health status measurements. Third is the type of measurement or Multi-item
Indexes. This is followed by scale development for item generation and item re-
duction. After the instrument has been developed, it must be evaluated. Wright
et al. describe the evaluation criteria as sensibility, reliability, validity and re-
sponsiveness. Feinstein defined sensibility as "a mixture of ordinary common
sense plus a reasonable knowledge of pathophysiology and clinical reality". Re-
liability means that the same result is obtained when the same phenomenon is
measured repeatedly by the same or different physicians. Validity means that the
measure represents what is being sought. This can be criterion or construct va-
lidity. Finally responsiveness is the ability to measure clinical change. It has al-
so been called sensitivity.
It is helpful to review first where we have been. The history of scoring be-
gins with man's attempts to quantitate his activities (Table 1). Indices or scores
have been developed for the status of a situation at a point in time, indices of
change, prognostic indices and for clinical guidelines. Scoring systems have
been developed for a number of health related matters (Table 2).
After the initial description of MOF [9] a number of excellent classifications
of MOF were developed. It was soon recognized by Knaus and his group that
there is a wide distribution of severity of illness in patients with similar organ
failure classifications [10]. Thus MOF scales were imprecise. There could be a
little or a lot of organ failure. Certain combinations of organ failures are more
lethal than others [11]. This led to the development of the Apache concept on
one hand [12] and the SIRS, MODS, approach on the other [13].
What is the purpose of a score? Why score? What are the issues - to predict
an outcome (measure prognosis), describe or classify the severity of illness, to
set criteria for clinical trials of new therapies, to assist, guide or stop therapy?
Will these improve care? Are the various programs interactive for therapy? Can
a worsening score allow one to stop therapy or organ support or an improved
score indicate survival? We can describe metabolic and cardioventilatory char-
acteristics which are consistent with illness or survival, but will they contribute
to clinical research and other trials. As we compare survival or death and conti-
nuity of care - probability may not be a certainty. Finally should we strive for
An Overview to Introduce Prognostic Indexes in MODS 37
consensus and a single universal scoring system? Can we ever all speak: the
same language?
For each area of medical activity many different scores have been developed.
Thus for sepsis there are at least 16 different scoring systems (Table 3). For
severity of illness scoring there are a multitude of systems (Table 4). To attempt
to document the severity of injury and compare trauma results, at least 19
Table 1. History of scoring
Game scores MOFscores

Romantic conquest by women and men Disease scores
WelIness scores Injury scores
Neonatal score - APGAR TISS-28
Sepsis scores TRISS - Trauma and Injury Severity Score
SIRS and MODS scores PISS - Predicted Injury Severity Score
TISS-76 - Therapeutic Intervention Scoring System
Table 2. Scoring systems
Health Status Evaluation

Severity of Illness
Sepsis
Injury Severity
Futility
Injured extremity
Table 3. Sepsis scoring
Sepsis Severity Score (SSS) - Stevens

Sepsis Score - Elebute - Stoner
Complete Septic Shock Score
Simplified Septic Shock Score
Sepsis Related Mortality Score
Risk of Operative Site Infection Formula
Surgical Stratification System for Intra-Abdominal Infections
Peritonitis Index
DTH - Delayed Hypersensitivity Skin Test Score
LPS - Cytokine Score - LPS - IL - IB, IL-6
SSS - Severity of Surgical Sepsis
SOFA - Sepsis Related Organ Dysfunction and/or Infection Model
Customized Probability Model
ODIN Model - Organ Dysfunction and/or Infection Model
Mannheim Peritonitis Score - MPS
Baltimore Sepsis Scale - BSS
Systemic Mediator - Associated Response Test - SMART
38 A.E. Baue
methodologies have been proposed (Table 5). There are at least ten scoring sys-
tems for the evaluation of general health status (Table 6). Even injured extremi-
ties have been scored (Table 7), and attempts are being made to develop a futili-
ty score (Table 8),
Table 4. Severity of illness scoring

APGAR Score
APACHE I, IT, III Acute Physiology and Chronic Health Evaluation System
SAPS I, II Simplified Acute Physiology Score
LODS The Logistic Organ Dysfunction System
MPMII Mortality Probability or Prediction Model (System 24, 48, 72)
MODS Score Multi-Organ Dysfunction Score
CARE Clinical Assessment, Research and Education System
CIRS Cumulative Index Rating Scale
PODS Probability of Death Score
MUM Multiattribute Utility Model
DPE Daily Prognostic Estimates
MOF Scoring Multiple Organ Failure Scoring
OSF Organ Systems Failure
MSOFScore Multiple System Organ Failure Score
The Parsonet Score
CIS Cell Injury Score of Hirasawa
HIS Hanover Intensive Score
Physiologic State Classification Siegel
HPI Hospital Prognostic Index
TISS Therapeutic Intervention Scoring System 76 Items
SCOUT Surgical Complication Outcome Score
New Intermediate TISS
Simplified TISS 28 Items
PNI Prognostic Nutritional Index
EPI Endocrine Prognostic Index
PRISM Pediatric Risk of Mortality
POPCS Pediatric Overall Performance Category Score
PCPC Pediatric Cerebral Performance Category
Murray Lung Injury Score
MISGS Medical Illness Severity Grouping System
CCSS Critical Care Scoring System
POSSUM The physiological and operative severity score for the
enumeration of mortality (and morbidity)
SMART Systemic Mediator - Associated Response Test
HCWPS Health Care Workers Predictors of Survival
TICS The ICU Coma Score
Childs-Turcotte Classification Liver Disease
Ranson's Criteria Pancreatitis
Forrester's Classification MIS
VS Ventilator Score
AOSFSS Acute Organ System Failure Scoring System
I have not attempted to provide references for each of these efforts. The ma-
jor programs will be cited and include: Apache II and III [4, 14], a simplified
Table 5. Injury severity scoring
ISS Injury Severity Score

AIS Abbreviated Injury Score
TI Triage Index
AL Anatomic Index - HICSDA-8 Code
PEPL Penetrating and Blunt Injury Code
ASCOT A Severity Characterization of Trauma
OIS Organ Injury Scales - for each organ and region
GCS Glasgow Coma Score
TS Trauma Score
PTS Poly trauma Score
TI Trauma Index
BI Bum Index
MTO Major Trauma Outcome Study
HTI Hospital Trauma Index
CRAMS Circulation, Respiration, Abdomen, Motor, Speech
MESS Mangled Extremity Score
ATI Abdominal Trauma Index
TRISS Trauma and Injury Severity Score
TNN The Neural Network
ISS Injury Impairment Score
FCI Functional Capacity Index
Table 6. Scoring systems for health status evaluation
SIP - Sickness Impact Profile (general health status)

Functional Independence Measure - FIM
The Barthel Activities of Daily Living Scale
Nottingham Health Profile
McMaster Health Index questionnaire
Quality of Well Being Score
MOS (medical outcome study) Health Status Measures MOS SF-20 and MOS SF-36
Medisgroups Classification
Computerized Severity Index
The Cardiac Risk Index
Table 7. Injured extremity scores
MESI Mangled Extremity Syndrome Index

MESS Mangled Extremity Severity Score
PSI Predictive Salvage Index
LSI Limb Salvage Index
40 A.E. Baue
Table 8. Futility score
Any fatal illness with a life expectancy of six months or less

immune failure - AIDS
severe single organ failure - not easily supportable
COPD requiring increasing O2 at rest and not an operative candidate
congestive heart failure, recurrent
cardiomyopathy - not a transplant candidate
metastatic malignancy, nonresectable multiple lesions brain, lung, liver, bone, post-radiation
therapy
mental status, unresponsiveness - disoriented as to time, place and person
cerebral vascular disease - dense stroke with inability to care for one's self but not brain dead
hepatic failure
progressive MOF
severe sepsis with any of the preceding
end stage metabolic musculru: disease
do not resuscitate order - no code
incapable of independent living with any of the preceding
Alzheimer's disease
persistent vegetative state - how long?
Fig. 1. A SOFA is a piece of furniture on which the cartoon character Dagwood Bumstead is lying
Fig. 2. A cartoon of the Tower of Babel with each construction worker speaking a different lan-
guage (recommending a different scoring system)
acute physiology score (SAPS II) [5], mortality probability models (MPM II)
[6,15, 16], the therapeutic intervention scoring system (TISS) [17], the multiple
organ dysfunction score (MODS) [18], developed by Marshall and the sepsis re-
lated organ failure assessment score (SOFA) developed by Vincent and col-
leagues [19, 20]. The emphasis of this group is on ideal variables which are ob-
jective, simple-easily avaiiable and reliable, obtained routinely and regularly in
every institution, specific for the function of the organ considered, a continuous
variable, independent of the type of patients, independent of the therapeutic in-
terventions. The expression SOFA is an interesting one. To those from the Unit-
ed States, SOFA means a couch or a piece of furniture to sit or lie upon (Fig. 1).
Recently LeGall and his group have developed two new instruments for
probability determination. The first approach is to customize existing models
such as SAPS II or MPM II [24] for subgroups of patients such as those with
early severe sepsis [21]. They propose this technique as an adjunct for clinical
trials of new therapeutic agents. More recently LeGall et al. for the ICU scoring
group have developed a new way to assess organ dysfunction in the ICU. They
call this the Logistic Organ Dysfunction System (LODS). On the first day in the
ICU they identify from 1 to 3 levels of organ dysfunction for 6 organ systems
and the relative severity among organ systems. Most of these programs will be
described by their developers later.
There have been a number of studies and publications of comparisons of
many of these scoring systems. These include a guide to prognostic scoring sys-
tems by Seneff and Knaus [23]. Castella et al. led a multicenter, multinational
study and concluded that Apache III, SAPS II and MPM II perform better than
42 A.E. Baue
their predecessors and all three showed good discrimination and calibration [24].
Roumen et al. compared seven scoring systems including Apache II and the In-
jury Severity Score in severely traumatized patients [25]. They concluded that
the ISS for example predicted complications such as ARDS and MOF whereas
Apache predicted mortality. Barie et al. found that the combined use of Apache
III and the MODS score predicted a prolonged stay in the ICU, but could not
predict outcome adequately in individual patients [26]. Rutledge raises the ques-
tion as to whether the ISS will differentiate between severe injury or poor care
[27]. Lewis points out that timing of the measurement will make a difference
[28]. Comparison of APACHE II and TRIS scores in trauma patients indicated
that both accurately predicted group mortality in ICU-trauma patients, but nei-
ther was accurate enough for prediction of outcome in individual patients [29].
Recently the problem of medical futility in ICU care has been described.
When is medical care and organ support no longer helpful? Civetta has defined
medical futility as "a situation in which further therapy seems useless" [30]. At-
tempts are being made to further define this and also to place such a concept in
an ethical setting in the ICU [31-33]. Much more will be heard about this in the
future. Knaus and others have reviewed the matter of whether objective esti-
mates of chances for survival should influence therapy to treat or withdraw
treatment [34]. As Knaus and his co-workers have stated "probability models
can never predict whether a patient will live or die with 100% accuracy" [23].
An exciting area of study will be whether predicted risk of mortality can help
to evaluate therapy. Many recent clinical trials of new therapeutic agents have
been negative. Initial retrospective evaluations by predicted risk have been en-
couraging [35].
So where does all of this leave us? How precise do we wish to be? We are
better at words and scores (personal classifications, descriptions and calculated
constructs) than we are at therapeutic advances which require development; trial
and proof. The cost of such programs must also be considered. Will we ever de-
velop a consensus and a single or several systems for universal use? I think not.
Even within the last year, new systems have been developed so that one can pre-
dict that this will continue. I used the comparison ofthe message about the Tow-
er of Babel from the Book of Genesis in the Old Testament of the Bible. In that
story, God punished mankind by confounding the single language of the people
into many languages so that they could no longer understand each other. Work
on building a tower to heaven thus ceased (Fig. 2). As LeGall wrote recently
"New systems to assess dysfunction are proposed almost every year, and each
system differs from the others in large or small ways" [22].
Will we ever develop a scoring system which is accurate enough for individ-
ual patients so as to predict mortality with complete certainty. I doubt it, but it is
dangerous to make absolute predictions. The vagaries and complexities of hu-
man disease are a source of continuous wonderment.
I could also use the evolution of terms for human illness and causes of mor-
bidity and mortality as an example. We began with terms such as infection.
Then sepsis and the sepsis syndrome were developed. The concept of multiple
organ failure was developed with scoring systems for it. Then the expression
systemic inflammatory response syndrome (SIRS) and multiple organ dysfunc-
tion syndrome (MODS) were developed, and everyone hopped on that band-
wagon. Bone has now extended the acronymic battle to include the terms CARS
(compensatory anti-inflammatory response syndrome), MARS (mixed antago-
nistic response syndrome) and CHAOS (cardiovascular compromise, homeosta-
sis, apoptosis, organ dysfunction, suppression of immune system). Where will it
end? The answer is that it will not end - ever. Man's need to strive and innovate
was well described by the poet Ralph Waldo Emerson in "For an Autograph"
where he wrote: "Though old the thought and oft expressed tis his at last who
says it best". Perhaps we should paraphrase Emerson to read "Tis his at last who
says it last".
References
1. Baue AE, Gullo A, Vincent JL (1995) Sepsis and organ dysfunction/failure: Know the
process and improve the art. In: Gullo A (ed) Organ Failure Academy: Sepsis and Organ Fail-
ure. APICE, Trieste, Italy
2. Suter P, Armaganidis A, Beaufils F et al (1994) Consensus conference organized by the
ESICM and the SRLF. Predicting outcome in ICU patients. Intensive Care Med 20:390-397
3. Suter P, Armaganidis A, Beaufils F et al (1994) Predicting outcome in intensive care unit pa-
tients: Second European consensus conference in intensive care medicine. Intensive Care
World 11(4):148-151
4. Knaus WA, Wagner DP, Draper EA et al (1991) The APACHE III prognostic system: Risk
prediction of hospital mortality for critically ill hospitalized adults. 100: 1619-1636
5. Le Gall J-R, Lemeshow S, Saulnier F (1993) A new Simplified Acute Physiology Score
(SAPS II) based on a EuropeanINorth American multicenter study. 270:2957-2963
6. Lemeshow S, Teres D, Klar J et al (1993) Mortality probability models (MPM II) based on
an international cohort of intensive care unit patients. JAMA 270:2478-2486
7. Wright JG, McLeod RS, Lossing A et al (1996) Measurement in surgical clinical research.
Surgery 119(3):241-244
8. FeinsteinAR (1987) Climinetrics New Haven CT, Yale University Press
9. Baue AE (1975) Multiple progressive or sequential systems failure - A syndrome of the 70s.
Arch Surg 110:779
10. Knaus WA, Draper EA, Wagner DP et al (1985) Prognosis in acute organ-system failure. Ann
Surg 202:685-693
11. Faist E, Baue AE, Dittmer H, Heberer G (1983) Multiple organ failure in poly trauma pa-
tients. J Trauma 23:775
12. Knaus WA, Zimmerman JE, Wagner DP et al (1981) APACHE - acute physiology and chron-
ic health evaluation: A physiologically based classification system. Crit Care Med 9:591-597
13. AACP/SCCM Consensus Conference Committee (1992) Definitions for sepsis and organ
failure and guidelines for the use of innovative therapies in sepsis. Chest 101: 1644
14. Knaus WA, Draper EA, Wagner DP et al (1985) APACHE II: a severity of disease classifica-
tion system. Crit Care Med 13:818-829
15. Zhu B-P, Lemeshow S, Hosmer DW et al (1996) Factors affecting the performance of the
models in the mortality probability model II system and strategies of customization: A simu-
lation study. Crit Care Med 24(1):57-63
44 A.E. Baue
16. Lemeshow S, KIar J, Teres D et al (1994) Mortality probability models for patients in the in-
tensive care unit for 48 or 72 hours: A prospective, multicenter study. Crit Care Med 22
(9): 1351-1358
17. Cullen DJ, Civetta JM, Briggs BA et al (1974) Therapeutic intervention scoring system: a
method for quantitative comparison of patient care. Crit Care Med 2:57-60
18. Marshall JC, Cook DJ, Christou NY et al (1995) Multiple organ dysfunction score: a reliable
descriptor of a complex clinical outcome. Crit Care Med 23:1638-1652
19. Vincent JL (1996) End-points of resuscitation: Arterial blood pressure, oxygen delivery,
blood lactate, or ... ? Intensive Care Med 22:3-5
20. Vincent JL, Moreno R, Takala J et al (1996) The SOFA (sepsis-related organ failure assess-
ment) score to describe organ dysfunction/failure. Int Care Med 22: 1-4
21. Le Gall JR, Lemeshow S, Leleu G (1995) Customized probability models for early severe
sepsis in adult intensive care patients. JAMA 273:644-650
22. Le Gall JR, KIar J, Lemeshow S (1996) The logistic organ dysfunction system: A new way to
assess organ dysfunction in the intensive care unit. JAMA 276:802-810
23. Seneff M, Knaus WA (1990) Predicting patient outcome from intensive care. A guide to
APACHE, MPM, SAPS, PRISM, and other prognostic scoring systems. J Intensive Care Med
5:33-52
24. Castella X, Artigas A, Bion J et al (1995) A comparison of severity of illness scoring systems
for intensive care unit patients: Results of a multicenter, multinational study. Crit Care Med
23(8):1327
25. Roumen RMH, Redl H, Schlag G et al (1993) Scoring systems and blood lactate concentra-
tions in relation to the development of adult respiratory distress syndrome and multiple organ
failure in severely traumatized patients. J Trauma 35(3):349-355
26. Barie PS, Hydo LJ, and Fischer E (1996) Utility of illness severity scoring for prediction of
prolonged surgical critical care. J Trauma 40(4):513-518
27. Rutledge R (1996) The injury severity score is unable to differentiate between poor care and
severe injury. J Trauma 40(6):944-950
28. Lewis FR (1996) Editorial Comment. J Trauma 40(6):950
29. Wong DT, Barrow PM, Gomez M, McGuire GP (1996) A comparison of the acute physiolo-
gy and chronic health evaluation (APACHE) II score and the trauma-injury severity score
(TRISS) for outcome assessment in intensive care unit trauma patients. Crit Care Med 24:
1642-1648
30. Civetta JM (1996) Futile care or caregiver frustration? A practical approach. Crit Care Med
24(2):346-251
31. Curtis JR, Park DR, Krone MR et al (1995) Use of the medical futility rationale in do not-at-
tempt-resuscitation orders. JAMA 273(2):124-128
32. Hoyt JW (1995) Medical Futility. Crit Care Med 23(4):621-622
33. Atkinson S, Bihari D, Smithies M et al (1994) Identification of futility in intensive care.
Lancet 344:1203-1206
34. Knaus WA, Rauss A, Alperovitch A et al (1990) Do objective estimates of chances for sur-
vival influence decisions to withhold or withdraw treatment? Med Decis Making 10: 163-171
35. Knaus WA, Harrell FA, LaBrecque JF et al (1996) Use of predicted risk of mortality to evalu-
ate the efficacy of anticytokine therapy in sepsis. Crit Care Med 24(1):46-56
36. Bone RC (1996) Sir Isaac Newton, sepsis, SIRS, and CARS. Crit Care Med 24(7): 1125-1128
Multiple Organ Dysfunction (MOD) Score
J.C. MARSHAlL
Articulation of the concept of multiple organ failure some two decades ago [1]
marked a seminal shift in our understanding of the process of care in the ICU.
Baue's landmark editorial made explicit an evolving recognition, that despite
the heterogeneity of conditions that lead to ICU admission, the subsequent clini-
cal course and postmortem findings are remarkably similar. Moreover, death,
when it occurs, is not the consequence of isolated lung, heart, or renal failure,
but rather reflects the necessary interdependence of multiple organ systems in-
volved in the maintenance of homeostasis.
Two features of this new syndrome were readily apparent. First, it was common-
ly [2, 3], though not invariably [4, 5], a consequence of occult unrecognized infec-
tion. Secondly, it portended significant morbidity and substantial mortality [6].
These two faces of multiple organ failure - one a manifestation of a patho-
physiologic process, and the other an expression of ICU morbidity - remain the
cardinal features of the clinical syndrome that must be reflected in any system
that seeks to describe organ dysfunction. Since organ dysfunction is a potential-
ly preventable complication of critical illness, and in particular of the sequelae
of infection, ischemia, and injury, it is particularly desirable that reliable and
validated measures of organ dysfunction be developed and tested.
Following the 1991 Consensus Conference of the ACCP/SCCM [7], and
based on previous work by Goris [8] and ourselves [5], we developed the Multi-
ple Organ Dysfunction (MOD) Score as an objective tool to quantify organ dys-
function as an outcome in critical illness [9] (Table 1).
Methodologic principles of the MOD score

The MOD Score was developed using a formal methodologic approach to maxi-
mize construct, content, and criterion validity.
Construct validity concerns the extent to which the score reflects organ dys-
function as it is seen by the intensivist. To maximize construct validity, we un-
dertook a systemic evaluation of 30 published reports of organ failure to define
what systems and what variables would seem to comprise the syndrome [10].
46 J.C. Marshall
Table 1. The multiple organ dysfunction (MOD) score
Organ system 0 1 2 3 4
Respiratory a (PO:IFI02 ratio) >300 226-300 151-225 76-150 5, 75
Renal b (serum creatinine) 5, 100 101-200 201-350 351-500 >500
Hepatic c (serum bilirubin) 5, 20 21-60 61-120 121-240 >240
Cardiovascular d (RIP ratio) 5, 10.0 10.1-15.0 15.1-20.0 20.1-30.0 >30.0
Hematologic e (platelet count) > 120 81-120 51-80 21-50 5, 20
Neurologic f (Glasgow Coma Score) 15 13-14 10-12 7-9 5,6
a The P021FI02 ratio is calculated without reference to the use or mode of mechanical ventilation, and without reference to
the use or level of PEEP
b The serum creatinine level is measured in mmollliter, without reference to the use of dialysis
C The serum bilirubin level is measured in mmolJliter
d The RIP ratio is calculated as the product of the heart rate and right atrial (central venous) pressure, divided by the mean
arterial pressure:
RIP ratio = .:.:H.:.:.eart.:.:..:.:.ra:,:..:te...;.:x___
RAP-,-=-
meanBP
e The platelet count is measured in platelets/mL 10-3
f The Glasgow Coma Score is preferably calculated by the patient's nurse, and is scored conservatively (for the patient
receiving sedation or muscle relaxants, normal function is assumed unless there is evidence of intrinsically altered
mentation)
Content validity reflects the extend to which the score encompasses the nature
of dysfunction in any given system. To maximize content validity, we estab-
lished a series of criteria to define the ideal descriptor of organ system dysfunc-
tion and measured candidate variables against these [10]. Finally, criterion va-
lidity reflects the extent to which the score reflects the process of organ failure
when measured by an independent gold standard. Lacking a definable biochem-
ical marker, we chose to use leu mortality as the standard against which to
evaluate criterion validity. leu rather than hospital mortality was selected since
the purpose of developing the score was not to permit the prediction of ultimate
survival, but rather to characterize a process whose expression occurs in the leu
and necessitates ongoing care there.
The MOD score is summarized in Table 1. It is similar to the recently pub-
lished systems including the SOFA Score [11], LOD Score [12], and Brussels'
Score [13]. Its differences reflect the conceptual challenges that remain to be re-
solved if a single widely accepted system is to be developed.
Organ dysfunction scales: unresolved questions

Should a scale reflect deranged physiology or therapeutic
intervention?
In seeking to produce a score derived purely from physiologic measures, we
have consciously stayed away from any consideration of therapy_ Values for
Multiple Organ Dysfunction (MOD) Score 47
each of the variables are recorded without consideration of therapeutic interven-

tion (the P021F102 ratio, for example, is calculated independent of the presence
of mechanical ventilation). This decision was made for several reasons. First we
wished to minimize the systematic error that would result because therapy dif-
fers from one centre to the next; such an approach maximizes the generalizabili-
ty of the score. Our purpose in developing the score was to produce a tool that
could be used to understand the pathophysiology of the multiple organ dysfunc-
tion syndrome, therefore measures reflecting physiology were used in prefer-
ence to those reflecting therapeutic approach. Physiologic measures are more
readily treated as continuous variables, thus increasing the sensitivity of the
score. Finally, modelling on our original database showed the effects of incorpo-
rating therapy on variable waiting to be minimal. Perhaps this is not surprising.
Evidence of inadequate oxygenation, for example, will generally lead the inten-
sivist to initiate ventilatory support to optimize the clinical situation.
When should variables be measured?

The model of organ dysfunction reflected in the MOD Score is one that views
organ dysfunction not as an acute and easily reversible abnormality, but as the
derangement that remains after maximal support has been instituted. Thus, a pa-
tient with hypotension that responded rapidly to volume replacement would not
be seen as having cardiovascular dysfunction, whereas another who is nor-
motensive only with maximal volume resuscitation and the use of vasopressors
would be considered to have significant cardiovascular dysfunction. Thus deter-
mination of the score should occur after, not prior to or during, resuscitation.
Equally such a model of stable dysfunction requires that the measurement of
each variable reflect a value that is representative of physiologic derangement
over time, to minimize the likelihood that abnormalities of respiratory function,
for example, reflect an acute reversible event such as desaturation during suc-
tioning. Thus, the variables recorded to calculate MOD scores are representative
values taken at a standard time of day rather than the worst values recorded dur-
ing a 24 hour period. Recording representative variables further minimizes the
bias that results because the frequency of ascertainment varies from one centre
to the next. Other approaches to identifying representative variables are logisti-
cally complex and were rejected for this reason.
How are variables selected and calibrated?

Unlike a prognostic score that is calibrated to maximize predictive ability, the
MOD Score was developed as an outcome measure. Thus, the variables chosen
and their weightings were selected not on the basis of their ability to predict
death, but on the basis of their ability to describe organ dysfunction in a clini-
cally relevant manner. Indeed if one accepts the estimate that 20% of leU
48 J. C. Marshall
deaths result from processes other than organ dysfunction [14], a valid organ
dysfunction measure should record a low score for patients who die without sig-
nificant organ failure, and thus perform significantly less well as a predictor of
ICU mortality.
How is cardiovascular dysfunction measured?

Perhaps the most striking difference between contemporary organ dysfunction
scores is in the variable selected to measure cardiovascular dysfunction. Our re-
view of published descriptive systems failed to identify an available marker that
met minimally acceptable criteria for incorporation into the MOD Score [10].
Therefore, we developed a novel descriptor termed the Pressure Adjusted Heart
Rate (PAR). The Pressure Adjusted Heart Rate is calculated as the product of
the heart rate, and the ratio 'of CVP to mean arterial pressure:
PAR = HRxCVP
MAP
Like the P02IFI0 2 ratio, the PAR provides a measure of physiology that is
less dependent on therapy. Prior to volume replacement, the value is low; as the
CVP rises, the PAR increases. A normal physiologic response - a reduction in
heart rate as the mean arterial pressure increases - will keep values of the PAR
low. On the other hand, persistent tachycardia and hypotension despite adequate
filling pressures produce high values. Moreover, tachycardia resulting from va-
soactive drugs will likewise increase the value of the PAR.
A major drawback of the PAR is that it has not been widely validated. It may
well be that other simpler and equally valid measures of cardiovascular dysfunc-
tion (for example, fluid balance or change in body weight) can be identified. We
are currently undertaking a prospective evaluation of the performance of the
PAR and several other candidate variables for the description of cardiovascular
dysfunction in hopes of identifying an optimal descriptive valuable.
Applications of the MOD score

A validated tool to measure organ dysfunction has a number of potential uses
both within the framework of clinical trials and in day to day ICU practise.
These will be briefly outlined.
Prognostication and severity of illness stratification

Calculated at the time of ICU admission, an organ dysfunction scale provides a
measure of global illness severity, expressed using the construct of the multiple
organ dysfunction syndrome. Although we found that the performance of the

MOD Score was comparable to that of APACHE II in predicting ICU outcome
[9], in general it would be expected that an organ dysfunction score will not be
as robust in predicting the probability of ICU survival. On the other hand, for a
clinical trial that uses organ dysfunction as an endpoint, such a score provides
both a potential stratification variable and a means of ensuring comparability of
study groups at the outset of the study.
Organ dysfunction as a point measure of illness severity

Organ dysfunction scores calculated on any given day of the ICU stay provide a
point measure of severity of illness and, by inference, of the intensity of thera-
peutic intervention. Thus, daily organ dysfunction scores can serve the same
role currently played by instruments such as the therapeutic intervention scoring
system that measure therapeutic intensity.
Measurement of global severity of organ dysfunction

By summing the worst values for each individual variable over the entire ICU
stay, an organ dysfunction score provides a measure of the global severity of ill-
ness over time. Such a measure incorporates not only pretreatment derange-
ments, but also those that develop during the ICU stay and has been the most
widely used approach to the measurement of organ failure.
Measurement of attributable ICU morbidity

Of greater interest in evaluating an ICU therapy is the ability to measure attrib-
utable morbidity during the ICU stay. Using the MOD Score, this can be accom-
plished by calculating a ~ODS - the difference between the overall MOD
Score and the MOD score on the day of admission. This difference represents
organ dysfunction developing following ICU admission and therefore potential-
ly amenable to modulation by ICU-based therapy. The ~ODS can be calculat-
ed over the entire ICU stay or over a defined time interval, such as the 28 day
period of a clinical trial.
Combined measures of morbidity and mortality

A composite measure that includes both morbidity and mortality is the Mortali-
ty-Adjusted MOD Score. Since survival with a high MOD Score is clearly
preferable to death with a low Score, all nonsurviving patients are automatically
assigned a maximal number of points, 24. For surviving patients, the number of
points scored is the number of points accumulated over the ICU stay. Such a
calculation converts the dichotomous variable of mortality into the maximum
50 J. C. Marshall
value of a continuous scale and therefore permits combined evaluation of mor-

bidity and mortality.
Conclusions
A number of similar systems have evolved for the quantification of organ dys-
function in critical illness. Their differences reflect both different concepts of
the nature of the syndrome and different potential uses. Nonetheless, their simi-
larities are striking and it is to be anticipated that consensus on a single, widely
applicable model will ultimately be achieved.
Perhaps one of the most important differences between existing systems is
their implicit intent in measuring pathophysiologic derangements of a disease
process on the one hand, and the therapy required to support these on the other.
While pathophysiology and therapy are intimately interrelated, it may well be
useful to evaluate each separately within a clinical trial. A purely physiologic
measure such as the MOD Score may have advantages in describing a disease
process and elucidating details of its epidemiology and fundamental biology. As
a continuous variable, it is sensitive, and because it avoids therapy, it avoids the
bias that would result from differing therapeutic practises.
On the other hand, the dichotomous decision to institute therapy or not re-
flects the fundamental reality of leu care from the perspective of the inten-
sivist. Therefore, a scale based on a series of dichotomous, therapy-dependent
variables reflecting these decisions may be intuitively more appealing to the
practitioner. Moreover, since the institution of therapy carries inevitable costs, a
therapy-based measure may provide a better reflection of the impact of a novel
therapy. Thus, it may well be desirable to look to developing two complen:ten-
tary systems - one based in physiology and similar to the MOD Score, and a
second reflecting a series of dichotomous therapeutic valuables.
The task of achieving consensus on the description and quantification of or-
gan dysfunction in critical illness is much more than a question of the simple
promotion of competing models. Whether organ dysfunction represents a single
disease process for which therapeutic intervention may be beneficial is un-
known. However it is clear that this process, however imperfectively defined, is
a leading cause of morbidity and mortality in the leu and responsible for sig-
nificant demands on national health care budgets. As a generic measure of leu
morbidity, the model of organ dysfunction provides a useful framework for
quantifying the adverse consequences of therapy. The need to do so in a valid
and methodologically rigorous fashion is compelling.
References
1. Baue AB (1975) Multiple, progressive, or sequential systems failure. A syndrome of the 1970s.
Arch Surg 11 0:779-781
2. Polk HC, Shields CL (1977) Remote organ failure: a valid sign of occult intraabdominal infec-
tion. Surgery 81:310-313
3. Fry DE, Pearlstein L, Fulton RML, Polk HC (1980) Multiple system organ failure. The role of
uncontrolled infection. Arch Surg 115: 136-140
4. Faist E, Baue AB, Dittmer H, Heberer G (1983) Multiple organ failure in polytrauma patients.
J Trauma 23:775-786
5. Marshall JC, Christou NY, Hom R, Meakins JL (1988) The microbiology of multiple organ
failure. The proximal GI tract as an occult reservoir of pathogens. Arch Surg 123 :309-315
6. Knaus WA, Draper EA, Wagner DP, Zimmerman JE (1985) Prognosis in acute organ system
failure. Ann Surg 202:685-693
7. Bone RC, Balk RA, Cerra FB et al (1992) Definitions for sepsis and organ failure and guide-
lines for the use of innovative therapies in sepsis. Chest 101:1644-1655
8. Goris RJA, te Boekhorst TPA, Nuytinck JKS, Girnbrere JSF (1985) Multiple organ failure.
Generalized autodestructive inflammation? Arch Surg 120: 1109-1115
9. Marshall JC, Cook DJ, Christou NV et al (1995) Multiple organ dysfunction score: A reliable
descriptor of a complex clinical outcome. Crit Care Med 23: 1638-1652
10. Marshall JC (1995) Multiple organ dysfunction syndrome (MODS). In: WJ Sibbald and J-L
Vincent (eds) Clinical trials for the treatment of sepsis. Springer, Berlin, pp 122-138
11. Vincent JL, Moreno R, Takala J et al (1996) The sepsis-related organ failure assessment
(SOFA) score to describe organ dysfunction/failure. Intens Care Med 22:707-710
12. Le Gall JR, Klar J, Lemeshow S et al for The ICU Scoring Group (1996) JAMA 276:802-810
13. Bernard GR, Doig G, Hudson G et al (1995) Quantification of organ dysfunction for clinical
trials in sepsis. Am J Respir Crit Care Med 151:A323 (Abstr)
14. Deitch EA (1992) MUltiple organ failure. Pathophysiology and potential future therapy. Ann
Surg 216:117-134
The Logistic Organ Dysfunction (LOD) System
J.R. LE GALL, J. KLAR, S. LEMESHOW
The assessment of the severity of organ dysfunction in the ICU is a critical tool
for conducting clinical trials, especially sepsis trials. The evaluation of new ther-
apies cannot be successfully achieved without checking the degree of organ dys-
function. It is not adequate to assess severity, or to describe a patient's condi-
tion, by simply counting the number of dysfunctioning organ systems.
In fact, new systems to assess organ dysfunction are proposed almost every
year, and each system differs from the others to a greater or lesser extent. As
early as 1980, Fry et al. proposed a system of 4 organ failures for surgical pa-
tients: pulmonary, hepatic, gastrointestinal, and renal failure [1]. In 1983,
Stevens described the Sepsis Severity Score comprising 7 failures, each with 5
severity levels [2]. The system of Marshall et al. contained metabolic failure
and took anergy into account [3]. The widely used Organ System Failure
(OSF) score was published in 1983 by Knaus et al. [4], and in 1989 [5] he-
patic failure was added. Fagon et al. added infection to the assessment of or-
gan dysfunction and called their system ODIN (Organ Dysfunction and/or In-
fection) [6]. Hebert et al. published a multiple organ failure scoring system
for patients who have sepsis syndrome [7]. Recently, Marshall et al. proposed
the Multiple Organ Dysfunction Score (MODS) based on a review of 30 re-
ports in the literature [8].
In many scoring systems, each organ dysfunction is graded from 1 to 4
points, or from 1 to 6 points, and a score is produced by adding the points.
These systems cannot adequately reflect patient severity. Not only are the
ranges defining the levels different from those found using statistical methods,
but weighting each organ system in the same way does not take into account the
differential prognostic significance of the involved organs.
In order to propose an objective system, we decided to use the large data
base of the European North American study (ENAS) and apply the statistical
technique of multiple logistic regression [9-11]. Although based on sophisticat-
ed statistical methods, our goal was to develop a system that was as simple as
possible to apply in the ICU. In developing a statistically based system, ranges
and weights of the variables defining levels of organ dysfunction can be deter-
mined objectively, the significance of severity levels for each organ can be iden-
54 J.R. Le Gall et al.
tified, and the levels of dysfunction can be weighted according to their relative
prognostic significance.
In the resulting Logistic Organ Dysfunction (LOD) System, the points for in-
dividual severity levels of each organ system reflect both the relative severity of
the levels within an organ system and the relative severity of the levels among
organ systems. The LOD score is a global score that can be calculated to sum-
marize the combined effect of dysfunction among several organs. In addition,
the LOD model is a logistic regression equation that can be used to translate the
score into a probability of mortality based on organ dysfunctions.
How was the LOn system created?

Data on 14,745 consecutive ICU admissions were collected in 137 medical, sur-
gical, or mixed ICUs in 12 countries on consecutive admissions to the ICUs
[12]. Eligible patients were aged 18 years or older; bum patients, coronary care
patients, and cardiac surgery patients were excluded. To develop and validate
the LOD System, 80% of the patients in the database were randomly selected to
constitute the developmental sample, and the remaining 20% composed the vali-
dation sample.
Variables were extracted from the database to define failure, based on a com-
bination of 12 variables, for six organ systems: cardiovascular (heart rate and
systolic blood pressure), hematologic (platelets and white blood cell count), he-
patic (bilirubin and prothrombin time), neurologic (Glasgow Coma Score), pul-
monary (ventilation/CPAP [continuous positive airway pressure] status and
Pa02IFi02 ratio), and renal (creatinine, urea, and urine output). The variables
had been recorded as the worst value in the first 24-hour period in the ICU ..
For sedated patients, the GCS was ascertained either from interviewing the
physician who ordered the sedation, or by reviewing the patient's medical
record before sedation. If a variable was not measured for a patient, it was as-
sumed to be within the range of normal. All variables except platelet counts and
prothrombin time (PT) were continuously scaled. Platelet counts were recorded
as being less than 50 x 1091L, and PT was recorded as being more than 3 sec-
onds over standard or less than 25% of standard.
The analysis was designed to first identify cut points that defined variable
ranges associated with changes in mortality rate.
When the points for each severity level were known, the LOD Score was cal-
culated by summing the points associated with each of the involved organ sys-
tems. The LOD score was then used as the single variable in a multiple logistic
regression equation of the form:
Logit = Bo + Bl x (LOD Score)

The Logistic Organ Dysfunction (LOD) System 55
The logit containing the LOD Score was then converted to a probability of
hospital mortality as:
e10git
Pr(Y = l/logit) =
1 + e10git
where Y equaled 1 for patients who died, Y equaled 0 for patients who lived, Pr
indicated probability, and e indicated a mathematical constant 2.7182818, which
represented the base of the natural logarithm.
The assessment of model performance was the final stage of the analysis. To
evaluate model calibration, Hosmer-Lemeshow goodness-of-Fit tests, compar-
ing observed with expected mortality, were performed [3]. To evaluate discrimi-
nation, area under the receiver operating characteristic (ROC) curve was calcu-
lated [13].
The LOD points that can be scored for each level of organ dysfunction are
shown for each organ system in Table 1. From the Table, it can be seen the neu-
rologic, cardiovascular, and renal dysfunction score the maximum of 5 LOD
points for the most severe level of dysfunction. Pulmonary and hematologic sys-
tem dysfunction score a maximum of 3 LOD points, and hepatic dysfunction
scores a maximum of 1 LOD point. The LOD score can range from 0 to 22
points. Figure 1 shows the distribution of the LOD score from 0 to 22 points in
the developmental sample. An LOD score of 0 indicates no organ dysfunction.
An LOD score of 1 is the score for the lowest level of severity for 1 organ sys-
tem dysfunction, and an LOD score of 22 points is the score for the highest lev-
el of severity for all 6 organ dysfunctions.
Of the 10,547 patients in the developmental sample, 1293 (12,3%) had no or-
gans dysfunction, 2723 (25,8%) had 1 organ dysfunction, 2615 (24,8%) had 2
organs in dysfunction, and 3916 (37,1%) had 3 or more organs in dysfunction.
Regardless of the number of organs in dysfunction, the LOD score varied
widely by the severity of the dysfunction. Depending on the involved organs
Table 1. The logistic organ dysfunction (LOD) system: Three levels of increasing severity with
corresponding points for each organ system
Severity Level
Organ System 0 1 2 3
LODPoints
Neurologic 0 3 5
Cardiovascular 0 3 5
Renal 0 3 5
Pulmonary 0 3
Hematologic 0 3
Hepatic 0
18
-
16
14
12 - I-r-
"e
0
IIi 10 -
C
CD
8
-
~
-
8 -
4
0
o 2 4 8
-nn,
8 10 12 14 16 18 20 22
LOO Score
Fig. 1. Distribution of the logistic organ dysfunction (LOD) score among 10,547 patients in the
developmental sample
and the level of severity, the LOD score can be as low as 1 or as high as 5 with 1
organ dysfunction.
It can be as low as 6 or as high as 22 with 6 organs in dysfunction.
The final scoring system for the LOD score is presented in Table 2. For each
organ dysfunction defined by more than 1 variable, only one of the variables
needs to be in the abnormal range for the LOD points to be assigned. All of the
variables defining an organ dysfunction must be within the normal range to re-
ceive 0 LOD points for that organ dysfunction (Table 3). To calculate the LOD
score for a patient, the points for each organ dysfunction are summed.
The application of the LOD score in the ICU can be illustrated using data for
a hypothetical patient as an example. Consider a patient admitted to the ICU for
septic shock with extreme oliguria. The WBC count is 2.0 x 109/L, the systolic
blood pressure is 60 nunHg, and there is no evidence of pulmonary, hepatic, or
neurologic dysfunction. The creatinine level is 88 Ilmol/L (1 mgldL).
In calculating the LOD score, being oliguric contributes 5 points to the LOD
score for renal dysfunction, the low WBC count contributes 1 LOD point for
hematologic dysfunction, and the systolic blood pressure contributes 3 LOD
points for cardiovascular dysfunction, for a total LOD score of 9 points.
Table 2. Scoring for the logistic organ dysfunction (LOD) system
~
LODPoints b
!G.
Increasing Organ Increasing '"n::to
Organ System Measures SeverityIDecreasing Dysfunction Severity/Increasing
0
Values Free Values cia
5 3 1 0 1 3 5
g;
CI
Neurologic '<
Glasgow Coma Score 3-5 6-8 9-13 14-15 ~6
'"
§'
n
Cardiovascular ::to
0
Heart rate, beats/min <30 30-139 ~140 ='
or and or
Systolic blood pressure, mmHg <40 40-69 70-89 90-239 240c269 ~270 5CI
'-'
Renal CI.l
Serum urea, mmollL (gIL) < 6 « 0.36) 6-9.9 (0.36-0.59) 10-19.9 (0.60-1.19) ~ 20 (~120) '<
or '"
Serum urea nitrogen, mmollL (mg/dL) < 6 « 17) 6-9.9 (17-< 28) 10-19.9 (28-<5 6) ~ 20 (~56) ~
and or or
Creatinine, 1llI10llL (mgL/dL) < 106 « 1.20) 106-140 (1.20-1.59) 3141 (31.60)
and or
Urine output, Ud <0.5 0.5-0.74 0.75-9.99 ~ 10
Pulmonary
Pa02 (mmHg) PI02 on MV or CPAP < 150 ~ 150 No ventilation
noCPAP
(Pa02[kPa]IFI02) (<19.9) (319.9) no IPAP
Hematologic
White blood cell count, x 10 9/L < 1.0 1.0-2.4 2.5-49.9 ~50.0
or and
Platelets, x 109/L <50 ~50
Hepatic
Bilirubin 1llI10llL (mgldL) < 34.2 « 2.0) ~ 34.2 (~ 2.0)
and or
Prothrombin time, s above « 25%) ~3(~25%) >3
standard (% of standard) U\
-...l
58 J. R. Le Gall et al.
Table 3. Variables and definitions for the logistic organ dysfunction (LOD) system
All variables must be measured at least once. If they are not measured, they are assumed to be
within the normal range for scoring purposes. If they are measured more than once in the first 24
h, the most severe value is used in calculating the score.
Neurologic System
Glasgow Coma Score: Use the lowest value; if the patient is sedated, the estimated Glasgow Co-
ma Score is 14 or 15.
Cardiovascular System
Heart rate: Use the worst value in 24 h, either low or high heart rate; if it varied from cardiac arre-
st (5 LOD points) to extreme tachycardia (3 LOD points), assign 5 LOD points.
Systolic blood pressure: Use the same method as for heart rate (e.g., if it varied from 60 to 250
mmHg, assign 3 LOD points). The patient is free of cardiovascular dysfunction if both heart rate
and systolic blood pressure are scored with 0 LOD points. This principle is the same for all organ
dysfunctions that may be definited by more than 1 variable.
Renal System
Serum urea or serum urea nitrogen level: Use the highest value in mmol/L or gIL for serum urea,
in mmol/L (mg/dL) of urea for serum urea nitrogen.
Creatinine: Use the highest value in flmol/L (mg/dL).
Urinary output: If the patient has been in the ICU for less than 24 h, make the calculation for 24 h
(e.g., 1 Ll8 h = 3 Ll24 h).
If the patient is on hemodialysis, use the pretreatment values.
Pulmonary System
If ventilated or under continuous positive airway pressure (CPAP), use the lowest value of the
Pa02lFI0 2 (fraction of inspired oxygen) ratio (whether Pa02 is mmHg or kPa). A patient who has
no ventilation or CPAP during the first day is free of pulmonary dysfunction.
Hematologic System
White blood cell count: Use the worst (high or low) white blood cell count that scores the highest
number of points.
Platelets: If there are several values recorded, find the lowest value and assign I LOD point if the
lowest value is less than 50 x 10 9IL.
Hepatic System
Bilirubin: Use the highest value in flIDol/L (mg/dL).
Prothrombin time (seconds or %): If there are several values recorded, assign I LOD point if the
prothrombin time was ever more than 3 s above standard or less than 25% of standard during the day.
Developing and validating the LOD model

The LOD score was first calculated for each of the 10,547 patients in the devel-
opmental sample by summing the points for each organ system based on the
recorded levels of each variable included in the system. The LOD score then
was used as the only term, along with a constant term, in a new logistic regres-
sion equation, resulting in a model that provided an estimate of the severity of
organ dysfunction as defined by the probability of hospital mortality.
The equation for the logit was = - 3.4043 + 0.4173 (LOD score).
This logit was the converted to a probability of hospital mortality for each
patient:
Pr (y = 1 logit) = e - 3.4043 + 0.4173 (LODscore)/1 + e - 3.4043 + 0.4173 (LODscore)
The goodness-of-fit and area under the ROC curve for this model were both
excellent in the developmental sample (Table 4), and the validation sample. The
probability of hospital mortality for each value of the LOD score is shown in
Table 5.
Table 4. Goodness-of-fit of the logistic organ dysfunction (LOD) model among 10,547 patients in
the developmental sample
Probability Survived, No Example 2

Observed Expected Observed Expected
0.00-0.032 51 1251.4 51 41.6
0.032-0.048 1781 1765.0 73 89
0.048-0.071 1172 1184.3 103 90.7
0.071-0.104 1108 1100.1 120 127.8
0.104-0.150 853 . 866.2 166 152.8
0.150-0.211 733 720.2 180 192.8
0.211-0.382 891 889.9 442 443.2
0.382-0.587 314 319.8 368 362.2
0.587-0.833 131 132.0 388 387.0
0.833-1.000 29 25.1 402 405.9
The steepest increases in the probability of mortality occur for LOD scores
from 1 to 32, with an approximative 10% increase in risk for each point increase
in the score. For an LOD score of 12 or more, the risk is over 80%; the risk
stays high but increase less rapidly as the score increases to the maximum of 22
points, which has an associated probability of mortality of 99.7%. There are
several scenarios by which a patient could receive an LOD score of 12 points or
more: either by several organs being involved at a moderate to severe level of
dysfunction or by the severity level of fewer organs being very high. In any such
scenario, the mortality risk is very high. Intermediate risk using the LOD Sys-
tem appears to occur in the range between approximately 5 to 10 points, and
there are numerous combinations of organs and severity levels that would result
in such a score.
The hypothetical patient described above had an LOD score of 9 points.
From Table 5, it can be seen that the probability of hospital mortality for that
patient would be 58.7%. While it is obvious that 3 organs were involved, the
LOD System weights the severity of dysfunction for the specific organ system
and provides a corresponding estimate of the probability of hospital mortality.
Use of the LOn system

The LOD score measures both the importance of the organ system relative to
the others and the degree of severity within that system. Most organ dysfunction
systems are scored with the worst severity level for each organ assigned the
same number of points, but giving the same number of points for a low GCS (5
LOD points) as for a high bilirubin level (1 LOD point) does not correctly char-
acterize patient's conditions.
Table 5. Conversion of the logistic organ dysfunction (LOD) score to a probability of hospital
mortality using the LOD model
Probability
LODScore
of Hospital Mortality, %
o 3.2
1 4.8
2 7.1
3 10.4
4 15.0
5 21.1
6 28.9
7 38.2
8 48.4
9 58.7
10 68.3
11 76.6
12 83.3
13 88.3
14 92.0
15 94.6
16 96.4
17 97.6
18 98.4
19 98.9
20 99.3
21 99.5
22 99.7
Of the 6 organ systems described by the LOD System, neurologic, cardiovas-

cular, and renal dysfunction were the most severe and received the maximum of
5 LOD points for the most severe level of dysfunction. Pulmonary and hemato-
logic dysfunction both received 3 points for the most severe level of dysfunc-
tion. Hepatic dysfunction received 1 point. It is notable that Fagon et al. found
that cardiovascular, renal, respiratory, and neurologic system dysfunction were
the most severe, while hematologic and hepatic system dysfunction were less
severe.
The most severe level of neurologic dysfunction, receiving 5 LOD points,
was defined by a GCS less than 6. Neurologic dysfunction was measured by the
actual GeS in patients who were not sedated patients. The criteria and weights
for neurologic dysfunction proposed for the MODS are somewhat similar to
those for the LOD System, although 4 levels of dysfunction are defined rather
than 3.
In the LOD System, cardiovascular system dysfunction could also be very
severe, with a state of severe shock adding 5 points to the LOD score. Adding
therapeutic measures such as the use of vasoactive drugs was not included in the
LOD definitions. The LOD score was developed using data from the first leU
day, and the physiological measurements represented patient's conditions prior
to therapy. The worst recorded values are those that receive the highest number
of LOD points. For example, if at different times on the first leU days a patient
has tachycardia of 150 beats per minute (l LOD point) and bradycardia of 25
beats per minute (5 LOD points), 5 points are added to the LOD score. After the
first leu day, when a patient is receiving continuous therapy, the problem of
scoring cardiovascular variables is, indeed, a difficult one. The variable pro-
posed for the assessment of cardiovascular dysfunction in the MODS is
pressure-adjusted heart rate (product of heart rate multiplied by the ratio of the
central venous pressure to the mean arterial pressure). This variable depends on
resuscitation and the use of blockers and pressors. The central venous pressure
is not recorded in all patients, which limits the value of this variable. Although
hypertension and bradycardia have not classicly been regarded as part of the
multiple organ system syndrome, they nevertheless reflect an abnormality in the
functioning of the cardiac system and were associated with a worse outcome
than was the case for patients without these factors in our study. This result sug-
gests that previous definitions of early cardiovascular dysfunction need to be
modified.
Renal dysfunction, as manifested by low urine output (oliguria) or high
serum urea levels, also receives 5 LOD points for the most severe level of dys-
function, which has been noted in other studies of renal dysfunction in intensive
care. There is no distinction between chronic and acute renal dysfunction in the
LOD scoring, as the focus is on the relevant physiological measurements with-
out having to rely on diagnostic assessments. Again, the decision as to what
constitutes the worst value is based on the number of points assigned. For exam-
ple, if a patient has oliguria of 0,4 LId (5 LOD points), 5 points are added to the
LOD score, regardless of the level of creatinine. To rely only on creatinine could
actually postpone the confirmation of renal dysfunction, since it may take sever-
al days to observe a rise in creatinine. In several assessment systems, serum cre-
atinine concentration is the only component of renal dysfunction measurement.
Serum urea or serum urea nitrogen, as well as daily urinary output, are mea-
sured in many countries and have a prognostic weight independent of creatinine.
The coefficients for both urea and urinary output demonstrated a stronger asso-
ciation with hospital mortality than the coefficients for creatinine, and when the
variables were considered in combination to define renal dysfunction, the asso-
ciation with mortality was even stronger.
62 J.R. Le Gall et a1.
Pulmonary dysfunction receives only 3 LOD points for the most severe leveL
Patients who have been assisted with neither ventilation nor CPAP are consid-
ered to be free of pulmonary dysfunction and receive 0 points towards the LOD
score. The PAOiFi02 ratio was also used in the MODS calculations to define
levels of pulmonary dysfunction; however, it was not clear whether all of their
population of 692 surgical patients were receiving mechanical ventilation,
which was not the case for the consecutive admissions that composed the ENAS
database.
Hematologic dysfunction also scores a maximum of 3 LOD points, with the
most severe level defined by a WBC count less than 1.0 x 1091L. This suggests
that a very low WBC count is not as strongly associated with mortality as the
most severe levels of dysfunction of other organs, all other things being equal.
The data for platelet counts were collected as a binary variable indicating only
whether platelet counts were low (less than 50 x 1091L), and this level of mea-
surement resulted in a severity level that receives only 1 LOD point. The MODS
uses only platelet counts, measured on a continuous scale, in the assessment of
hematologic system dysfunction. Platelet counts less than 50 x 1091L showed a
strong association with mortality in that study, consistent with the LOD System
categorization.
Hepatic dysfunction scores a maximum of I LOD point. This suggests that
early hepatic dysfunction by itself is not strongly associated with mortality, but
its occurrence in association with the dysfunctioning of the other organ systems
worsens the prognosis in an ICU patient. Unlike the MODS, hepatic dysfunc-
tion contributed the least to the scoring of multiple organ dysfunction in the
LOD System, allowing a maximum of 1 LOD point. Using PT to assess hepatic
dysfunction incorporated the measurement of a variable that may be abnormal
even when the bilirubin is within normal limits. Since our analysis was restrict-
ed to the first 24 hours in the ICU, it would be expected that hepatic dysfunction
would be more heavily weighted later in the ICU stay than during the first ICU
day. In future LOD research, data for platelet count and PT should be collected
as a continuous measurement to confirm whether the current cut points are best
suited to reflect the association with mortality.
Although developed using the same database, there are important differences
between the SAPSII and the LOD systems. The former takes into account not
only several physiologic parameters, but also includes age, the type of patient
admission, and several comorbidities. The LOD System was designed to char-
acterize 6 distinct organ systems and uses only physiological measurements to
do so. The information from the physiological measurements is grouped in a
manner that permits the characterization of organ dysfunction, both as to the
number of affected organs and the degree of dysfunction for each organ. In the
LOD System, 1 abnormal element is sufficient for the classification of organ
dysfuction.
Multiple organ dysfunction is not necessary for the application of the LOD
System, as it applies to both single or multiple organ dysfunctions. This makes
the LOD System more broadly applicable, since less than one third of patients
in an leU may have 2 or more organs in dysfunction, with the majority having
only 1 organ dysfunction. In our database, which comprised several tertiary
care units, 26% of patients had 1 organ dysfunction, and 62% had 2 or more or-
gans in dysfunction. The concept of multiple organ dysfunction implies the in-
volvement of multiple organs, rather than a single organ, but the LOD System
grades organ dysfunction in such a way that severity due to organ dysfunction
can be quantified, whether 1 to 6 organs are involved. As for the severity scores
[15, 16], an objective organ dysfunction system is probably superior to the pre-
vious ones.
Conclusion
The proposed LOD System, which was developed using statistical methods that
determined the relative weights of the several organ systems and of the levels of
severity within each organ system, also produces an estimate of the risk of mor-
tality that demonstrated excellent calibration and discrimination. The LOD Sys-
tem can be used for the assessment of organ dysfunctions developing later in the
leU stay, particularly among elective surgery patients who are often free of or-
gan dysfunction when admitted to the leu. The large ENAS database from
which the system was developed, however, comprised a mix of medical patients,
emergency surgical patients, and elective surgical patients who manifested mea-
surable levels of organ dysfunction on the first leU day. Many of the leUs in
the ENAS database were tertiary care units, and patients entered them at a rela-
tively advanced stage of disease, as reflected by the 62% of patients with 2 or
more organs in dysfunction on the first day in the leU.
Any system to assess organ dysfunction that uses first leU day variables
must be validated for use at other time periods, including the LOD System, and
future studies must be designed for that purpose. The validity of the estimate of
the probability of mortality from the conversion of the LOD score to a probabil-
ity using the LOD model on subsequent leU days has not been tested.
Studies of the association with mortality of an LOD score that is collected
daily in the leU need to be undertaken. Also, further studies that take into ac-
count the duration of dysfunction will be needed to estimate the probability of
hospital mortality at later points in the leU stay. Duration of dysfunction is
commonly associated with a worsening prognosis, even if a patient's condition
is unchanging, since absence of improvement is a negative sign.
Having a tool to quantify the severity of organ dysfunction is necessary in
order to evaluate the effectiveness of treatment, not only on mortality but on the
resolution of organ dysfunction. Successful resolution of organ dysfunction,
however, has not been clearly defined. Some researchers have proposed the
number of days free from organ dysfunction as a marker of resolution for an
outcome measure. For such a purpose, the number of days with a zero for the
LOD score could be calculated.
As with all models designed for use in a dynamic and changing environment,
the LOD System must be kept up-to-date and applicable in the face of changing
case mix and ICU therapies. In its present form, the LOD System we have pro-
posed is based on objectively derived coefficients that weight the severity of or-
gan dysfunction differentially both among the 6 organ systems and within each
organ system. The results of our analysis by defining the relative association of
levels of severity with hospital mortality, suggest that the LOD System has great
potential as a tool with which to assess the real severity of organ dysfunction
among general medical and surgical ICU patients.
References
1. Fry DE, Pearlstein L, Fulton RL, Polk HC Jr (1980) Multiple system organ failure: the role
of uncontrolled infection. Arch Surg 115: 136-140
2. Stevens LE (1983) Gauging the severity of surgical sepsis. Arch Surg 118:1190-1192
3. Marshall JC, Christou NV, Horn R, Meakins JL (1988) The microbiology of multiple organ
failure: the proximal gastrointestinal tract as an occult reservoir of pathogens. Arch Surg 123:
309-315
4. Knaus WA, Draper EA, Wagner DP, Zimmermann JE (1985) Prognosis in acute organ system
failure. 202:685-693
5. Knaus WA, Wagner DP (1989) Multiple systems organ failure epidemiology and prognosis in
critical care clinics. In: Pinsky MR, Matuschak GM (eds) Multiple systems organ failure. WB
Saunders, Philadelphia, pp 221-232
6. Fagon JY, Chastre J, Novara A et al (1993) Characterization of intensive care unit patients us-
ing a model based on the presence or absence of organ dysfunction and/or infection: the
ODIN model. Intensive Care Med 19:137-144
7. Hebert RC, Drummond AJ, Singer Jet al (1993) A simple multiple organ system failure scor-
ing system predicts mortality of patients who have sepsis syndrome. Chest 104:230-235
8. Marshall JC, Cook DJ, Christou NV et al (1995) Multiple organ dysfunction score: a reliable
descriptor of a complex clinical outcome. Crit Care Med 23:1638-1652
9. Le Gall JR, Lemeshow S, Saulnier F (1993) A new simplified Acute Physiology Score (SAPS
II) based on a European/North American multicenter study. JAMA 270:2957-2963
10. Lemeshow S, Le Gall JR (1994) Modeling the severity of illness of ICU patients: a system
update. JAMA 272: 1049-1055
11. Hosmer DW, Lemeshow S (1989) Applied logistic regression. John Wiley & Sons, New York, NY
12. Le Gall JR, Klar J, Lemeshow S et al for the ICU scoring group (1996) The logistic organ
dysfunction system: A new way to assess dysfunction in the intensive care unit. JAMA 276:
802-810
13. Hanley JA, McNeil BJ (1982) The meaning and use of the area under a receiver operating
characteristic (ROC) curve. Radiology 143:29-36
Will the Use of Scores Benefit the Individual Patient
and Improve Therapy?
R. LEFERING, R.J.A. GORIS
Since the introduction of a 10 point scale for newborn infants [1] in 1953 by Ap-
gar numerous scores and scales have been developed in medicine. A score is an
attempt to integrate many different information like blood pressure, heart rate or
laboratory data into a single one-dimensional numerical value. The more com-
plex a clinical situation is the more efforts are made to develop and refine score
systems. Therefore, many scores for patients in the intensive care unit (leU)
have been published.
Scores should help to classify diseases or health states, facilitate compar-
isons to other cases, or enhance decision-making. Scores are used to predict fu-
ture events like prognosis or outcome, and economists and administrators use
scores to achieve a justifiable reimbursement policy.
The central point of a score system is the reduction and summary of diverse
clinical data. The reduction of information can help to concentrate on the essen-
tials, but this reduction naturally causes a loss of information. Many different
clinical situations are projected on the same score value. Thus a score is like a
pair of glasses giving a selective view of the patient's clinical situation. Regard-
ing an individual patient, this sacrifice of information implies that scores will
never replace the underlying data but provide a context allowing interpretation.
In the following paragraphs, clinical situations where scores might have a di-
rect effect on the individual patient are presented. Knowledge about the prob-
lems and limitations of score systems is a prerequisite for their application.
Indirect and direct effects in the individual patient

The indirect benefit of score systems is widely known and demonstrated in
many papers. First of all, research results can be presented much easier and re-
sults of clinical trials become more objective and reproducible if scores were
used (inclusion criteria, description of patients, assessment of outcome). Sec-
ond, scores can be used to perform quality audits. The actual outcome in a
group of patients can be compared to what is expected in these patients where
the expected outcome is based on a severity classification with scores. Third,
66 R. Lefering, R.J.A. Goris
dealing with scores undoubtedly has also an educational effect. Especially

younger physicians will benefit from these considerations in teaching rounds.
The above mentioned use of scores may indirectly affect patient care by in-
troducing a new therapy or enhancing the level of education. A direct effect is
present if a therapeutic or diagnostic procedure is induced (or withdrawn, or
denied) due to a score value, which would not have been done without this in-
formation. Direct effects of scores in intensive care may occur in the following
situations:
- triage
- initiation of procedures
discharge from the ICU
termination of therapy.
Triage decisions are well-known in catastrophes or in the battlefield where
the discrepancy between availability and need of care is usually huge. But simi-
lar considerations are sometimes necessary in intensive care. The Society of
Critical Care Medicine Ethics Committee defined "triage" as "finding the most
appropriate disposition for a patient based on an assessment of the patient's ill-
ness and its urgency" [2]. They further state that "priority for admission to an
ICU should correlate with the likelihood that ICU care will benefit the patient
substantially more than non-ICU care. Patients with poor or very good progno-
sis should not be admitted" [2]. It is very imaginable that decisions to admit a
patient to an ICU or discharge a patient to a normal ward can be supplemented
by a severity of illness classification based on scores: "Triage involves a com-
mon logic that includes probability estimates of outcome" [2]. On the other
hand it is clear as well, that those estimates based on scores will never be the
only criterion for admission or discharge policies.
The initiation of certain procedures is a further area where scores might have
a direct influence. Murray and coworkers investigated whether provision of
computer-based prediction of outcome would alter patient management [3].
1025 ICU patients, admitted after severe head injury, were studied during three
time periods. The probability for survival was calculated. During the baseline
phase the outcome estimate was withheld, while nurses and physicians were
given this information in tRe second phase. Phase three was again without pre-
diction in order to evaluate the continuation of behavioural changes. Aspects
like intubation and/or ventilation, administration of osmotic agents, or intracra-
nial pressure monitoring were chosen as indicators for possible changes in pa-
tient management. They found that the average frequency of use of these proce-
dures was similar or slightly increased in patients with good and moderate prog-
nosis, while there was a substantial decrease of 39% in patients with a poor
prognosis (risk of death> 80%). This tendency even continued after the com-
puter prognosis had been withdrawn in phase three. Thus outcome predictions
derived from score values might well have a direct influence on the initiation of
certain procedures.
Will the Use of Scores Benefit the Individual Patient and Improve Therapy? 67
Besides the initiation of certain procedures the termination or withdrawal of

therapy in critically ill patients is another point of controversial discussion. In-
tensive care is costly, resources are limited, and the number of patients needing
intensive monitoring and therapy increase. But beyond economic considerations
it is an ethical question whether it is justified to artificially prolong dying. Tech-
nological advances in intensive care can prolong patients' lives but also their
suffering. Therefore it is a challenge to identify hopelessly ill patients who will
not benefit from further maximal therapy [4]. But how can one identify those
patients? Chang et al. published a prediction model based on slightly modified
daily Apache II assessment [5, 6]. All patients predicted to die in his test set (n =
112) and validation set (n = 212) actually died so that he had no false positive
predictions. The published criteria were subsequently tested by other investiga-
tors. Rogers et al. found in 3350 ICU patients that actual mortality rates among
patients predicted to die vary between 48% and 69% only (depending on the cri-
terion used) [7]. Our own observations in 1575 ICU patients identified 42 pa-
tients predicted to die according to the Chang criteria. 40 of them actually died
(95%), but two patients survived.
Score values can give only probability estimations of outcome based on pre-
vious observations of many similar cases [8]. A 100% certainty is impossible.
Therefore, decisions to withdraw treatment will never be based on a score value
alone. The individual situation including the patient's wishes and preferences as
well as the expected benefit from further therapy have to be taken into account
seriously [2]. The score-based outcome prediction is only one small part in the
complex puzzle of the decision process.
Limitations and problems

Any method, including scores, has some inherent limitations. Like positive
blood culture results may be due to contamination of the sample, score values
can also be misinterpreted. Those who apply scores especially in the individual
patient should well know about these problems.
Interpretation
Most severity of disease classification systems like Apache II, SAPS, etc. can be
transformed into a risk of death estimate. Applied to an individual patient, this
prognosis may cause confusion, especially if a patient with a good prognosis ac-
tually dies. A 10% risk of death estimate does not exclude a life-threatening sit-
uation. It only means that only one out of ten cases with a similar score value
will die. The prognostic information of a score is a probability based on previ-
ously observed cases with similar score values. Therefore, the validity of its
prognosis can only be tested in a group of patients.
68 R. Lefering, R.J.A. Goris
Fixed components
A score value usually is composed of several items, identified by experts or sta-
tistical procedures, each adding a predefmed value to the total sum. For exam-
ple, a serum potassium below 0.6 mMollL adds 2 points to the total Apache II
score. But in a specific clinical situation a certain derangement may be much
more important for survival than suggested by two score points. A scoring sys-
tem that considers all clinical circumstances would require a rather complex
model and would still miss some unforeseen situations. Therefore, a score value
will always need a clinical interpretation.
Applicability
A structured collection of data is the first step in the development of a scoring
system. Patients with a certain disease or other similar characteristics (e.g. ad-
mission to ICU, trauma) are documented, and with statistical methods a formula
for a score is derived from the data. The application of that score to new patients
will yield most reliable results if these patients are as comparable as possible to
the initial group of patients. The use in patients with a different disease as well
as in a small subgroup of patients may cause a substantial bias. The Apache II
score, for example, was developed in a general ICU population. Its application
to trauma patients only causes a well-documented under-estimation of risk of
death [9]. But even if patients are comparable, the transfer of a score into a dif-
ferent setting (e.g. in another country) may cause inaccuracy. In general, the ap-
plication of a score system primordially requires a validation study.
Validy of measurement
A score is only as good as its components. If some measurements are very inaccu-
rate or show a high inter-rate variability, the final score value will have the same
shortcomings. A good example is the Glasgow Coma Score (GCS) in intensive
care patients. This score was developed by Teasdale and Jennett to describe the
severity of a head injury [10]. In these patients it has a high prognostic value. But
most ICU patients are intubated and sedated, and a valid assessment of their con-
sciousness is impossible. In this situation, one can a) record the actual state re-
gardless of being influenced by drugs (i.e. 3 points), b) continue to document the
last valid measurement (e.g. 8 points); c) estimate at best guess the presumed pre-
sent state without sedation (e.g. 11 points), or d) regard the present assessment of
consciousness as missing and thus deny a discount (i.e. 15 points). Although solu-
tion d) was emphasized by the authors of Apache II, all four will be found in the
real world. The GCS is also part of the Multiple Organ Failure (MOF) score from
Goris et al. [11], but as a result of a recent validation study in different centers it
was decided to omit the GCS assessment because of its inaccuracy in intensive
care patients.
Will the Use of Scores Benefit the Individual Patient and Improve Therapy? 69
Overestimation of accuracy
The ability of a score to correctly predict e.g. mortality, usually is expressed in
terms of sensitivity (proportion of patients correctly predicted to die) and speci-
ficity (proportion of survivors correctly predicted to survive). These estimates
are based on previous samples of ICU patients. But accuracy measures like sen-
sitivity and specificity can strongly be influenced by the composition of the
sample they are based on. If many patients with a low risk such as patients for
short-term postoperative surveillance are contained in a sample of ICU patients,
specificity might artificially be increased due to an "easy" prognosis of survival.
In difficult clinical situations where clinicians need some additional information
scores perform much worse than expected from their published accuracy. This is
a further limitation of scores for use in the individual patient.
Conclusion
The scientific assessment and the practical use of score systems will undoubted-
ly continue or even increase. Scores can facilitate comparisons, enhance quality
management, and improve clinical research. The usefulness of scores in the in-
dividual patient is limited due to the fact that scores reduce many dimensions of
a clinical situation into a single value. Therefore, a score may supplement but
will never replace clinical judgement, and those who use scores as a clinical
routine should well know about the problems and limitation of scores.
References
1. Apgar V (1953) A proposal for a new method of evaluation of newborn infant. Curr Res
Anesth 32:260
2. Society of Critical Care Medicine Ethics Committee (1994) Consensus statement on the triage
of critically iII patients. JAMA 271: 1200-1203
3. Murray LS, Teasdale GM, Murray GD et al (1993) Does prediction of outcome alter patient
management? Lancet 34 I: 1487-1491
4. Atkinson S, Bihari D, Smithies M et al (1994) Identification of futility in intensive care.
Lancet 334: 1203-1206
5. Chang RWS, Jacobs S, Lee B (1988) Predicting outcome among intensive care unit patients
using computerized trend analysis of daily Apache II scores corrected for organ system failure.
Intens Care Med 14:558-566
6. Chang RWS (1989) Individual outcome prediction models for intensive care units. Lancet
ii:143-146
7. Rogers J, Fuller HD (1994) Use of daily Acute Physiology and Chronic Health Evaluation
(APACHE) II scores to predict individual patient survival rate. Crit Care Med 22:1402-1405
8. Lemeshow S, KIar J, Teres D (1995) Outcome prediction for individual intensive care patients:
useful, misused, or abused? Intens Care Med 21 :770-776
9. Lefering R, Dicke S, Bottcher, Neugebauer E (1997) Der Apache II Score bei Traumapatienten
- Eine systematische Unterschatzung der Prognose. Intensivmed 34:426-431
10. Teasdale G, Jennett B (1974) Assessment of coma and impaired consciousness. Lancet ii:
81-84
11. Goris RJA, te Boekhorst TPA, Nuytinck JKS, Gimbrere JSF (1985) Multiple organ failure.
Generalized autodestructive inflammation? Arch Surg 120: 1109-1115
I MEDIATORS AND PROSTAGLANDINS I
Old and New Mediators in Sepsis
The Enigma Is Not Yet Resolved
~.~ONELLI,~.PASSAJUELLO
Sepsis may be defined as a systemic inflammatory response to an infection. The

syndrome is characterized by alterations in temperature, leukocytosis, hypoten-
sion, hypoperfusion with cellular injury and organ failure often resulting in
death.
The incidence of sepsis and sepsis related conditions has continued to in-
crease during the last 30 years with a mortality rate of 35% [1].
Sepsis may be associated with infection by all classes of micro-organisms in-
cluding Gram positive and Gram negative bacteria, fungi, parasites, protozoa
and viruses.
These agents induce the synthesis and release of endogenous inflammatory
and immunomodulating cytokines, which are essential to modulate the host re-
sponse to infections.
Release of cytokines
TNF alpha and Interleukin beta (IL-l beta) are known to exert an important pro-
inflammatory action, activating neutrophils [2] and coagulation [3], stimulating
neutrophil adhesion to the endothelium [4] and inducing apoptosis [5].
After endotoxin challenge TNF is the first cytokine to appear [6], peaking 90
minutes after endotoxin injection with transient release that varies in a wide
range of concentrations (from few to thousands of pg/ml). Following the release
of TNF, the circulating concentrations of other cytokines as IL-6, IL-8
macrophages inflammatory protein (~IP-l alpha), monocyte chemoattractant
protein-l (~CP-l) increase as well [7, 8].
TNF is then responsible for the release of these secondary cytokines. Neu-
tralisation of TNF activity in experimental endotoxemia prevented the release of
IL-6 and IL-8 [9].
Blocking IL-6 activity does not affect the induction of other cytokines, in-
dicating that IL-6 and IL-8 are more distal agents in the inflammatory cas-
cade [10].
74 M. Antonelli, M. Passariello
TNF and IL-6 were also shown to be implicated in the local inflammatory
reaction in course of septic and non septic ARDS [11].
IL-8 can be produced from several types of cells, specifically: endothelial
cells, macrophages, and polymorphonuclear leukocytes. It induces chemotaxis
and activation of neutrophils.
The primary stimulus for this circulating IL-8 is not fully understood, but IL-
l and TNF are important inducers of IL-8 expression in many cell types [12].
IL-6 is a cytokine that can be produced by macrophages, lymphocytes, en-
dothelial cells, and other tissues; it plays an essential role during the acute
phase of the inflammatory response, acting synergistically with other com-
pounds [13]. This peptide has been shown to be predictive for outcome in sep-
tic patients [13, 14].
Other cytokines such as IL-I0 and granulocyte colony stimulating factor (G-
CSF) are also released after sepsis or endotoxin administration [15, 16].
In animal models of sepsis G-CSF pre-treatment reduces endotoxin induced
mortality and organ failure [17].
In humans G-CSF has been found to be beneficial in patients with pneumo-
nia, bum injury and major trauma [18, 19].
The mechanism by which G-CSF regulates inflammatory reaction is not yet
fully understood. In recent studies on human endotoxemia, G-CSF increased
both pro- and anti-inflammatory responses [20].
In sepsis and experimental endotoxemia, circulating levels of IL-I0 are in-
creased, especially in meningococcal septic shock [21].
IL-IO seems to playa protective role in endotoxemia, as neutralisation of en-
dogenously produced IL-lO resulted in increased mortality in endotoxin chal-
lenged mice. In humans administration of IL-I0 causes a suppression of TNF
and IL-l beta production, while the production of their receptor antagonists re-
mained unaltered [22].
Leukocyte activation
The emergence of leukocytes from the vascular to the interstitial compartment
necessitates that the leukocytes come in contact with the blood vessel wall. Ini-
tially the leukocyte must be displaced from the blood stream towards the periph-
ery of the vessels, a process which is mediated by the radial dispersal forces
within the blood vessels [23]. In vivo observations suggest that this process of
margination requires the interaction between red blood cells and leukocytes
[24]. Following their margination, leukocytes undergo a series of adhesive inter-
actions that starts with a rolling movement along vessel endothelium and is fol-
lowed by a firm adhesion and subsequently, migration through the vessel wall
(Fig. 1).
Old and New Mediators in Sepsis - The Enigma Is Not Yet Resolved 75
Rolling Adherence Migration
L selectin CDlllCD18 (beta integrins)
®
P and E selectin ..
ILS,n.6, PM, G-CSF
Fig. 1. After inflammatory stimulation the expression of selectins on endothelial cells increases,
inducing an enhanced contact between endothelium and neutrophils. Leukocyte integrins promote
the adherence of neutrophils onto the endothelial layer. Migration of neutrophils is induced by
chemotactic agents as interleukins 8 and 6, platelet activating factor and granulocyte colony stim-
ulating factor, etc
Leukocyte rolling in post-capillary venules may be enhanced by many sub-

stances and pathological conditions. Leukocytes rolling is mediated by a family
of adhesion molecules located on both endothelial cells and leukocytes, known
as selectin family of cell adhesion receptors. This family includes E and P se-
lectin, which are expressed on the surface of stimulated endothelial cells. L se-
lectins are expressed on the surface of all leukocytes [25]. The selectins are
structurally similar carbohydrate-binding lecitins.
Firm adhesion between leukocytes and endothelial cells is mediated by the
supergene immunoglobulins located on the surface of endothelial cells and their
ligands, beta 2 integrins (CDII/CDI8), which are expressed on most leukocytes
[26]. Intracellular adhesion molecule I (ICAMI) has been implicated as a key
modulator of leukocyte recruitment in several inflammatory models. The two al-
pha subunits of CDII/CDI8, CDIIa and CDIIb, have been proposed as ligands
for ICAMl. After intravenous endotoxin administration and following the stim-
ulation by TNF, or chemotactic factors such as IL-8, G-CSF and platelet activat-
ing factor (PAP), CDllb/CDI8 expression and activation on neutrophils In-
creases [27].
Activation of the coagulation

A severe complication of sepsis is disseminated intravascular coagulation
(DIC), which results from a deranged activation of the coagulation system with
enhanced activation of coagulation, depression of the inhibitory mechanisms of
coagulation, and inhibition of the fibrinolytic system.
Moreover, the consumption of clotting factors and platelets, due to the con-
tinuous activation of the coagulation system, may lead to severe bleeding disor-
ders. The coagulopathy observed in sepsis have been well defined in studies of
endotoxin and TNF-infused volunteers.
The common pathway of the coagulation system can be activated through the
intrinsic and extrinsic way.
The intrinsic route is considered to result in the production of inflammatory
mediators as bradykinin, which may be involved in the onset of hypotension
during bacteremia [28].
The activation of the intrinsic pathway is well demonstrated in a consider-
able number of septic patients, however in human experimental endotoxemia re-
sults appear contradictory.
Animal studies of experimental endotoxemia have shown that TNF is a main
mediator of endotoxin induced activation of fibrinolysis and IL-6 is relevant for
the induction of the procoagulant response during endotoxemia [29].
Derangement of platelet count and activity represents another hallmark of
septic condition. Hypopiastrinemia is in fact common during sepsis.
Furthermore, a recent clinical study have demonstrated that in septic patients
platelets become activated and are hyperadhesive to other vascular cells includ-
ing neutrophils and endothelial cells. This seems to induce sequestration of
platelets and microcirculatory arrest, thus the development of multiple organ
failure [30]. .
Apoptosis
Apoptosis is a form of cellular death, effected through the expression of an en-
dogenous, genetically-regulated program, that results in the rapid elimination of
the cell without evoking a significant inflammatory response. First recognised
more than a century ago, this phenomenon is now believed to be central to the
normal development of multi-cellular organisms, mediating such phenomena as
tissue remodelling during embriogenesis, the deletion of autoreactive T cell
clones during immune maturation, and the elimination of transformed cells by
natural killer cells. It is also the mechanism through which cells with a fixed life
span, such as epithelial cells and neutrophils die.
Early features of apoptosis include cytoplasmic blebbing, loss of normal
cell-cell adbesion and elimination of specialized structures such as microvilli
APOPTOSIS nonnal cell
I
a
Nuclear and cell shrinkage
I Nuclear fragmentation and fonnation
of apoptotic bodies
Ingestion of bodies by phagocyte
Fig. 2. Cell dying of apoptosis condenses and is phagocytosed by local macrophages, without in-
ducing inflammation. The absence of a local response and the rapid evolution render the process
less evident hystologically
(Fig. 2). In contrast to the cell dying a necrotic death, the apoptotic cell shrinks,
its chromatin condenses and the nucleus fragments as a result of internucleoso-
I
mal cleavage of DNA. Apoptosis occurs rapidly, and once triggered it is com-
plete in 45 minutes.
Signalling pathways for apoptosis are incompletely defined. Generation of a
ceramide from membrane sphingomyelin constitutes a key second messenger
system leading to apoptosis [31]. Ceramide has also been reported to activate
MAP kinase (MAPK) [32], and to mediate Fas- and TNF-induced apoptosis
through activation of SAP kinase (SAPK).
Although apoptosis in critically ill patients has not been extensively studied,
the available data show interesting abnormalities.
Lymphopenia is in fact common concomitant with trauma in critical illness.
Studies in patients with major thermal injury show increased rates of apoptosis
in peripheral blood lymphecytes [33].
Granulocyte macrophage-colony stimulating factor (GM-CSF) seems to
modulate cell survival in systemic circulation [34].
An evolving understanding of the role of inflammatory cell apoptosis in the
expression and termination of the host response to acute life threatening stimuli
should provide the intensive care knowledge with new weapons to modulate the
inflammation of the critically ill patient.
Conclusion
Due to the continuous improvement of knowledge on molecular and biological
mechanisms of sepsis and sepsis related problems (only partially exposed in the
present review), all researchers and clinicians fell victim of the illusion to be
able of modulating the therapeutic approach.
Indeed, our present understanding of sepsis and organ failure needs to be re-
vised, as the negative results of new therapies for these disorders suggest. Previ-
ous theories for the pathogenesis of these situations are incomplete for different
reasons:
1. The models used to study these alterations are not analogous to the clinical
situation.
2. Patients with less severe manifestations are often overlooked.
3. Patient's pre-existing conditions are not enough taken into account.
A considerable body of evidence indicates that together an important proin-
flammatory reaction, an anti-inflammatory response contributes to the onset of
sepsis and organ failure.
At a local site of injury or infection and during the initial appearance of pro-
and anti-inflammatory mediators in the circulation, the beneficial effects of
these compounds counterbalance their harmful effects. Only when the balance
between these two forces is lost these substances may become harmful. The se-
quelae of an unbalanced systemic inflammatory reaction include derangement
of microcirculation, shock, transudation into organs and defects of coagulation.
An unbalanced systemic compensatory anti-inflammatory response often results
in anergy and immunosuppression. The proinflammatory and anti-inflammatory
conditions may ultimately reinforce each other, creating a state of destructive
immunologic dissonance [35]. At the present time all the therapies that imply
the block of a mediator or an inflammatory cascade should be carefully revised,
as our present knowledge does not offer an adequate guarantee for their applica-
tion in clinical practice.
References
1. Kreger BE, Craven DE, McCabe WR (1980) Gram negative bacteremia IV. Re-evaluation of
clinical features and treatments in 612 patients. Am J Med 68:344-355
2. Fujishima S, Aikawa N (1995) Neutrophil-mediated tissue injury and its modulation. Inten-
sive Care Med 21:277-285
3. Levi M, ten Cate H, van der Poll T, van Deventer SJH (1993) Pathogenesis of disseminated
intravascular coagulation in sepsis. JAMA 270:975-979
4. Gamble JR, Harlan JM, Klebanoff SJ, Vadas MA (1985) Stimulation of the adherence of neu-
trophils to umbilical vein endothelium by recombinant TNF. Proc Natl Acad Sci USA 82:
8667-8671
5. Verheij M, Bose R, Xin HL et al (1996) Requirement for ceramide-initiated SAPKlJNK sig-
nalling in stress-induced apoptosis. Nature 380:75-79
6. Michie HR, Manogue KR, Spriggs DR et al (1988) Detection of circulating tumor necrosis
factor after endotoxin administration. N Engl J Med 318: 1481-1486
7. Matrich GO, Danner RL, Ceska M, Suffredini AF (1991) Detection of IL-8 and TNF in nor-
mal humans after intravenous endotoxin. The effect of anti-inflammatory agents. J Exp Med
173: 1021-1024
8. Sylvester I, Cuffredini AF, Boujoukos AJ et al (1993) Neutrophil attractant protein 1 and
monocyte chemoattractant protein 1 in human serum. J Immunol 151 :3292-3298
9. Van der Poll T, Levi M, van Deventer SHJ et al (1994) Differential effects of anti-TNF anti-
bodies on systemic inflammatory responses in experimental endotoxemia in chimpanzees.
Blood 83 :446-451
10. Van der Poll, Levi M, Hack Ce et al (1994) Elimination of IL-6 attenuates coagulation activa-
tion in experimental endotoxemia in chimpazees. J Exp Med 179:1253-1259
11. Raponi G, Antonelli M, Gaeta G et al (1992) TNF in serum and in BAL of patients at risk for
ARDS. J Crit Care 7:183-188
12. Finn A, Naik S, Klein N et al (1993) Interleukin-8 release and neutrophil degranulation after
paediatric cardiopulmonary bypass. J Thorac Cardiovasc Surg 105:234-241
13. Bauer J, Herrmann F (1991) Interleukin-6 in clinical medicine. Ann Hematol 62:203-207
14. Antonelli M, Raponi GM, Martino P et al (1995) High IL-6 serum levels are associated with
septic shock with severe leUkopenia due to malignancies. Scand J Infect Dis 27:381-384
15. Suffredini AF, Reda 0, Banks SM et al (1995) Effects of recombinant dimeric TNF receptors
on human inflammatory responses following intravenous endotoxin administration. J Im-
munol 155:5038-5045
16. Von der Mohlen MAM, Kimmings NA, Wedel N et al (1995) Inhibition of endotoxin-induced
cytokine release and neutrophil activation in human using a recombinant endotoxin binding
protein (rBPI23). J Infect Dis 172: 144-151
17. Gorgen I, Hartung T, Leist M et al (1991) Granulocyte colony-stimulating factor treatment

protects rodents against LPS-induced toxicity via suppression of systemic TNF alpha. J Im-
munoI149:918-924
18. Sartorelli KH, Silver GM, Gamelli RL (1991) The effect of C-GSF upon burn-induced defec-
tive neutrophil chemotaxis. J Trauma 31:523-529
19. Nelson S, Farkas S, Fotheringham N et al (1996) Filgrastin in the treatment of hospitalized
patients with community acquired pneumonia. Am J Respir Crit Care Med 153 (abs)
20. Pollmacher T, Korth C, Mullington J et al (1996) Effects of G-CSF on plasma cytokine and
cytokine receptor levels and on the vivo host response to endotoxin in healthy men. Blood
87:900-905
21. Derkx B, Marchant A, Goldman M et al (1995) High levels of IL-IO during the initial phases
of fulminant meningococcal septic shock. J Infect Dis 171:229-232
22. Hun RD, Radwaski E, O'Connell SM et al (1996) Pharmacokinetics and immunomodulatory
properties of intravenous administred recombinant human IL-I0 in healthy volunteers. Blood
87:699-705
23. Schmid-Schonbein GW, Usami S, Skalak R et al (1980) TIie interaction of leukocytes and
erythrocytes in capillary and postcapillary vessels. Microvasc Res 19:45-70
24. Mayrovitz HN, Rubin R (1985) Leukocyte distribution to arteriolar branches: Dependence on
microvascular blood flow. Microvasc Res 29:282-294
25. Kishimoto TK, Jutilia MA, Berg EL et al (1989) Neutrophil Mac.1 and MEL-14 adhesion
proteins inversely regulated by chemotactic factors. Science 245:1238-1241
26. Anderson DC (1995) The role of Beta 2 integrins and intracellular adhesion molecule type 1
in inflammation. In: Granger DN, Schmid, Schonbein (eds) Physiology and pathophysiology
of leukocyte adhesion. 1st edn. Oxford University Press, New York, pp 3-42
27. Windsor ACJ, Walsh CJ, Mullen PG et al (1993) TNF alpha blockade prevents neutrophil
CD18 receptorupregulation and attenuates acute lung injury in porcine sepsis without inhibi-
tion of neutrophil oxygen radical generation. J Clin Invest 91: 1459-1468
28. Pixley RA, de la Cadena R, Page JD et al (1993) The contact system contributes to hypoten-
sion, but not to DIC in lethal bacteremia. J Clin Invest 91:61-68
29. Van der Poll T, Levi M, Hack CE et al (1994) Elimination ofIL-6 attenuates coagulation acti-
vation in experimental endotoxemia in chimpanzee. J Exp Med 179:1253-1259
30. Gawaz M, Dickfeld T, Bogner et al (1997) Platelet function in septic multiple organ dysfunc-
tion syndrome. Intensive Care Med 23:379-385
31. Hannun YA, Obeid LM (1995) Ceramide: an intracellular signal for apoptosis. Trends
Biochem Sci 20:386-401
32. Wiegmann K, Scheutze K, Machleidt T et al (1994) Functional dichotomy of neutral and
acidic sphingomyelinases in TNF signalling. Cell 78:I005-1015
33. Teodorczyk-Injeyan JA, Cembrznska-Nowak M, Llani S et al (1995) Immune deficiency fol-
lowing thermal trauma is associated with apoptotic cell death. J Clin Immunol 15:318-328
34. Chitnis DC, Dickerson C, Munster AM et al (1996) Inhibition of apoptosis in polymorphonu-
clear neutrophils from bum patients. J Leuk BioI 59:835-839
35. Bone (1996) Immunologic dissonance: a continuing evolution in our understanding of the
systemic inflammatory response (SIRS) and the multiple organ dysfunction syndrome
(MODS). Ann Intern Med 125:680-687
Modulating Prostaglandin Metabolism in
M.R. PINSKY
Sepsis is associated with a cacophony of activation and inhibi1

aspects of the host's inflammatory pathways. Because sepsis is
by pro-inflammatory stimuli such as pancreatitis, trauma, endt
ous studies have focused on anti-inflammatory therapies in th{
patients with sepsis and septic shock. The mechanisms by whic
processes induce inflammation and their regulation are central
Endotoxin infusion in the dog produces a generalized toxen
is characterized by a shock state associated with diffuse org:
lactemia, and altered circulating levels of formed blood elemer
morphonuclear leukocytes (PMN) and thrombocytes [1]. We
shown that the associated generalized peripheral vascular para]
tation-induced hyperdynamic state appear to reflect a loss of s)
tone [22] occurring too early to be due entirely to nitric oxide (I
inducible nitric oxide synthase [34]. Endotoxin induces thes(
namic effects through a complicated process that includes the •
lease of inflammatory substances of the cyclooxygenase or lipc
ways [2]. Although prior activation of pro-inflammatory medi,
in a more proximal position in the inflammatory cascade appe,
ble for the initiation of the inflammatory response [3] inhibitio
nase and lipooxygenase pathways can prevent many of the hem
of endotoxin administration in sheep [3, 4]. Presumably these
occur despite an initial endotoxin-induced activation of both I
cytokines and white blood cells [5]. However, the mechanism
cyclooxygenase or lipooxygenase inhibition alters the inflamm:
endotoxin is unclear. We have previously shown that pretreatrr
fen, a reversible cyclooxygenase inhibitor, though abolishing th
endotoxemia on blood pressure and cardiac output, does not pr
ated increasing in 02 consumption or lactic acidosis [21]
processes in the expression of the endotoxic state may be contre
ferent mediators.
The systemic manifestations of endotoxemia are essentially
host in response to endotoxin through the synthesis and release
sis factor-a (TNF-a) [6]. TNF-a both stimulates further cytokil
82 M.R. Pinsky
has direct cytotoxic effects in vitro [7]. Furthermore, TNF-a infusion can mimic
many of the hemodynamic effects of acute endotoxemia in the dog [8]. After
TNF-a release, further stimulation of the monocyte pool results in the synthesis
and release of numerous other cytokines that have diverse regulatory functions
within the inflammatory process [9]. Interleukin-6 (IL-6) appears to be central
in the global metabolic response to sepsis in man [10]. However, IL-6 by itself
does not appear to induce any significant hemodynamic alteration [11]. We pre-
viously assayed a large battery of cytokines over time, including TNF-a, IL-l,
IL-2, IL-6, and interferon-y from the blood of humans with sepsis and found
that only TNF-a and IL-6 are consistently elevated at any time in those subjects.
Furthermore, if elevate TNF-a and IL-6 levels are sustained, they are highly
predictive of the subsequent development of multiple system organ failure and
death [12]. Subsequently, we demonstrated that variations in systemic IL-6 lev-
els parallel variations in PMN up-regulation of the potent cell adhesion mole-
cule CD-llb [l3] in critically ill humans. These data suggest that sustained
serum cytokine elevations induce activation of circulating immune effector cells
inducing end-organ dysfunction and death in man.
Exactly how cytokines release induces subsequent cardiovascular changes
and organ injury is not only a complex process but poorly understood. It is
known, however that ecosinoid metabolism is intimately related to this process
through the elaboration of both prostanoid (PG), leukotriene (LT) and platelet
activating factor (PAF) metabolism. Ecosinoids, such as LTB4, LTD4, throm-
boxane A2, PGF2a and PGE 1 are all released into tissue compartments and their
metabolic breakdown products are measurable both in these compartments and
in the blood of both septic patients and experimental sepsis animals. These
ecosinoid metabolites induce PMN adhesion and migration, increased capillary
permeability.
Potentially, inhibition of ecosinoid metabolism may blunt the initial systemic
inflammatory response minimizing subsequent organ injury and secondary cy-
tokine up-regulation. However, the pattern of cytokine response to acute endo-
toxin infusion in the dog is unclear and the effect of pretreatment with either a
cyclooxygenase or lipooxygenase inhibitor has not been described.
Northover and Subramanian [14] first demonstrated that non-steroidal anti-
inflammatory drugs ameliorate the cardiovascular effects of endotoxin in the ca-
nine model, presumably by inhibiting cyclooxygenase. We subsequently demon-
strated that pretreatment with ibuprofen, a competitive cyclooxygenase inhibitor
blocked the hemodynamic but not metabolic consequences of acute endotox-
emia in an acute canine model [21]. Ibuprofen pretreatment may beneficially af-
fect long-term outcome after endotoxin exposure either by preventing the initial
endothelial injury induced by the endotoxin-stimulated release of cytokine [15]
or by altering the subsequent response of formed cellular elements [16] as
prostaglandins inhibit human monocyte production of cytokines [17]. According
to the above hypothesized mechanism, we would predict that although the initial
serum cytokine levels in response to endotoxin infusion would be unaffected or
Modulating Prostaglandin Metabolism in Sepsis 83
possibly increased by ibuprofen pretreatment, subsequent cytokine levels would

decrease rapidly toward baseline values and be associated with normalization of
hematological and metabolic markers of system stress. Because TNF-a has a
serum half-life of less than 20 minutes [6], serum TNF-a levels measured at a
later time should be reduced toward pre-insult levels in animals pretreated with
ibuprofen. Furthermore, associated markers of a systemic inflammatory re-
sponse, such as circulating polymorphonuclear leukocytes and thrombocytes,
oxygen supply-demand ratios, and arterial blood lactate levels, should parallel
serum cytokine kinetics.
Although cyclooxygenase inhibition with ibuprofen has been extensively
studied in models of endotoxic shock, leukotriene D4 synthetase inhibition with
diethylcarbamazine (DEC) has not. Recent studies in murine models of endo-
toxic shock suggest that lipooxygenase inhibition can modify the inflammatory
response to endotoxin [18]. Furthermore, this modulation appears to be depen-
dent on baseline hepatocytic function [19]. Because IL-6 appears to function
primarily as a metabolic cytokine, and because the liver is one of its primary tar-
gets, we further wished to define the effect of DEC pretreatment on TNF-a and
IL-6 kinetics in a dog model of endotoxic shock.
Potentially, prior cyclooxygenase or 5-lipooxygenase inhibition should not
alter the initial cytokine or WBC response to a septic challenge, but would limit
the duration of any systemic cytokine elevation in response to that challenge by
limiting feed back activation of the initial pathways of inflammation. To exam-
ine these interactions, we pretreated dogs with either ibuprofen, a cyclooxyge-
nase inhibitor, or DEC, a 5-lipooxygenase inhibitor prior to inducing endotoxic
shock with a bolus infusion of E. coli endotoxin.
In control animals, not surprisingly, endotoxin infusion resulted in an imme-
diate decrease in blood pressure, cardiac output, and white blood cell count
(WBC) and platelet levels and an increase in heart rate (HR). Serum TNF-a and
IL-6 levels increased following both the minor surgical manipulation and endo-
toxin infusion. The levels of both cytokines decreased progressively over the
next three hours. The response to endotoxin infusion in the ibuprofen group was
different from that in the control group. Interestingly, blood pressure and serum
lactate levels unchanged, while cardiac output, WBC and platelet levels all de-
creased to levels similar to control animals. Furthermore, the increase in serum
TNF-a and IL-6 levels after endotoxin infusion were similar to that in control
dogs after endotoxin infusion, although the IL-6 levels started at a higher initial
value in the ibuprofen group. Presumably, ibuprofen increased IL-6 levels inde-
pendent of endotoxin. Unlike ibuprofen pretreatment, DEC infusion induced a
transient decrease in blood pressure and an increase in both cardiac output and
IL-6 levels, consistent when a selective drug-induced vasodilatation plus a non-
specific activation of IL-6 release. However, DEC pretreatment had no other
measurable effect on hemodynamic profile data before endotoxin infusion. In-
terestingly, serum IL-6 levels were the highest following endotoxin challenge in
this group when compared with the two others. Interestingly, when we fluid
84 M.R. Pinsky
challenged these dogs and calculated a V0 2ID0 2 relation, we saw that prior to
endotoxin challenge V0 2 remained constant, whereas following endotoxin chal-
lenge, the V0 2ID0 2 relation was linear such that they co-varied. Blocking the
arachidonic acid pathways did not alter this change. These data suggest that
arachidonic acid metabolites playa more complex role in the regulation and ex-
pression of acute endotoxemia than just as amplifiers of the inflammatory signal.
Infusion of endotoxin results in a rapid decrease in arterial pressure and car-
diac output, and a progressive increase in blood lactate levels [1]. Furthermore,
Pinsky and Matuschak [22] demonstrated that the immediate decrease in arterial
pressure and cardiac output was due to a reduction in venous return, presumably
due to venous pooling, whereas in the fluid-resuscitated dog, the changes in
steady-state arterial pressure and cardiac output could be explained by a de-
crease in arterial closing pressure with a minimal change in incremental resis-
tance. Pinsky [21] subsequently demonstrated that ibuprofen pretreatment abol-
ished the observed decreases in both arterial pressure and blood flow occurring
after a bolus infusion of endotoxin but did not prevent the endotoxin-induced in-
crease in V0 2 . Changes in D02 do not normally alter V0 2 in a resting animal
whose D0 2 exceeds 10 to 12 mllkg-J/min- J of oxygen [20]. Our data on the
V0 2ID0 2 relation before infusion of endotoxin support these findings. Howev-
er, post-endotoxin levels of blood lactate were increased, while V0 2 and D0 2
co-varied. These data agree with those from our previous study on patients in
septic shock, in which changes in D0 2 proportionally altered V0 2 only when
blood lactate levels were high [23]. Interestingly, post-endotoxin levels of V0 2
never exceeded pre-endotoxin levels, even at maximal intravascular volume lev-
els during the fluid challenge runs, despite similar D02 levels after volume infu-
sion. These data suggest that the covariance of V0 2 and D0 2 in our model may
reflect impaired oxygen extraction in the tissues. Previous research has shown
that endotoxin infusion induces a transient reduction in WBC and platelet levels
because of the margination of these formed blood elements in the periphery
[24]. Our data agree with those of others and serve as baseline data to compare
the effects of ibuprofen and DEC.
Products of both cyclooxygenase and lipooxygenase metabolism appear to
be central to the hemodynamic response to endotoxin because a) their levels are
increased in the blood and tissues after endotoxemia [4, 9, 19, 25], b) they can
cause many of the observed hemodynamic, cellular, and metabolic effects seen
during endotoxemia, if given alone [9, 26]; c) in disease states where clearance
of these inflammatory mediators is impaired, survival from septicemia is dra-
matically reduced [18, 25], and d) specific inhibitors of both metabolic path-
ways improve survival in animals given endotoxin [24, 27]. As previously re-
ported [28, 30], ibuprofen increased pulmonary vascular resistance. The cause
of this increase may be the inhibition of synthesis of vasodilatory escosinoids,
such as PgE J and PgI2. Numerous investigators have studied the effects of cy-
clooxygenase and 5-lipooxygenase inhibition and/or LTD4 antagonists on endo-
toxin-induced changes in pulmonary and systemic hemodynamics [4]. These
studies demonstrated that neither cyclooxygenase inhibitors nor LTD4-antago-

nists prevent endotoxin-induced changes in WBC count despite the observed
marked protective effects of these therapies on pulmonary and systemic hemo-
dynamics. Interestingly, a calcium channel blocker, verapamil, is one of the few
agents shown to prevent both endotoxin-induced changes in hemodynamics and
WBC changes [29]. Although these studies were done in a different animal
species (sheep), both the time course and magnitude of the hemodynamic re-
sponses were similar.
Importantly, our preliminary study supports the previous findings of others
[30], who reported that pretreatment with ibuprofen can block many of the ob-
served acute hemodynamic effects of bolus endotoxin infusion in the dog. How-
ever, ibuprofen pretreatment does not appear to alter either the blood component
or cytokine response to endotoxin infusion over an initial 3 hour interval. Fur-
thermore, others have shown that ibuprofen pretreatment in humans may aug-
ment the TNF-a and JL-6 response to low-dose endotoxin infusion [5, 31]. Ac-
cordingly, ibuprofen appears to mask the effects rather than inhibit the activa-
tion of proximal inflammatory mediators and their effects on formed blood ele-
ments. Thus, the observed beneficial effects of ibuprofen pretreatment in the
management of animals with endotoxic shock appears to be due to actions that
are indirectly related to the short-term release of inflammatory mediators into
the blood. Furthermore, in patients with documented septic shock [12], TNF-a
levels decrease considerably within the first 4 hours without treatment with cy-
clooxygenase inhibitors.
As previously described, DEC induced a transient hypotensive response
characterized by tachycardia, decreased arterial pressure, and increased cardiac
output, suggesting that selective loss of arterial vasomotor tone induces this ef-
fect [19]. Furthermore, we found that serum IL-6 levels were markedly in-
creased after administration of DEC. DEC did not abolish the hemodynamic re-
sponse to endotoxin, but rather exaggerated it by a sustained tachycardia. DEC
may reduce arterial resistance, thereby increasing peripheral blood flow. In sup-
port of this hypothesis, blood lactate levels, although increased after exposure to
endotoxin, returned to pre-endotoxin levels by 3 hours. DEC does not appear to
affect the TNF-a response to endotoxin. Serum IL-6 levels tended to be higher
after endotoxin, but these differences were insignificant.
Interestingly, we saw significant increases in serum TNF-a and especially
IL-6 levels in those dogs pretreated with DEC. To our knowledge, no one has
ever demonstrated an enhanced inflammatory response to endotoxin with
lipooxygenase inhibition. Furthermore, DEC has been shown to reduce, not
augment, the inflammatory response to endotoxin in the rat with acute liver fail-
ure [19]. There are three explanations for the DEC findings. First, products of
lipooxygenase metabolism may function to minimize the host response to endo-
toxin. However, this explanation seems unlikely because normal lipooxygenase
pathway end-products are inflammatory in nature, whereas cyclooxygenase
pathway end-products have both inflammatory and anti-inflammatory properties
86 M.R. Pinsky
[9]. Second, DEC may incompletely inhibit all lipooxygenase actIvity [32].
DEC is a competitive inhibitor of lipooxygenase. Its pharmacological actions
may allow partiallipooxygenase activity to continue or shunting of arachidonic
acid metabolism toward increased production of platelet activating factor to oc-
cur. Although interesting to speculate upon, this explanation can be neither re-
futed nor proved. Third, DEC may also induce tissue injury, such that its effects
become synergistic with those of endotoxin. We usually observed some degree
of hypotension after DEC infusion in all animals. This hypotensive response
could be minimized by decreasing the DEC infusion rate.
No animal model accurately represents human sepsis. However, numerous
investigators have attempted to explore the pathophysiologic processes opera-
tive in sepsis by examining the effects of endotoxin infusion on the canine car-
diovasculature [1, 22, 24, 27, 28]. Although the cardiovascular response to endo-
toxin appears to be both time- [1, 22] and dose-dependent [19], most studies
have shown that endotoxin produces an immediate decrease in arterial pressure
and cardiac output that is associated with a progressive increase in serum lactate
levels, with subsequent cardiovascular effects being dependent on the method of
administration and amount of endotoxin infused [1, 19]. This hypotensive effect
becomes more pronounced as the amount and rate of endotoxin infusion in-
creases. Furthermore, the fall in cardiac output seen during endotoxemia is a
function of circulating blood volume [22]. In our study, we performed an ex-
change transfusion to acquire autologous iso-hematocrit blood for the fluid chal-
lenge runs and were surprised to find that exchange transfusion increased serum
levels of both TNF-a and IL-6 and that the rise in TNF-a levels of some ani-
mals were similar to those seen after endotoxin infusion. The cause of this effect
is unclear. Since we used hydroxyethylstarch infusion as the oncotic balance of
the blood removal, we may have altered blood monocyte function. However, in
a study in rats, Schmand et al. [33] demonstrated the hydroxyethlystarch aug-
mented IL-6 expression by macrophages. It is unclear if DEC and hydroxyethyl-
starch co-stimulated release of IL-6 in that animal sub-group but it can partially
explain the increases in IL-6 seen in the blood following exchange transfusion
in all animals. Furthermore, the immunocompetent cells in the reserved blood
may also have been producing TNF-a and IL-6, which would then increase
blood cytokine levels further during re-infusion. Subsequent studies in humans
have failed to document measurable increases in TNF-a, IL-6 or IL-8 levels in
shed blood over similar time interval [35]. Furthermore, it is doubtful that the
stored blood was producing significant amounts of cytokines because serum
TNF-a and IL-6 levels decreased during the fluid challenge run before endotox-
in infusion; if mediators were also being infused with the blood, these levels
should have increased.
Clinical studies, which have attempted to alter prostaglandin metabolism in
sepsis, are few. Unfortunately, all have proven negative in their ability to im-
prove outcome [36]. The initial studies of PgE2 infusions in acute respiratory
distress syndrome and the recent study of ibuprofen treatment for sepsis clearly
demonstrate that the non-specific use of these agents in sepsis is not recom-
mended. Clearly, if these agents which either supplant or inhibit arachidonic
acid metabolism are to be used clinically, they will need to be given within the
framework of continual monitoring of a specific end effect, which at present has
not been defined.
References
1. Weil MH, MacLean LD, Vischer MB et al (1959) Studies on the circulatory changes in the dog
produced by gram-negative microorganisms. J Clin Invest 35: 1191-1198
2. Kunkel SL, Spengler M, May MA et al (1988) Prostaglandin E z regulates macrophage-derived
tumor necrosis factor gene expression. J BioI Chern 263:5380-5384
3. Okusawa S, Gelfand JA, Ikejima T et al (1988) Interleukin-l induces a shock-like state in rab-
bits. Synergism with tumor necrosis factor and the effects of cyclooxygenase inhibition. J Clin
Invest 81: 1162-1172
4. Ertel W, Morrison MH, Ayala A, Chaudry IH (1991) Eicosanoids regulate tumor necrosis fac-
tor synthesis after hemorrhage in vitro and in vivo. J Trauma 31:609-615
5. Martich GD, Danner RL, Ceska M, Suffredini AF (1991) Detection of interleukin 8 and tumor
necrosis factor in normal humans after intravenous endotoxin: the effect of anti-inflammatory
agents. JExpMed 173:1021-1024
6. Beutler B, Krochin N, Milsark IW et al (1986) Control of cachectin (tumor necrosis factor)
synthesis: mechanisms of endotoxin resistance. Science 232:977-980
7. Dinarello CA, Cannon JG, Wolff SM et al (1986) Tumor necrosis factor (cachectin) is an
endogenous pyrogen and induces production of interleukin 1. J Exp Med 163: 1433-1450
8. Tracey KJ, Beutler B, Lowry SF et al (1986) Shock and tissue injury induced by recombinant
human cachectin. Science 234:470-474
9. Fletcher JR (1982) The role of prostaglandins in sepsis. Scand J Infect Dis SuppI31:55-60
10. Calandra T, Gerain J, Heumann D et al (1991) High circulating levels of interleukin-6 in
patients with septic shock: evolution during sepsis, prognostic value, and interplay with other
cytokines. Am J Med 91 :23-29
11. Preiser JC, Schmartz D, Van der Linden P et al (1991) Interleukin-6 administration has no
acute hemodynamic or hematologic effect in the dog. Cytokine 3: 1-4
12. Pinsky MR, Vincent JL, Deviere J et al (1993) Serum cytokine levels in human septic shock:
relation to multiple systems organ failure and mortality. Chest 103:565-575
13. Rosenbloom A, Pinsky MR, Bryant JL et al (1995) Leukocyte activation in the peripheral
blood of patients with cirrhosis of the liver and SIRS: Correlation with serum Interleukin-6
levels and organ dysfunction. JAMA 274:58-65
14. Northover BJ, Subramanian G (J 982) Analgesic-antipyruvate mugs as antagonists of endotox-
ic shock in dogs. J PathoI83:463-471
15. Leeper-Woodford SK, Carey PD, Byrne K et al (1991) 3rd: Ibuprofen attenuates plasma tumor
necrosis factor activity during sepsis-induced acute lung injury. J Appl Physiol 71 :915-923
16. Kettelhut IC, Fiers W, Goldberg AL (1987) The toxic effects of tumor necrosis factor in vivo
and their prevention by cyclooxygenase inhibitors. Proc Natl Acad Sci USA 84:4273-4277
17. Knudsen PJ, Dinarello CA, Strom TB (1986) Prostaglandins posttranscriptionally inhibit
monocyte expression of interleukin 1 activity by increasing intracellular cyclic adenoside
monophosphate. J Immunol137:3189-3194
18. Ahmed T, Weichman B, Wasserman MA et al (1988) Prevention and reversal of endotoxin-
induced pulmonary hypertension by a leukotriene antagonist. Eur Respir J 1: 142-152
19. Matuschak GM, Pinsky MR, Klein EC et al (1990) Effects of D-galactosamine-induced acute
liver injury on mortality and pulmonary responses to Escherichia coli lipopolysaccharide.
Modulation by arachidonic acid metabolites. Am Rev Respir Dis 141: 1296-1306
88 M.R. Pinsky
20. Van Snick J, Cayphas S, Vink A et al (1986) Purification and NHz-terminal amino acid
sequence of a T-cell-derived lymphokine with growth factor activity for B-ceU hybridomas.
Proc Nat! Acad Sci USA 83:9679-9683
21. Pinsky MR (1992) The cardiovascular response in canine endotoxic shock: effect of ibuprofen
pretreatment. Circ Shock 37:323-332
22. Pinsky MR, Matuschak GM (1986) Cardiovascular determinants of the hemodynamic
response to acute endotoxemia in the dog. J Crit Care 1:18-31
23. Vincent JL, Roman A, De Backer D, Kahn RJ (1990) Oxygen uptake/supply dependency.
Effects of short-term dobutamine infusion. Am Rev Respir Dis 142:2-7
24. Balk RA, Jacobs RF, Tryka AF et al (1988) Low dose ibuprofen reverses the hemodynamic
alterations of canine endotoxin shock. Crit Care Med 16: 1128-1131
25. Galanos C, Freudenberg MA, Reutter W (1979) Galactosamine-induced sensitization to the
lethal effects of endotoxin. Proc Nat! Acad Sci USA 76:5939-5943
26. Ogletree ML (1987) Overview of physiological and pathophysiological effects of thrombox-
aneA2 . Fed Proc 46:133-138
27. Jacobs ER, Soulsby ME, Bone RC et al (1982) Ibuprofen in canine endotoxin shock. J Clin
Invest 70:536-541
28. Kadowitz PJ, Chapnick BM, Joiner PD, Hyman AL (1975) Influence of inhibitors of
prostaglandin synthesis on the canine pulmonary vascular bed. Am J Physiol 229:941-946
29. Ahmed T, D'Brot J, Weasserman MA et al (1988) Effect of Verapamil on pulmonary and
eicosanoid responses to endotoxin in awake sheep. J Appl Physiol 64: 1700-1708
30. Balk RA, Jacobs RF, Tryka AF et al (1988) Effects of ibuprofen on neutrophil function and
acute lung injury in canine endotoxin shock. Crit Care Med 16: 1121-1127
31. Spinas GA, Bloesch D, Keller U et al (1991) Pretreatment with ibuprofen augments circulat-
ing tumor necrosis factor-alpha, interleukin-6, and elastase during acute endotoxinemia. J
Infect Dis 163:89-95
32. Mathews WR, Murphy RC (1982) Inhibition of leukotriene biosynthesis in mastocytoma cells
by diethylcarbamazine. Biochem PharmacoI31:2129-2132
33. Schmand JF, Ayala A, Morrison MH, Chaudry IH (1995) Effects of hydroxy ethyl starch after
trauma-hemorrhagic shock: restoration of macrophage integrity and prevention of increased
circulating interleukin-6Ievels. Crit Care Med 23:806-814
34. Kilbourn RO, Gross SS, Jubran A et al (1990) N-methyl-L-arginine inhibits tumor necrosis
factor-induced hypotension: Implications for the involvement of nitric oxide. Proc Nat! Acad
Sci USA 87:3629-3632
35. Rosenbloom A, Pinsky MR, Bryant JL et al (1995) Leukocyte Activation in the Peripheral
Blood of Patients With Cirrhosis of the Liver and SIRS: Correlation With Serum Interleukin-6
Levels and Organ Dysfunction. lAMA 274:58-65
36. Bernard GR, Wheeler AP, Russel lA et al (1997) The effects of ibuprofen on the physiology
and survival of patients with sepsis. The Ibuprofen in Sepsis Study Group. N Engl 1 Med 336:
912-918
I 0°2 /\/°2 IN SEPSIS I
Oxygen Delivery and Consumption Relationships in
Sepsis - The Role of Inotropic and Vasoactive Drugs
0. BOYD
Introduction
Historical background
Oxygen delivery (D0 2) and oxygen consumption (V0 2) relationships are a con-
fusing subject with contradictory research findings. Whole body D02, also
termed oxygen availability, is calculated: D02 = Hb x Sa02 x CO x 0.134,
where Hb = hemoglobin concentration, g/dl; Sa02 = arterial oxygen saturation,
%; CO = cardiac output, lImin. V0 2 can either be calculated indirectly, Hb x
(Sa02 - Sv02) x CO x 0.134, where SV02 = venous oxygen saturation, or be
directly measured as the difference between inspired and expired gases. In nor-
mal individuals V0 2 is closely regulated to provide for the bodies needs and
remains relatively constant over a wide range of D02. If D02 falls however, a
critical point is reached after which V0 2 also falls. A different situation might
occur in critically ill patients, particularly septic patients, where V02 appears to
be related to D02 over a wide range of D02, a phenomenon termed "supply
dependence" [1].
The "supply dependence" of V0 2 on D02 allows possible rational therapy -
D02 is increased by inotropes to increase V0 2, to prevent an oxygen debt and
hypoxic damage. It was suggested that an "oxygen flux test" should be conduct-
ed in all critically ill patients and two early clinical trials also tended to show
improved outcome in septic patients [2, 3].
Problems and questions

The biggest problem is that no studies have convincingly demonstrated that
manipulating D02 and V0 2 in septic patients has a positive effect on outcome
(this is quite a different situation to the peri-operative patient) [4]. This paper
considers the problems inherent in the interpretation of D02 and V02 data and
then looks at the role of vasoactive medication in human sepsis and septic
shock.
92 O. Boyd
The interpretation of D02 and V02 data

Experimental methodology
The methods that have been used to investigate the relationship between D02
and V0 2 may be flawed. Many early studies induced changes in D0 2, e.g. by
addition of PEEP, and subsequently showed changes in V0 2 . However, D0 2
changes were over a narrow range, and changes in V0 2 over wider ranges can-
not be implied. Moreover, showing reduction in both parameters may not allow
the conclusion that both will increase together.
Data analysis
Pooled data is often used for analysis, this may show correlation between the
parameters for the population, whereas there may be no correlations for the
parameters for any individual. A further mathematical consideration is the use
of a shared measurement variable, cardiac output, in the calculation of both D02
and V0 2 [5]. However, in the clinical situation the importance of mathematical
coupling may have been over emphasised as a number of trials which have
manipulated D0 2 and measured V0 2 indirectly have not shown any correlation
between D0 2 and V0 2 [6].
Changes in oxygen demand

Many interventions can effect oxygen demand. The normal physiological
response to an increase in demand is to increase oxygen supply, giving an iden-
tical positive relationship between D02 and V0 2 as in the "supply dependent"
situation. However, the relationship here will be demand led, not supply led. For
example, temperature causes an increased V0 2 [7]. Changes in sedation and
paralysis also change oxygen demand. We demonstrated that variations in D0 2
and V0 2 that occurred during variations in sedation level in critically ill post-
operative patients mimicked "supply dependence" [8]. Moreover, there are wide
swings in metabolic rate in ICU patients, occurring due to changes in arousal,
the resultant variations in oxygen transport can mimic supply dependence [9].
A further complication is the fact that increasing blood flow to some organ
systems will increase their metabolic demand. The heart is an obvious example,
increased blood flow requires increased work and increased myocardial V0 2 . In
the kidney, increased renal blood flow and glomerular filtration requires
increased metabolic work and V0 2 for active solute absorption. In these situa-
tions a "supply dependent" relationship between D02 and V0 2 actually exists,
but is not due to a previously unmet demand for oxygen. On the contrary, an
increase in supply is driving the need for increase in V0 2 .
Oxygen Delivery and Consumption Relationships in Sepsis 93
The thermogenic effect of drugs

It has been known for many years that catecholamines are potent stimuli for
increasing metabolic rate in humans [10], due to increases in cellular metabolic
rate and flow changes. Many of the interventions used to increase D02 in criti-
cally ill patients involve the use of catecholamines, and dobutamine has been
advocated as an agent to be used as part of the "oxygen flux test" in the critical-
ly ill. However, epinephrine, norepinephrine, dopamine, dobutamine, and com-
binations of agents have all been shown to increase V0 2 in normal volunteers. If
both flow and V0 2 are increased by the addition of a catecholamine in normals,
the conclusion that when both occur in the critically ill there is evidence of oxy-
gen debt does not seem logical.
More recently two studies have re-evaluated the prognostic value of the rela-
tionship between D0 2 and V0 2 . Bihari et al. showed that a positive relationship
between D02 and V0 2 indicated a poor prognosis [11]. In contrast, Vallet et al.
showed that if V0 2 is increased by a short-term infusion of dobutamine in septic
patients the prognosis for the patients is much better [12]. They suggest that in
patients who show an increase in VO b cellular metabolism is responding as
normal, but a failure of response indicates a cellular metabolic defect. Although
not all studies show a survival difference between responders and non-respon-
ders [13], Vallet's paper is potentially of major importance as it questions the
whole rational for increasing D0 2 and V0 2 as a therapeutic tool.
The influence of inotropes and vasodilators on D02 and V0 2

Despite the large number of studies in this area, in both patients and animal
models, no consensus of opinion has emerged, and apparently similar studies
often give conflicting results. The reasons for this may be detailed above, but
even when there is some agreement the clinical relevance is not known. There is
also doubt as to the bioavailability of the agents used. Moreover, it is by no
means certain that the variations in V0 2 are of importance in themselves. In
some studies V0 2 has remained unchanged while supposed markers of organ
perfusion, such as gastric intramucosal pH (pHi), have improved.
Dohutamine
Dobutamine is the most widely studied agent, and has been traditionally shown
to increase both D0 2 and V0 2 in patients with sepsis and septic shock [14], par-
ticularly if lactate was high. However, the therapeutic importance of this is in
doubt. In the early trials Edwards et al. [2] showed improved outcome in
patients with septic shock but only compared to historical controls. Tuchschmidt
et al. [3], demonstrated improved survival in the group treated with "goal-direct-
ed" therapy, but this did not reach statistical significance. More recently larger
94 o. Boyd
studies have also failed to show improvement in outcome in critically ill

patients, many of whom were septic, when the investigators tried to attain the
goals for cardiac index and D02 [6, 15]. Furthermore, in Hayes' study [6] there
were no changes in \702 in the group treated to attain the higher D02.
Other recent studies have also shown that the effect of dobutamine may be
variable. In patients with sepsis and ARDS, dobutamine increased D02 and \70 2
in only 8 of 12 patients, and this had no relationship to lactate levels [16]. In
another study patients with severe sepsis showed no increase in \702, whether or
not they had a high lactate, and regardless of whether dobutamine or military
antishock trousers were used to increase D02 [17]. However, in other studies the
relationship between rise in D02 and rise in \702 continues to be found [13].
Other vasoactive agents,

Dopexamine has been recommended to increase splanchnic and renal blood flow
and oxygenation [18], but there is little published work on D0 2 /\702 changes.
Smithies et al. showed a rise in pHi and improvement in liver function accompa-
nying an increase in D02 in 10 septic patients [19]. In another study of 29 septic
patients dopexamine led to an increase in D02 and a 4-8% increase in \70 2 [20].
In this study oxygen extraction ratio fell and the authors suggested that the rise
in \702 might be due to calorigenic effects and increased myocardial \702.
Norepinephrine is widely used for the maintenance of blood pressure in sep-
tic shock, but once again its effects on D02 and \702 have been less well docu-
mented. Schreuder et al. reported that norepinephrine increased \70 2 but not
D02 in patients with septic shock, presumably due to rise in perfusion sec-
ondary to rise in mean arterial blood pressure [21]. However, in a study of 28
patients with septic shock, 15 received norepinephrine, and while there were no
changes in global D02 or \702, but there were significant rises in splanchnic
D02 and \70 2 without change in pHi [22]. In contrast, a study comparing
dopamine and norepinephrine in septic shock patients showed increases of D02
and \702 with both agents but an increase in pHi only with norepinephrine, this
despite a higher cardiac output with dopamine [23]. If dopamine is substituted
for dobutamine or norepinephrine in stable patients with septic shock, there is
no further increase in \70 2 despite increases in D0 2 [24].
Bollaert et al. investigated the use of epinephrine in 13 patients with septic
shock, and showed that both D02 and \70 2 could be elevated [25]. Other vaso-
pressor medication has also been used in septic shock. In a cross-over study of
12 patients with septic shock a nitric oxide synthetase inhibitor was adminis-
tered, although mean arterial pressure rose there was a fall in cardiac output and
D02 [26]. However, in a trial of methylene blue in patients with septic shock,
short term infusion resulted in increase in MAP, without detrimental changes in
cardiac output, D02 or \702 [27], and another study has even shown rise in D02
and \702 [28].
Oxygen Delivery and Consumption Relationships in Sepsis 95
Conclusion
Although in the 1980's it appeared that understanding of the relationship
between D02 and V02 in septic patients might herald the onset of new and suc-
cessful therapeutic options, the early promise has not been fulfilled. No clear
picture of the effect of inotropes or vasoactive medication on D02 and V0 2 in
human sepsis has emerged, and no trials have demonstrated that manipulation of
these parameters has any effect on mortality.
References
1. Cain SM (1984) Supply dependancy of oxygen uptake in ARDS: myth or reality? Am J Med
Sci 288: 119-124
2. Edwards JD, Brown CS, Nightingale P et al (1989) The use of survivors cardiorespiratory val-
ues as therapeutic goals in septic shock. Crit Care Med 17: 1098-1103
3. Tuchschmidt J, Fried J, Astiz M et al (1992) Elevation of cardiac output and oxygen delivery
improves outcome in septic shock. Chest 102:216-220
4. Boyd OF, Bennett ED (1996) Enhancement of perioperative tissue perfusion as a therapeutic
strategy for major surgery. New Horizons 4:453-463
5. Chittock DR, Russell JA (1996) Oxygen delivery and consumption during sepsis. Clinics in
Chest Medicine 17:263-278
6. Hayes MA, Timmins AC, Yau EHS et al (1994) Elevation of systemic oxygen delivery in the
treatment of critically ill patients. N Engl J Med 330: 1717 -1722
7. Pemerstorfer T, Krafft P, Fizgerald R et al (1995) Optimal values for oxygen transport during
hypothermia in sepsis andARDS. Acta Anaesth Scand 107 [SupplJ:223-227
8. Boyd 0, Grounds RM, Bennett ED (1992) The dependency of oxygen consumption on oxygen
delivery in critically ill post-operative patients is mimiced by variations in sedation. Chest 101:
1619-1624
9. Villar J, Slutsky AS, Hew E et al (1990) Oxygen transport and oxygen consumption in critical-
ly ill patients. Chest 98:687-692
10. Cori CF, Buchwald KW (1930) Effect of continuous injection of epinephrine on the carbohy-
drate metabolism, basal metabolism and vascular system of normal man. Am J Physiol 95:
71-78
11. Bihari D, Smithies M, Gimson A et al (1987) The effects of vasodilation with prostacyclin on
oxygen delivery and uptake in critically ill patients. N Engl J Med 317 :397 -403
12. Vallet B, Chopin C, Curtis SE et al (1993) Prognostic value of the dobutamine test in patients
with sepsis syndrome and normal lactate values: a prospective multicenter study. Crit Care
Med 21:1868-1875
13. De Backer D, Berre J, Zhang H et al (1993) Relationship between oxygen uptake and oxygen
delivery in septic patients: effects of prostacyc1ine versus dobutamine. Crit Care Med 21:
1658-1664
14. Gilbert EM, Haupt MT, Mandanas RY et al (1986) The effect of fluid loading, blood transfu-
sion, and catecholamine infusion on oxygen delivery and consumption in patients with sepsis.
Am Rev Respir Dis 134:873-878
15. Gattinoni L, Brazzi L, Pelosi P et al (1995) A trial of goal-oriented hemodynamic therapy in
critically ill patients. N Engl J Med 333: 1025-1032
16. Krachman SL, Lodato RF, Morice R et al (1994) Effect of dobutamine on oxygen transport
and consumption in the adult respiratory distress syndrome. Int Care Med 20: 130-137
17. Mira JP, Fabre JE, Baigorri F et al (1994) Lack of oxygen supply dependency in patients with
severe sepsis. A study of oxygen delivery increased by military antishock trouser and dobuta-
mine. Chest 106:1524-1531
96 o. Boyd
18. Meier-Hellmann A, Reinhart K (1995) Effects of catecholamines on regional perfusion and

oxygenation in critically ill patients. Acta Anaesth Scand 39 [Suppl] 107:239-248
19. Smithies M, Yee TH, Jackson L et al (1994) Protecting the gut and liver in the critically ill:
effects of dopexamine. Crit Care Med 22:789-795
20. Hanneman L, Reinhart K, Meier-HelImannA et al (1996) Dopexamine hydrochloride in septic
shock. Chest 109:756-760
21. Schreuder WO, Schneider AJ, Groeneveld ABJ et al (1989) Effect of dopamine vs. norepi-
nephrine on hemodynamics in septic shock. Emphasis on right ventricular performance. Chest
95:1282-1288
22. Meier-Hellmann A, Specht M, Hannemann L et al (1996) Splanchnic blood flow is greater in
septic shock treated with norepinephrine than in severe sepsis. Int Care Med 22: 1354-1359
23. Marik PE, Mohedin M (1994) The contrasting effects of dopamine and norepinephrine on sys-
temic and splanchnic oxygen utilization in hyperdynamic sepsis. JAMA 272: 1354-1347
24. Hannemann L, Reinhart K, Grenzer 0 et al (1995) Comparison of dopamine to dobutamine
and norepinephrine for oxygen delivery and uptake in septic shock. Crit Care Med 23:
1962-9170
25. Bollaert PE, Bauer P, Audibert G et al (1990) Effects of epinephrine on hemodynamics and
oxygen metabolism in dopamine-resistant septic shock. Chest 98:949-953
26. Petros A, Lamb G, Leone A et al (1994) Effects of a nitric oxide synthesis inhibitor in humans
with septic shock. Cardiovasc Res 28:34-39
27. Preiser JC, Lejeune P, Roman A et al (1995) Methylene blue administration in septic shock: a
clinical trial. Crit Care Med 23:259-264
28. Daemen-Gubbels CR, Groeneveld PH, Groeneveld AB et al (1995) Methylene blue increases
myocardial function in septic shock. Crit Care Med 23:1363-1370
I COAGULATION I
Disturbances and Modulators of Coagulation
and Fibrinolysis in Sepsis
L.G. THIJS
In patients with sepsis abnormalities in hemostatic variables indicating systemic

activation of coagulation and early activation with subsequent inhibition of fibri-
nolysis, are common features [1-5]. The most important clinical manifestation
of these processes is disseminat'ed intravascular coagulation (DIC), which is
characterized by generation and deposition of fibrin in the microvasculature
with widespread microvascular thrombosis in various organ systems. The bipha-
sic effect on the fibrinolytic system may result in impaired fibrin dissolution and
aggravate the formation of microthrombi. There is evidence that DIe con-
tributes to the development of (multiple) organ dysfunction and failure [3, 5]. In
addition, depletion of coagulation proteins and platelets, mainly due to the ex-
tensive and ongoing coagulation activation, may induce severe bleeding compli-
cations [6]. The impressive extreme of the clinical spectrum of manifestations of
DIe is purpura fulminans, which presents with hemorrhagic skin necrosis and
peripheral gangrene and is most often secn in meningococcal sepsis [7].
Sepsis appears to be the most common cause of acute DIe and DIe is a rela-
tively frequent complication of sepsis with major implications for morbidity and
mortality. Septic patients who develop DIe have a higher mortality rate and
have more organ dysfunction than patients who do not have signs of DIe [3].
The reported prevalence of DIe in large clinical trials in septic patients varies
between 7.5 and 49%, depending on the type of patients and the definition used
for DIe [8-13]. Generally the prevalence is higher in the more severe forms of
sepsis, e.g. septic shock, than in the milder forms. In a large prospective study
the incidence of DIe in patients with sepsis was 16%, in severe sepsis 18% and
in septic shock 33% [14].
Coagulation and fibrinolysis

In the classical concept of coagulation thrombin is generated by either the ex-
trinsic (tissue factor-dependent) pathway or the intrinsic (contact activation-de-
pendent) pathway of coagulation. Although both pathways can be activated in
sepsis, recent studies using highly specific and sensitive assays for activation
100 L.G. Tbijs
products of coagulation indicate that activation of coagulation in sepsis is pri-

marily driven by the tissue factor-dependent pathway. In non-human primates
challenged with endotoxin, the blocking of the extrinsic pathway by simultane-
ous infusion of monoclonal antibodies directed against tissue factor or factor
vn almost completely inhibits thrombin generation and fibrinogen-to-fibrin
conversion [15, 16]. Conversely, the blocking of the intrinsic pathway in septic
baboons with infusion of a monoclonal antibody against factor XII did not pre-
vent coagulation activation [17]. Therefore, the concept of coagulation has
changed and the extrinsic and intrinsic pathway have essentially merged into
one series of interrelated reactions that are initiated by expression of tissue fac-
tor [18]. During sepsis tissue factor can be expressed on monocytes and en-
dothelial cells. Monocytes generate after exposure to endotoxin measurable
quantities of mRNA encoding the tissue factor molecule which is then rapidly
synthesized and surface exposed [19]. Monocytes isolated from blood of pa-
tients with meningococcal sepsis show increased tissue factor expression and
highest expression was associated with a high mortality rate [20]. Endotoxin can
also in vitro enhance tissue factor expression on endothelial cells, a process that
is slower than in monocytes and takes a few hours [21, 22].
Cell surface-expressed tissue factor binds and activates factor VII, thereby
forming the factor VIla-tissue factor complex which activates factor X and fac-
tor IX. Factor Xa together with factor Va converts prothrombin to thrombin (and
fragment Fl + 2). Factor IXa together with factor VIlla can activate additional
factor X. Thrombin can catalyze the conversion of factor XI to factor Xla which
activates factor IX. Thrombin cleaves fibrinogen, yielding monomeric fibrin
which then polymerizes to form the fibrin clot [23].
Coagulation is regulated by natural inhibitors of coagulation: the antithrom-
bin III (ATIII)-heparin sulfate system, the protein C (PC)-protein S (PS) system
and tissue factor pathway inhibitor (TFPI). ATIII binds to and inactivates throm-
bin generated during the clotting process and forms the thrombin-antithrombin
(TAT) complex and also inactivates other activated coagUlation factors [23].
TAT complexes are sensitive markers of in vivo thrombin generation [24]. This
complex formation is accelerated by exogenous heparin but also by endogenous
heparin sulfates produced by endothelial cells. Synthesis of these glycosamino-
glycans is inhibited by endotoxin [25].
Protein C is activated by complexes of thrombin with the endothelial cell
surface protein thrombomodulin. Activated PC is an important inhibitor of the
coagulation cofactors Va and VIlla and also enhances fibrinolysis by binding
and inactivating plasminogen activator inhibitor-l (PAl-I) [23]. In vitro, endo-
toxin can suppress thrombomodulin expression on endothelial cells [21]. Also,
granulocyte proteases and oxygen radicals which are generated during sepsis
can in vitro inactivate thrombomodulin [26]. Activated PC can be inhibited by a
heparin-dependent protein C inhibitor and by aI-antitrypsin [27]. Protein S
serves as a cofactor for activated PC and exists in plasma in two forms: as free
protein which is active and as an inactive form complexed with C4b-binding
Disturbances and Modulators of Coagulation and Fibrinolysis in Sepsis 101
protein (C4BP). C4BP may, therefore, serve as an inhibitor of PS activity. PS

functions by enhancing the cell surface anticoagulant activity of activated Pc.
C4BP is an acute phase protein and its level may increase in sepsis and bind PS.
It therefore seems that in sepsis various mechanisms may impair activity of
the PC-PS system and thereby promote coagulation.
TFPI produced by endothelial cells is a potent but slow inhibitor of the factor
VIla-tissue factor complex and can also inhibit factor Xa directly [28,29]. The
importance of these natural inhibitors has been demonstrated in a number of
studies in non-human primates. High dose ATIII [30], activated PC [31] and re-
combinant TFPI [32, 33] reduce the coagulopatbic and cell injury responses to
an infusion of a lethal dose of live E. coli and prevent death. Conversely, block-
ing of PC activation with an anti-PC monoclonal antibody [31] or infusion of
C4BP [34] in this model infused with a sublethal dose of E. coli increases organ
injury and promotes mortality.
Fibrinolysis plays an important role in regulating the formation and the re-
moval of microthrombi. The fibrinolytic system is an endogenous system which
seems to preserve the microcirculation from irreversible damage. Fibrinolysis is
initiated mainly by the release of tissue-type plasminogen activator (t-PA) re-
leased from endothelial cells. The role of another plasminogen activator uroki-
nase plasminogen activator (u-PA) which can be released from various cell
types has not been clearly defined [35]. t-PA converts the zymogen plasminogen
into the active enzyme plasmin which enzymatically degrades fibrin into fibrin
degradation products. Also activation of the contact system which is initiated by
factor XII activation can activate fibrinolysis [36].
Fibrinolysis is attenuated at two levels. First, t-PA (and also u-PA) activity is
inhibited by plasminogen activator inhibitor-l (PAl-I) which binds to and inac-
tivates t-PA. Endothelial cells are capable of releasing PAI-l [37]. On the other
hand activated PC enhances fibrinolysis by binding to and neutralizing PAl-I.
Second, plasmin binds to its endogenous inhibitor az-antiplasmin to form plas-
min-antiplasmin (PAP) complexes, whereby plasmin activity is neutralized. The
presence of PAP complexes in plasma is a direct indication of in vivo plasmin
generation. In vitro, cell cultures of human endothelial cells show no change or
a decrease in t-PA release [38,39], whereas the release of PAI-l is increased
[38, 40]. Endotoxin, therefore, seems to inhibit the fibrinolytic capacity of en-
dothelial cells.
Early dynamics of coagulation and fibrinolysis in experimental

septic models: role of cytokines
Studies in human volunteers and non-human primates challenged with a low
dose endotoxin have clarified the early route and dynamics of activation of co-
agulation and fibrinolysis. In these models a rise in plasma concentration of pro-
102 L.G. Thijs
thrombin fragment Fl + 2 and TAT complexes is observed, becoming evident

from 2 hours after the challenge [15,41-43]. Activation of coagulation is, how-
ever, preceded by a rapid and transient activation of fibrinolysis. Following en-
dotoxin injection levels of t-PA rise rapidly starting at 1 hour postinjection and
paralleled by a rise in PAP complexes, indicating in vivo plasmin generation
[15,41-45]. This is followed by an increase in PAl-l levels starting at about 2
hours postinjection [15, 41-45]. Apparently, fibrinolysis is inhibited as evi-
denced by a fall of PAP levels after the rise in PAl-levels, whereas coagulation
activation is still proceeding. Thus, several hours after endotoxin injection coag-
ulation is ongoing and fibrinolysis is offset, creating a procoagulant state. Such
a dysbalance of coagulation and fibrinolysis has also been observed in non-hu-
man primates following infusion of (sub)lethal doses of live E. coli [46].
During sepsis many endogenous mediator systems are activated [47]. These
activated mediators are largely responsible for the clinical signs and symp-
toms of sepsis. Release of cytokines (TNFa, IL-l, IL-6, IL-8, etc.) from acti-
vated macrophages and endothelial cells is an early event after a bacterial
challenge [48].
There is much evidence that the effects of endotoxin on coagulation and fib-
rinolysis are mediated by cytokines. First, TNF and IL-l are able to enhance in
vitro the expression of tissue factor in cultured human endothelial cells [49, 50]
and monocytes [51]. Also, these cytokines can in vitro downregulate thrombo-
modulin activity in endothelial cells, thereby inhibiting their anticoagulant prop-
erties [52, 53]. TNF and IL-l can also affect the fibrinolytic properties of en-
dothelial cells, the overall effect being an impairment of fibrinolysis, mainly by
stimulating the release of PAI-l [38,54]. Second, a bolus injection of recombi-
nant TNFa in healthy individuals induces a sustained thrombin generation as
evidenced by elevated levels of prothrombin fragment Fl + 2 [55]. This is pre-
ceded by early activation of fibrinolysis mediated by a rise in t-PA, followed by
sustained inhibition due to a rise in PAI-l levels [56]. Similar changes in coagu-
lation and fibrinolysis have been observed in patients with malignancies infused
with TNF for 24 hours [57,58]. The TNF-induced signs of activation of coagu-
lation and fibrinolysis are similar to those induced by endotoxin except that they
occur about 1-2 hours earlier, suggesting that endotoxin acts via TNF release.
Third, administration of endotoxin in combination with pentoxifylline, a com-
pound that can inhibit release of TNF by macrophages and mononuclear cells in
non-human primates inhibited endotoxin-induced activation of coagulation and
strongly attenuated the fibrinolytic response [15, 45]. These observations sug-
gest a central role for TNF as activator of coagulation and fibrinolysis. However,
in the same experimental model an anti-TNF monoclonal antibody did not influ-
ence endotoxin-induced activation of coagulation as reflected by unchanged in-
creases in levels of prothrombin fragment Fl + 2 and TAT complexes [42].
Whereas coagulation was not inhibited, fibrinolysis was completely suppressed
[42,45]. In addition, administration of recombinant IL-la in the experimental
model induces a rise in TAT complex levels and activates fibrinolysis [59].
The pattern is similar to that seen after TNF administration i.e. early fibrinol-
ysis is initiated by a rise in t-PA levels and counteracted by a subsequent rise in
PAI-I levels resulting in a transient elevation of levels of PAP complexes and
this is followed by sustained activation of coagulation [59]. Recombinant IL-I
receptor antagonist (IL-Ira) infused in patients with sepsis [60] and non-human
primates with lethal bacteremia [59] can attenuate the coagulation response as
assessed by TAT complex levels.
Although IL-Ira had an attenuating effect on the rise of t-PA and PAI-I lev-
els in the experimental bacteremic model, the early rise of PAP complex levels
was not influenced [59]. Similarly, IL-Ira in septic patients had no effect on
PAP levels [60]. Also anti-IL-6 monoclonal antibodies can significantly reduce
the increase in levels of prothrombin fragment FI + 2 and TAT complexes in
non-human primates subjected to low dose endotoxin [43] but do not suppress
plasmin generation [43,45]. Th,ese studies indicate that TNF, IL-I and IL-6 are
involved in the activation of coagulation and TNF and IL-I in the activation of
fibrinolysis. There are however, very complex interactions between these vari-
ous cytokines (and other mediators) and their role has not been completely elu-
cidated. It has convincingly been shown that thrombin generation can activate
fibrinolysis [61]. However, in a mild endotoxemia model prevention of thrombin
generation by administration of a tissue factor neutralizing monoclonal antibody
or direct neutralization of thrombin by continuous infusion of recombinant
hirudin had no effect on activation of fibrinolysis [45]. In this model, therefore,
induction of coagulation and fibrinolysis appears to be regulated independently
[42-45]. It has been suggested that this also occurs in patients with sepsis [5].
Abnormalities in coagulation and fibrinolysis in patients with sepsis

In patients with severe sepsis signs of activation of coagUlation and fibrinolysis
as judged by changes in common hemostatic variables occur frequently. During
the process of intravascular coagulation there is consumption of fibrinogen, re-
duction in platelet count and increases in prothrombin and activated partial
thromboplastin times, changes that are most pronounced in DIe [3]. Activation
of fibrinolysis results in various fibrin degradation products. The early dynamics
of coagulation and fibrinolysis have been studied in patients with severe
chemotherapy-induced neutropenia, who are prone to severe septic complica-
tions, allowing serial measurements of hemostatic variables before and during
evolving sepsis [62, 63]. These studies have shown that at the onset of sepsis
levels of factor VII activity and antigen rapidly decline, followed by a rise in
prothrombin fragment FI + 2 and TAT complex levels and associated with a
rapid fall in ATIII activity [62]. Also, tPA antigen levels and PAI-I activity and
levels of PAI-l antigen rapidly rise, associated with a decrease in ~-antiplas
min activity and a rise in total fibrin(ogen) degradation products [63]. These
changes were particularly seen in patients who progressed to septic shock. Al-
104 L.G. Thijs
though this study concerned only neutropenic patients, which may limit general-
ization of the findings, it shows that the activation of coagulation and fibrinoly-
sis are an early phenomenon in sepsis. Levels of factor Vll are usually lowered
in patients with severe sepsis [1, 3, 7, 62, 64] and levels of TAT complexes are
elevated in the majority of these patients, even in the absence of clinical signs of
DIC [4,5,27,65-68]. Levels of most natural coagulation inhibitors are marked-
ly lowered most likely related to consumption with an increased turnover during
the coagUlation process. ATIII levels are generally decreased during severe sep-
sis [1-4, 7, 27, 35, 62, 67]. The lowest levels are observed in septic shock [1, 62,
67] and they are sensitive markers of an unfavourable outcome [3, 62].
Serial measurements show a slow spontaneous recovery toward normal val-
ues in survivors, whereas levels usually remain low in non-survivors [1, 3, 4].
Also protein C concentrations are decreased in patients with severe sepsis [1-4,
7,27,35,64,67], a phenomenon which is most pronounced in septic shock even
in the absence of clinically overt DIC [1, 67]. In nonsurvivors usually a severe
and prolonged decrease in PC activity is observed, whereas in survivors serial
measurements show a gradual return towards normal levels [1, 3, 4]. Although
consumption of PC during the coagUlation process is a major mechanism, also
downregulation of thrombomodulin in endothelial cells may interfere with PC
activation. In septic patients there is evidence that PC activity is also regulated
by binding to PC inhibitor and <Xl-antitrypsin, thereby forming complexes by
which PC might loose its antithrombotic properties. Levels of PC inhibitors are
decreased and these complexes appear in the circulation during sepsis [27].
Plasma concentrations of both total and free protein S may also be lowered,
but usually less pronounced than the decrease of PC and ATIII [3, 7, 64]. On the
other hand, levels of TFPI generally increase in sepsis [69, 70], and its kinetics
apparently differ markedly from the kinetics of AT III, PC and PS in sepsis.
Loss of activity of most of the inhibitors during the coagulation process may by
itself contribute to a sustained procoagulant state.
The contact system consisting of the zymogens factor Xll, factor XI,
prekallikrein and the substrate procofactor high-molecular-weight kininogen,
can also be activated during sepsis. This is demonstrated by the finding of low
plasma levels of prekallikrein and factor XI and Xll [1, 2, 68, 71] and elevated
concentrations of complexes between kallikrein and its inhibitors Cl-inhibitor
and <X2-macroglobulinand between factor Xlla or Xla and Cl-inhibitor [67, 71,
72]. Activation of this system, however, does not seem to be involved in coagu-
lation activation, but rather contributes to the development of hypotension in
sepsis by the release of the potent vasodilator bradykinin [17]. Also, activation
of the contact system may activate the fibrinolytic system [36].
Activation of the fibrinolytic system in sepsis can be demonstrated by the
presence of circulating PAP complexes which can be found in many patients [5,
65,66]. Although levels of t-PA antigen are usually elevated in patients with se-
vere sepsis [2,4, 35, 63], t-PA-activity can often not be detected or is only mild-
ly elevated [2, 35], indicating inhibition of t-PA activity, most likely by PAl-I.
Indeed, plasma levels of PAI-l antigen [1,4,35, 73-76] and PAl-l activity [2,
35, 63, 74] are strongly elevated in patients with the severe forms of sepsis. Al-
though in uncomplicated sepsis PAI-l levels may be in the normal range [1], in
septic shock they are usually markedly elevated. In most studies high PAI-l lev-
els are highly predictive for an unfavourable outcome [1, 63, 73-75]. PAl-con-
centrations may be highest on admission to decline in the following days [75,
76] even in nonsurvivors [75], or remain elevated or even increase in non sur-
vivors [1,4,74].
Levels of arantiplasmin, another regulator of plasmin activity and therefore
fibrinolysis, may be in the normal range [1, 2, 4] or are decreased [1, 67, 75-77].
Low concentrations are particularly observed in patients with septic shock [1,
63, 76] indicating a stronger activation of fibrinolysis in the more severe forms
of sepsis. In conclusion, coagulation and fibrinolysis are usually activated in pa-
tients with sepsis and this is most pronounced in septic shock. However, fibri-
nolysis is strongly inhibited creating a dysbalance between coagulation and fib-
rinolysis resulting in a procoagulant state in sepsis which may contribute to the
development of organ failure and mortality.
Dysbalance between coagulation and fibrinolysis

As discussed before, studies in human endothelial cells have shown that endo-
toxin can, directly or indirectly through the actions of TNF and IL-l, stimulate
the procoagulant properties of vascular endothelium in vitro and reduce its anti-
coagulant properties: enhancement of tissue factor expression, downregulation
of proteins C and S. In addition, endotoxin, TNF and IL-l suppress the fibri-
nolytic capacity of vascular endothelial cells mainly by stimulation of PAI-l
production and release.
In human volunteers and non-human primates challenged with either a low
dose endotoxin or TNF early activation of fibrinolysis is observed before signif-
icant thrombin generation can be detected, and fibrinolysis is already offset by
the release of PAl-1 when thrombin generation becomes maximal. Thus, there is
a remarkable dysbalance between activation of coagulation and fibrinolysis cre-
ating a procoagulant state several hours after the challenge, which is also ob-
served in non-human primates infused with (sub)lethal dose of live E. coli mi-
croorganisms. In septic patients coagulation and fibrinolysis are activated with
subsequent strong inhibition of fibrinolysis which may explain why TAT levels
are usually more elevated than PAP levels [5, 78]. The (dys)balance between co-
agulation and fibrinolysis can be assessed by the ratio of levels of TAT and PAP
complexes. In many patients with severe sepsis the TATIPAP ratio is increased,
reflecting a procoagulant state [5, 65, 66]. Septic patients with organ dysfunc-
tion have higher TATIPAP ratios than those who do not develop organ dysfunc-
tion [5]. Moreover, the rise of the TATIPAP ratio may precede the development
of organ dysfunction. Also, the TATIPAP ratio is higher in nonsurvivors than in
106 L.G. Thijs
survivors of a septic insult [5, 65]. These observations suggest that this dysbal-
ance contributes to the development of organ dysfunction and to mortality. A
positive correlation between the TATIPAP ratio and PAI-l levels also suggests
that impaired fibrinolysis is the most important mechanism for this dysbalance
[5]. Also at the local levels in the lungs, enhanced coagulation and suppressed
fibrinolysis, as assessed in broncho-alveolar lavage fluid has been found in pa-
tients with ARDS and this dysbalance may be an important factor in the patho-
genesis of ARDS which is a common complication of sepsis [79, 80]. In conclu-
sion, in sepsis a dynamic process of coagulation and fibrinolysis is ongoing with
a dysbalance between coagulation and (impaired) fibrinolysis. These abnormali-
ties have major pathophysiological and therefore clinical implications as the fib-
rinolytic capacity to counteract widespread vital organ microvascular thrombo-
sis becomes insufficient which seems to contribute to· the development of (mul-
tiple) organ failure and mortality.
Therapeutic implications
The current treatment of DIe in sepsis includes vigourous treatment of the un-
derlying infection and supplementation of clotting factors (fresh frozen plasma)
and platelets when considered necessary. As there are no controlled clinical tri-
als which have demonstrated superiority of any therapeutic regimen, additional
therapy in case of clinically important thromboembolic or bleeding complica-
tions is difficult and usually guided by clinical judgement. Heparin has been
widely used to block coagulation in conjunction with supplementation of clot-
ting proteins, but its use is still controversial, although most authors advise to
administer heparin in low-dose continuous infusion [18]. In an experimental
sepsis model heparin could ameliorate organ damage [81] but there is no evi-
dence that this also occurs in human sepsis. Experimental studies have not
shown that heparin can alter the shock response or lethal effects of E. coli infu-
sion although it is effective in blocking the DIe response [82]. Since the in-
hibitors of coagulation are consumed during the process of coagulation, substi-
tution of these inhibitors may be a therapeutic option. Numerous anecdotal re-
ports have shown promising effects of substituting the acquired ATIII deficiency
in sepsis with infusion of ATIII concentrates. High dose of ATIII given before
infusion of a lethal dose of E. coli in baboons could reduce the intensity of both
coagulopathic and cell injury responses and prevent death [30, 83]. Randomized
clinical studies in patients with DIe demonstrated a significant attenuation of
DIe after ATIII treatment [84, 85]. Only one study, however [84], included pa-
tients with trauma or sepsis and no placebo-control group was included. The on-
ly published double-blind placebo-controlled trial of ATIII was performed in pa-
tients with septic shock [86]. This study showed that the duration of DIe was
reduced in the ATIII-treated group and mortality was reduced by 44% although
this did not reach statistical significance. A large-scale clinical trial is necessary
in order to establish whether treatment with ATIIl is effective. ATIII has not on-
ly antithrombotic properties but may also have an effect on the inflammatory
response. In cultured endothelial cells ATIII can promote pro stacycline release
which may inhibit cytokine release and may have an effect on activated leuko-
cytes [87]. Thus, ATIII may have an additional protective effect.
Activated protein C can prevent in the baboon model the coagulopathic, he-
patotoxic and lethal effects of infusion of lethal doses of live E. coli [31]. Pre-
liminary studies of protein C substitution in a small series of patients with DIC,
albeit not related to sepsis, have shown promising results [88].
In experimental septic models infusion of TFPI reduces the coagulation re-
sponse and diminishes cell injury and also significantly reduces the rise in IL-6
levels, suggesting also an effect on the inflammatory response [32, 33]. Also
mortality is largely prevented. The clinical effects of TFPI are now being evalu-
ated in a phase II clinical trial in patients with sepsis.
In addition to substitution of inhibitors of coagulation, interference with cy-
tokine activities may be another therapeutic strategy. Although highly effective
in the animal model, large scale clinical trials in septic patients using several of
such strategies have provided disappointing results concerning the effects on
mortality [10, 11, 13] but effects on coagulation abnormalities have not been re-
ported in detail. In one study using an anti-TNF monoclonal antibody no effect
was observed on coagulation and fibrinolysis variables [89]. New strategies
more specifically aimed at treatment or prevention of DIC in sepsis may include
inhibition of coagulation factor activity e.g. by anti-tissue factor antibodies [15,
90] or anti factor VIla antibodies [16]. It may be expected that the increased
knowledge of the pathophysiology of coagulation and fibrinolysis will ultimate-
ly result in effective therapeutic modalities for septic patients.
References
l. Hesselvik JF, Blombiick M, Brodin Bet al (1989) Coagulation, fibrinolysis, and kallikrein sys-
tems in sepsis: relation to outcome. Crit Care Med 17:724-733
2. Voss R, Matthias FR, Borkowski G et al (1990) Activation and inhibition of fibrinolysis in sep-
tic patients in an internal intensive care unit. Br J HaematoI75:99-105
3. Fourrier F, Chopin C, Goudemand Jet al (1992) Septic shock, multiple organ failure, and dis-
seminated intravascular coagulation. Compared patterns of antithrombin III, protein C, and
protein S deficiencies. Chest 10 1:816-823
4. Lorente JA, Garcia-Frade LJ, Landin L et al (1993) Time course of hemostatic abnormalities
in sepsis and its relation to outcome. Chest 103: 1536-1542
5. Kidokoro A, Iba T, Fukunaga M et al (1996) Alterations in coagulation and fibrinolysis during
sepsis. Shock 5:223-228
6. Levi M, ten Cate H, van der Poll T et al (1993) Pathogenesis of disseminated intravascular
coagulation in sepsis. JAMA 270:975-979
7. Fijnvandraat K, Derkx B, Peters M et al (1995) Coagulation activation and tissue necrosis in
meningococcal septic shock: severely reduced protein C levels predict a high mortality.
Thromb Haemost 73: 15-20
108 L.G. Thijs
8. Ziegler EJ, Fisher CJ, Sprung CL et al (1991) Treatment of gram-negative bacteremia and sep-
tic shock with HA-IA human monoclonal antibody against endotoxin. N Engl J Med 324:
429-436
9. Greenman RL, Schein RMH, Martin MA et al (1991) A controlled clinical trial ofE5 Murine
monoclonal IgM antibody to endotoxin in the treatment of gram-negative sepsis. JAMA 266:
1097-1102
10. Fisher CJ, Dhainaut JFA, Opal SM et al (1994) Recombinant human Interleukin 1 receptor
antagonist in the treatment of patients with sepsis syndrome. JAMA 271: 1836-1843
11. Abraham E, Wunderink R, Silvennan H et al (1995) Efficacy and safety of monoclonal anti-
body to human tumor necrosis factor a in patients with sepsis syndrome. JAMA 273:934-941
12. Bone RC, Baik RA, Fein AM et al (1995) A second large controlled clinical study of E5, a
monoclonal antibody to endotoxin: Results of a prospective, multicenter, randomized, con-
trolled trial. Crit Care Med 23 :994-1 005
13. Cohen J, Carlet J, Intersept study group (1996) INTERSEPT: an international, multicenter,
placebo-controlled trial of monoclonal antibody to human tumor necrosis factor-a in patients
with sepsis. Crit Care Med 24:1431-1440
14. Rangel-Frausto MS, Pittet D, Costigan M et al (1995) The natural history of the systemic
inflammatory response syndrome (SIRS). A prospective study. JAMA 273: 117 -123
15. Levi M, ten Cate H, Bauer KA et al (1994) Inhibition of endotoxin-induced activation of coag-
ulation and fibrinolysis by pentoxifylline or by a monoclonal anti-tissue factor antibody in
chimpanzees. J Clin Invest 93:114-120
16. Biemond BJ, ten Cate H, Levi M et al (1995) Complete inhibition of endotoxin-induced coag-
ulation activation in chimpanzees with a monoclonal Fab fragment against factor VIIIVIIa.
Thromb Haemost 73:223-230
17. Pixley RA, DeLa Cadena R, Page JD et al (1993) The contact system contributes to hypoten-
sion but not to disseminated intravascular coagulation in lethal bacteremia. J Clin Invest 91:
61-68
18. Ten Cate H, Brandjes DPM, Wolters HJ et al (1993) Disseminated intravascular coagulation:
pathophysiology, diagnosis, and treatment. New Horizons 1:312-323
19. Gregory SA, Morrissey JH, Edgington TS (1989) Regulation of tissue factor gene expression
in the monocyte procoagulant response to endotoxin. Mol Cell BioI 9:2752-2755
20. 0sterud B, Flregstad T (1983) Increased tissue thromboplastin activity in monocytes of
patients with meningococcal infection: related to an unfavourable prognosis. Thromb Haemost
49:5-7
21. Moore KL, Andreoli SP, Esmon NL et al (1987) Endotoxin enhances tissue factor and sup-
presses thrombomodulin expression of human vascular endothelium in vitro. J Clin Invest
79:124-130
22. Colucci M, Balconi G, Lorenzet R et al (1983) Cultured human endothelial cells generate tis-
sue factor in response to endotoxin. J Clin Invest 71 : 1893-1896
23. Furie B, Furie BC (1992) Molecular and cellular biology of blood coagulation. N Engl J Med
326:800-806 .
24. Wada H, Minamikawa K, Wakita Yet al (1993) Hemostatic study before onset of disseminated
intravascular coagulatiQn. Am J Hematol43: 190-194
25. Kobayashi M, Shimada K, Ozawa T (1990) Human recombinant interleukin I-beta and tumor
necrosis factor alfa-mediated suppression of heparin-like compounds on cultured porcine aor-
tic endothelial cells. J Cell PhysioI144:383-390
26. Abe H, Okajima K, Okabe H et al (1994) Granulocyte proteases and hydrogen peroxide syner-
gistically inactivate thrombomodulin of endothelial cells in vitro. J Lab Clin Med 123:874-881
27. Alcaraz A, Espana F, Sanchez-Cuenca J et al (1995) Activation of the protein C pathway in
acute sepsis. Thromb Res 79:83-93
28. Rapaport SI (1991) The extrinsic pathway inhibitor: a regulator of tissue factor-dependent
blood coagulation. Thromb Haemost 66:6-15
29. Girard TJ, Warren LA, Novotny WF et al (1989) Functional significance of the Kunitz-type
inhibitory domains of lipoprotein-associated coagulation inhibitor. Nature 338:518-520
30. Taylor Jr FE, Emerson Jr TE, Jordan R et al (1988) Antithrombin-III prevents the lethal effects
of Escherichia coli infusion in baboons. Circ Shock 26:227-235
31. Taylor Jr FE, Chang A, Esmon CT et al (1987) Protein C prevents the coagulopathic and lethal
effects of Escherichia coli infusion in baboons. J Clin Invest 79:918-925
32. Creasey AA, Chang ACK, Feigen L et al (1993) Tissue factor pathway inhibitor reduces mor-
tality from Escherichia coli septic shock. J Clin Invest 91:2850-2860
33. Carr C, Bild GS, Chang ACK et al (1995) Recombinant E. coli-derived tissue factor pathway
inhibitor reduces coagulopathic and lethal effects in the baboon Gram-negative model of sep-
tic shock. Circ Shock 44: 126-137
34. Taylor F, Chang A, Ferrel G et al (1991) C4b-binding protein exacerbates the host response to
Escherichia coli. Blood 78:357-363
35. Philippe J, Offner F, Declerck PJ et al (1991) Fibrinolysis and coagulation in patients with
infectious disease and sepsis. Thromb Haemost 65:291-295
36. Jansen PM, Pixley RA, Brouwer M et al (1996) Inhibition of factor XII in septic baboons
attenuates the activation of complement and fibrinolytic systems and reduces the release of
interleukin-6 and neutrophil elastase. Blood 87:2337-2344
37. Sprengers ED, Kluft C (1987) Plasminogen activator inhibitors. Blood 62:381-387
38. Emeis JJ, Kooistra T (1986) Interleukin-l and lipopolysaccharide induce an inhibitor of
tissue-type plasminogen activator in vivo and in cultured endothelial cells. J Exp Med 163:
1260-1266
39. Hanss M, Collen D (1987) Secretion of tissue-type plasminogen activator and plasminogen
activator inhibitor by cultured human endothelial cells: modulation by thrombin, endotoxin,
and histamine. J Lab Clin Med 109:97-104
40. Colluci M, Paramo JA, Collen D (1985) Generation in plasma of a fast acting inhibitor of pi as-
minogen activator in response to endotoxin stimulation. J Clin Invest 75:818-824
41. Van Deventer SJH, Buller HR, Ten Cate JW et al (1990) Experimental endotoxemia in
humans: analysis of cytokine release and coagulation, fiblinolytic and complement pathways.
Blood 76:2520-2526
42. Van der Poll T, Levi M, van Deventer SJH et al (1994) Differential effects of anti-tumor necro-
sis factor monoclonal antibodies on systemic inflammatory responses in experimental endo-
toxemia in chimpanzees. Blood 83 :446-451
43. Van der Poll T, Levi M, Hack CE et al (1994) Elimination of Interleukin 6 attenuates coagula-
tion activation in experimental endotoxemia in chimpanzees. J Exp Med 179: 1253-1259
44. Suffredini AF, Harpel PC, Parrillo JE (1989) Promotion and subsequent inhibition of plas-
minogen activation after administration of intravenous endotoxin to normal subjects. N Engl J
Med 320: 1165-1172
45. Biemond BJ, Levi M, ten Cate H et al (1995) Plasminogen activator and plasminogen activator
inhibitor I release during experimental endotoxaemia in chimpanzees: effect of interventions
in the cytokines and coagulation cascades. Clin Science 88:587-594
46. De Boer JP, Creasy AA, Chang A et al (1993) Activation patterns of coagulation and fibrinoly-
sis in baboons following infusion with lethal or sublethal dose of Escherichia coli. Circ Shock
39:59-67
47. Hack CE, Thijs LG (1991) The orchestra of mediators in the pathogenesis of septic shock: A
review. In: Vincent JL (ed) Update in Intensive Care and Emergency Medicine. Springer, New
York, pp 232-246
48. Redl H, Schiag G, Bahrami S et al (1991) Plasma neutrophil-activating peptide-V
Interleukin-8 and neutrophil elastase in a primate bacteremia model. J Infect Dis 164:383-388
49. Bevilacqua MP, Pober JS, Majeau GR et al (1984) Interleukin-l (IL-l) induces biosynthesis
and cell surface expression of procoagulant activity in human vascular endothelial cells. J Exp
Med 160:618-623
50. Conway EM, Back R, Rosenberg RD et al (1989) Tumor necrosis factor enhances expression
of tissue factor in RNA in endothelial cells. Thromb Res 53:231-241
110 L.G. Thijs
51. Conkling PR, Greenberg CS, Weinberg JB (1988) Tumor necrosis factor induces tissue
factor-like activity in human leukemia cell line U937 and peripheral blood monocytes. Blood
72:128-133
52. Nawroth PP, Handley DA, Esmon CT et al (1986) Interleukin-l induces endothelial cell proco-
agulant while suppressing cell surface anticoagulant activity. Proc Natl Acad Sci USA 83:
3460-3464
53. Conway EM, Rosenberg RD (1988) Tumor necrosis factor suppresses transcription of throm-
bomodulin gene in endothelial cells. Mol Cell BioI 8:5588-5592
54. Schleef RR, Bevilacqua MP, Sawdey M et al (1988) Cytokine activation of vascular endotheli-
um: effects on tissue-type plasminogen activator and type I plasminogen activator inhibitor. J
BioI Chern 263:5797-5803
55. Van der Poll T, Buller HR, ten Cate H et al (1990) Activation of coagulation after administra-
tion of tumor necrosis factor to normal subjects. N Engl J Med 322: 1622-1627
56. Van der Poll T, Levi M, Buller HR et al (1991) Fibrinolytic response to tumor necrosis factor
in healthy subjects. J Exp Med 174:729-732
57. Bauer KA, ten Cate H, Barzegar Set al (1989) Tumor necrosis factor infusions have a proco-
agulant effect on the hemostatic mechanism of humans. Blood 74: 165-172
58. Van Hinsbergh VWM, Bauer KA, Kooistra T et al (1990) Progress of fibrinolysis during tumor
necrosis factor infusion in humans, concomitant increase in tissue-type plasminogen activator,
plasminogen activator inhibitor type-I, and fibrin(ogen) degradation products. Blood 76:
2284-2289
59. Jansen PM, Boermeester MA, Fischer E et al (1995) Contribution of interleukin-1 to activation
of coagulation and fibrinolysis, neutrophil degranulation, and the release of secretory-type
phospholipase A2 in sepsis: Studies in nonhuman primates after interleukin-1 ex administration
and during lethal bacteremia. Blood 86: 1027 -1034
60. Boermeester MA, van Leeuwen PAM, Coyle SM et al (1995) Interleukin-1 blockade attenu-
ates mediator release and dysregulation of the hemostatic mechanism during human sepsis.
Arch Surg 130:739-748
61. Giles AR, Nesheim ME, Herring SW et al (1990) The fibrinolytic potential of the normal pri-
mate following the generation of thrombin in vivo. Thromb Haemost 63:476-481
62. Mesters RM, Mannuccio PM, Coppola R et al (1996) Factor Vila and antithrombin III activity
during severe sepsis and septic shock in neutropenic patients. Blood 88:881-886
63. Mesters RM, FlOrke N, Ostermann et al (1996) Increase of plasminogen activator inhibitor
levels predicts outcome of leukocytopenic patients with sepsis. Thromb Haemost 75:902-907
64. Leclerc F, Hazelzet J, Jude B et al (1992) Protein C and S deficiency in severe infectious pur-
pura of children: a collaborative study of 40 cases. Intensive Care Med 18:202-205
65. Thijs LG, de Boer JP, de Groot MCM et al (1993) Coagulation disorders in septic shock. Inten-
sive Care Med 19:58-S15
66. Takahashi H, Tatewaki W, Wada K et al (1990) Thrombin versus plasmin generation in dis-
seminated intravascular coagulation associated with various underlying disorders. Am J
Hemato133:90-95
67. Hesselvik JF, MaIm J, Dahlback Bet al (1991) Protein C, Protein Sand C4b-binding protein in
severe infection and septic shock. Thromb Haemost 65: 126-129
68. Wuillemin WA, Fijnvandraat K, Derkx BHF et al (1995) Activation of the intrinsic pathway of
coagulation in children with meningococcal septic shock. Thromb Haemost 74: 1436-1441
69. Brandtzaeg P, Sandset PM, Jo¢ GB et al (1989) The quantitative association of plasma endo-
toxin, antithrombin, protein C, extrinsic pathway inhibitor and fibrinopeptide A in systemic
meningococcal disease. Thromb Res 55:459-470
70. Warr TA, Rao LVM, RapaportSI (1989) Human plasma extrinsic pathway inhibitor activity:
II. Plasma levels in disseminated intrava!;Cular coagulation and hepatocellular disease. Blood
74:994-998
71. Nuijens JH, Huijbregts CC, Eerenberg-Belmer AJ et al (1988) Quantification of plasma factor
XIIa-C I-inhibitor and kallikrein-C I-inhibitor complexes in sepsis. Blood 72: 1841-1848
72. Kaufman N, Page JD, Pixley RA et al (1991) az-macroglobulin-kallikrein complexes detect

contact system activation in hereditary angioedema and human sepsis. Blood 77:2660-2667
73. Pralong G, Calandra T, Glauser MP et al (1989) Plasminogen activator inhibitor 1: a new prog-
nostic marker in septic shock. Thromb Haemost 61 :459-462
74. Gando S, Kameue T, Nanzaki S et al (1996) Disseminated intravascular coagulation is a fre-
quent complication of systemic inflammatory response syndrome. Thromb Haemost 75:
224-228
75. Kruithof EKO, Calandra T, Pralong G et al (1993) Evolution of plasminogen activator
inhibitor type I in patients with septic shock - correlation with cytokine concentrations. Fibri-
nolysis 7:117-121
76. Brandtzaeg P, J00 GB, Brusletto B et al (1990) Plasminogen activator inhibitor 1 and 2,
alpha-2-antiplasmin, plasminogen, and endotoxin levels in systemic meningococcal disease.
Thromb Res 57:271-278
77. Okajima K, Yang W-P, Okabe H et al (1991) Role of leukocytes in the activation of intravascu-
lar coagulation inpatients with septicemia. Am J Hemat 36:265-271
78. Ono S, Mochizuki H, Tamakuma S (1996) A clinical study on the significance ofplatelet-acti-
vating factor in the pathophysiology of septic disseminated intravascular coagulation in
surgery.AmJ Surg 171:409-415
79. Idell S, James KK, Levin EG et al (1989) Local abnormalities in coagulation and fibrinolytic
pathways predispose to alveolar fibrin deposition in the adult respiratory distress syndrome. J
Clin Invest 84:695-705
80. Bertozzi P, Astedt B, Zenzius L et al (1990) Depressed bronchoalveolar urokinase activity in
patients with adult respiratory distress syndrome. N Engl J Med 322:890-897
81. Tanaka T, Tsujinaka T, Kambayashi J-I et al (1990) The effect of heparin on multiple organ
failure and disseminated intravascular coagulation in a sepsis model. Thromb Res 60:321-330
82. Coalson JJ, Benjamin B, Archer LT et al (1978) Prolonged shock in the baboon subjected to
infusion ofE. coli endotoxin. Circ Shock 5:423-437
83. Emerson TE Jr, Fournel MA, Redens TB et al (1989) Efficacy of antithrombin III supplemen-
tation in animal models of fulminant Escherichia coli endotoxemia or bacteremia. Am J Med
87:27S-33S
84. Blauhut B, Kramar H, Vinazzer H et al (1985) Substitution of antithrombin III in shock and
DIC: a randomized study. Thromb Res 39:81-89
85. Maki M, Terao T, lkenoue T et al (1987) Clinical evaluation of antithrombin III concentrate
and BI 6.013 for disseminated intravascular coagulation in obstetrics. Gynecol Obstet Invest
23:230-240
86. Fourrier F, Chopin C, Huart JJ et al (1993) Double-blind, placebo-controlled trial of
antithrombin III concentrates in septic shock with disseminated intravascular coagulation.
Chest 104:882-888
87. Yamauchi T, Umed F, Inoguche T (1991) Antithrombin III stimulates prostacyclin production
by cultured aortic endothelial cells. Biochem Biophys Res Comm 163:1404-1411
88. Okajima K, Imamura H, Koga Set al (1990) Treatment of patients with disseminated intravas-
cular coagulation by protein C. Am J Hemat 33:277-278
89. Salat C, Boekstegers P, Holler E et al (1996) Hemostatic parameters in sepsis patients treated
with anti-TNFa-monoclonal antibodies. Shock 6:233-237
90. Taylor Jr FB, Chang A, RufW et al (1991) Lethal E. coli septic shock is prevented by blocking
tissue factor with monoclonal antibody. Circ Shock 33: 127 -134
EXPERIMENTAL
AND CLINICAL APPROACH
Therapies Directed against TNF-a and IL-I during
Sepsis
K. REINHART, W. KARzAI
Sepsis is a systemic inflammatory response to an infectious stimuli [1]. Sepsis is

diagnosed when clinical and/or laboratory parameters confirm the presence of
an inflammatory response and, identifies an infection as the cause of this re-
sponse. The infection can be located in any organ or tissue of the body and can
be caused by bacteria, fungi, parasites or viruses. Sepsis has been documented
in all age groups and may occur both in immunocompetent and immunosup-
pressed patients. Although our understanding of the pathophysiology of sepsis
has improved considerably and despite availability of modern medical technolo-
gy, the morbidity and mortality associated with this syndrome has not been
much affected in the last decade [2, 3].
Sepsis begins when infections cannot be contained within tissues and infec-
tious agents or toxins invade the blood stream and cause a profound systemic in-
flammatory reaction. This reaction involves cells (neutrophils, macrophages,
monocytes and endothelial cells), plasma factors (complement and coagulation
system), and products of cells (cytokines, eicosanoids, nitric oxide etc.) [4-8].
These various components of the inflammatory response are important in gaug-
ing the inflammatory response to the magnitude of the infectious stimuli and
have complex interrelationships and interactions with both potentiating and in-
hibiting effects [9]. Excessive amounts of these mediators, however, can cause
vascular paralysis, increased vascular permeability, myocardial depression, renal
failure, respiratory failure, encephalopathy, coagulation abnormalities, and,
when untreated or untreatable, can lead to death. Thus host inflammatory medi-
ators playa central role in the pathogenesis of sepsis-associated morbidity and
mortality.
Conventional therapy of sepsis consists of eradication of infections with an-
tibiotics, elimination of septic foci with surgical procedures, and supportive in-
tensive care management of organ failure. Since inflammatory mediators are in-
strumental in sepsis-associated organ dysfunction, much research has focused
on modulating the host inflammatory response as an adjunctive therapeutic op-
tion in sepsis. Most of this research has focused on inhibiting components of the
host inflammatory response with appropriate antibodies or receptor antagonists.
The cytokines, peptides that regulate the amplitude and duration of the host
inflammatory response, are an important group of inflammatory mediators [9].
116 K. Reinhart, W. Karzai
Tumor necrosis factor alpha (TNF-a), interleukin-l (IL-l), interleukin-6 (IL-6),

and interleukin-8 (IL-8) are cytokines often associated with sepsis. These cy-
tokines are released from various cells (monocytes, macrophages, endothelial
cells, etc.) in response to infectious stimuli and bind to specific receptors of oth-
er cells changing their behavior and defining their role in the inflammatory re-
sponse. During infection and inflammation, the host not only produces cy-
tokines with predominantly pro-inflammatory properties, it also produces
counter-inflammatory cytokines like interleukin-IO (lL-IO) and IL-4 and solu-
ble receptors or receptor antagonists of pro-inflammatory cytokines [9-13].
Among other factors released during the inflammatory response to infections
are bioactive phospholipids such as platelet activating factor (PAP) [14].
Tumor necrosis factor alpha

TNF-a, a 17 -kDa polypeptide produced primarily by mononuclear phagocytes,
is implicated in the pathogenesis of sepsis [15]. Administration of endotoxin in
humans leads to an early and substantial increase in TNF-a levels [16] TNF-a
challenge in dogs simulates many abnormalities and systemic manifestations of
human sepsis [17]. Inhibition of TNF-a with appropriate antibodies improves
survival in animal models of bacteremic and endotoxic shock [18, 19]. Lastely,
TNF-a has been detected in the blood of septic patients [20, 21]. Because of
these findings, clinical trials were conducted to test whether inhibiting TNF-a
can also improve survival in human sepsis. The trials were conducted with vari-
ous TNF-a neutralizing agents such as monoclonal antibodies or antibody-frag-
ments directed against TNF-a or with soluble TNF-a-receptors with neutraliz-
ing capabilities.
Trials with murine monoclonal antibodies

In a phase II study, the efficacy of three doses (ranging from 0.1 to 10 mg/kg)
of a murine monoclonal antibody against TNF-a (CB0006) [22] was studied in
80 patients with severe sepsis or septic shock. No overall 28-day survival bene-
fit was found although patients with high TNF-a levels seemed to have some
benefit from the 10 mg/kg antibody treatment. More than 900 patients with
sepsis or septic shock were included in a multicenter North American study [2]
which used three different concentrations (0, 7.5, and 15 mg/kg) of a murine
monoclonal antibody to TNF-a. This trial found no overall reduction in 28-day
mortality among the three groups. A second international trial [23] with the
same antibody used three concentrations (0, 3, 15 mg/kg) and enrolled over 500
patients in septic shock patients. This trial too failed to show any significant ef-
fect of the antibody on 28-day survival in septic shock patients as compared to
controls.
Therapies Directed against TNF-a and IL-l during Sepsis 117
Human or humanized monoclonal antibodies

In a multicenter dose-ranging study, Reinhart et al. [24] used a monoclonal anti-
body fragment directed against TNF-a in patients with severe sepsis or septic
shock. This trial enrolled over 120 patients but also failed to show any beneficial
effects of the antibody fragment on 28-day survival. A post hoc analysis of this
trial suggested that patients with baseline increased IL-6 concentrations (greater
than 1000 pg/ml) may have had some benefit from the drug. This hypothesis
was tested in a prospective randomized study but was stopped because it also
failed to reduce sepsis-associated mortality (personal communication). The
American arm of this study however is still ongoing because an interim analysis
of this trial did not urge to stop this trial due to futility.
Fusionproteins
Three doses of a soluble fusion protein (p75 kd TNF-a receptor and Fc portion
of IgG 1) [3], which binds and neutralizes TNF-a, were used in patients with se-
vere sepsis. In this study, mortality increased with increasing doses (0.15, 0.45,
and 1.5 mg/kg) of the fusion protein. In a most recent study [25], yet another
soluble fusion protein (RO 45-2081; p55 TNF-a receptor and portion of IgG1),
was studied in septic patients. The study used three different doses of the fusion
protein and included several hundred patients. In this study, as in previous stud-
ies, overall survival was not significantly different among the groups studied,
however the subgroup of patients with severe sepsis and early septic shock may
have had benefit.
These negative results question the rationale of therapies directed against
TNF-a.
However, antibodies directed against TNF-a have proven clinical efficacy in
special groups of patients with infectious disease. Antibiotic therapy of patients
with relapsing fever (Borrelia recurrentis infection) is followed by fever, persis-
tent hypotension, and rigors (Jarisch-Herxheimer reaction). This response is as-
sociated with increases in circulating cytokine (TNF-a, IL-6, IL-8) levels. Pre-
treatment with antibodies against TNF-a suppressed the Jarisch-Herxheimer re-
action and reduced circulating cytokine levels in these patients [26].
Interleukin-l
The term IL-l is used for both IL-l a and IL-1~, both produced by mononuclear
cells and sharing similar biological properties [27]. Infusion of IL-l simulates
symptoms of sepsis [27] and endotoxin administration in humans causes eleva-
tion of IL-1 levels [28]. In addition, IL-1 is detected in serum of patients with
sepsis or septic shock [29-31]. IL-1 has a naturally occurring receptor antagonist
(IL-lra) which has been used in laboratory and clinical trials to inhibit IL-l ac-
tivity. In laboratory animals, IL-lra was highly efficacious in increasing survival

during both endotoxemic and bacteremic shock [32, 33]. The foregoing discus-
sion was the rationale behind using recombinantly IL-lra to decrease the mor-
bidity and mortality of human sepsis.
Clinical trials of IL-Ira in sepsis

In an initial clinical trial [34] 99 patients with sepsis received either IL-lra
(Synergen Inc, Boulder, CO, USA) - in three dose ranges - or placebo. The
study found not only that the drug was well tolerated but also demonstrated a
dose-related improvement in 2S-day survival. This improvement was accompa-
nied by decreases in severity of injury score (APACHE II) and decreases in IL-6
concentrations (within the first 24 hours). However, a second trial [35] of IL-lra
which enrolled over SOO septic patients, failed to show efficacy and did not im-
prove 2S-day survival when compared to a control group. Since this trial sug-
gested improved survival in a subgroup of patients with one or more organ fail-
ures, a new trial was started which enrolled patients with characteristics of the
subgroup which seemed to benefit from IL-Ira treatment. However, this trial too
failed to show clinical efficacy [36].
Why clinical trials directed against TNF -(1 and IL-l may have failed
Some of many possible reasons of why these promising immunomodulatory
therapies of sepsis have failed may be as follows:
1. Mediators involved in the pathogenesis of sepsis have both beneficial and
deleterious effects. For example, TNF-a is a cytokine which is important in
regulating host defense mechanisms. For example, TNF-a promotes
transendothelial migration of neutrophils to sites of inflammation and infec-
tion [37] and promotes superoxide production of neutrophils which is im-
portant in killing bacteria [3S]. Rats pretreated with recombinant TNF-a one
week prior to abdominal infection through cecal ligation and puncture all
survived whereas mortality was 50% in rats not pretreated with TNF-a [39].
Furthermore, inhibiting TNF-a activity in Listeria monocytogenes infec-
tions has worsened outcome [40]. These and other studies [41] underline the
importance of TNF-a in acute bacterial infections. Likewise, other "media-
tors of sepsis" have important functions within the specific and non-specific
host defense to infections.
2. One important idea behind directing therapies against these mediators was
that their excessive release during severe infections leads to the severe sys-
temic manifestations of sepsis. Although overwhelming release of pro-in-
flammatory mediators may contribute to the septic response, a massive com-
pensatory anti-inflammatory reaction which occurs concomitantly seems to
be also important in the development of sepsis [9, 11-13].
Therapies Directed against TNF-a and IL-1 during Sepsis 119
3. The patient population included in these trials were very heterogeneous.

Age, immune status, and genetic predisposition may all alter the inflamma-
tory response to infections and lead to different responses to anticytokine
therapies. In one experimental model [42], for example, older animals were
ten times more sensitive to endotoxin administration than younger animals
and released more TNF-a and more nitric oxide in response to endotoxin
than younger animals. Genomic polymorphism within the TNF-a gene lo-
cus may also influence plasma TNF-a levels during sepsis and influence
subsequent survival or non-survival of patients [43]. These studies show the
importance of patient characteristics in the immune response to infections
and suggest that these differences may also lead to different responses dur-
ing immunomodulatory therapies.
4. The criteria used in enrolling patients in immunomodulatory trials may be
important when diagnosing sepsis as a clinical syndrome but may not be
predictive of the patients inimune status. We may need immunologic para-
meters such as HLA-classification, circulating cytokine levels, or new para-
meters like procalcitonin [44] which better identify subgroups of septic pa-
tients who may benefit from pro- or anti-inflammatory therapies.
In conclusion, despite the implication of TNF-a and IL-l in the pathogenesis
of sepsis and sepsis-associated organ failure, therapies directed against these cy-
tokines have failed to show clinical efficacy. Future studies may have to better
identify circumstances under which these therapies may become an adjunctive
therapeutic option in: septic patients.
References
1. American College of Chest Physicians/Society of Critical Care Medicine Consensus Confer-
therapies in sepsis. Crit Care Med 20:864-874
2. Abraham E, Wunderink R, Silverman H et al (1995) Efficacy and safety of monoclonal anti-
body to human tumor necrosis factor alpha in patients with sepsis syndrome. A randomized,
controlled, double-blind, multicenter clinical trial. JAMA 273:934-941
3. Fisher CJ, Agosti JM, Opal SM et al (1996) Treatment of septic shock with the tumor necro-
sis factor receptor: Fc fusion protein. The Soluble TNF Receptor Sepsis Study Group. N Engl
J Med 334:1697-1702
4. Schreuder WO, Schneider AJ, Groenveld ABJ et al (1989) Effects of dopamine vs norepi-
nephrine on hemodynamics in septic shock. Chest 95:1282-1288
5. Bone RC (1992) Phospholipids and their inhibitors: a critical evaluation of their role in the
treatment of sepsis. Crit Care Med 20:884-890
6. Bone RC (1992) Inhibitors of complement and neutrophils: A critical evaluation of their role
in the treatment of sepsis. Crit Care Med 20:891-898
7. Bone RC (1992) Modulators of coagulation. A critical appraisal of their role in sepsis. Arch
InternMed 152:1381-1389
8. Cobb JP, Cunnion RE, Danner RL (1993) Nitric oxide as a target for therapy in septic shock.
9. Blackwell TS, Christman JW (1996) Sepsis and cytokines: current status. Br J Anaesth
77:110-117
to. Schroder J, Stuber F, Gallati H et al (1995) Pattern of soluble TNF receptors I and n in sep-
sis. Infection 23:143-148
11. Zeni F, Vindirnian M, Pain P et al (1995) Antiinflammatory and proinflammatory cytokines
in patients with severe sepsis [letter]. J Infect Dis 172: 1171-1172
12. Goldie AS, Fearon KC, Ross JA et al (1995) Natural cytokine antagonists and endogenous
antiendotoxin core antibodies in sepsis syndrome. The Sepsis Intervention Group. JAMA
274:172-177
13. Moldawer LL (1993) Interleukin-l, TNF alpha and their naturally occurring antagonists in
sepsis. Blood Purif 11:128-133
14. Dhainaut JF, Tenaillon A, Le Tulzo Y et al (1994) Platelet-activating factor receptor antag-
onist BN 52021 in the treatment of severe sepsis: a randomized, double-blind, placebo-
controlled, multicenter clinical trial. BN 52021 Sepsis Study Group. Crit Care Med 22:
1720-1728
15. Tracey KJ, Cerami A (1993) Tumor necrosis factor: an updated review of its biology. Crit
Care Med 21:S415-22
16. Michie HR, Manogue KR, Spriggs DR et al (1988) Detection of tumor necrosis factor after
endotoxin administration. N Engl J Med 318:1481-1486
17. Eichenholz PW, Eichacker PQ, Hoffman WD et al (1992) Tumor necrosis factor challenges
in canines: Patterns of cardiovascular dysfunction. Am J PhysioI263:H668-H675
18. Tracey KJ, Fong Y, Hesse"DG, et al (1987) Anti-cachectintTNF monoclonal antibodies pre-
vent septic shock during lethal bacteraemia. Nature 330:662-664
19. Beutler B, Millsark IW, Cerami AC (1985) Passive Immunization against cachectinltumor
necrosis factor protects mice from lethal effects of endotoxin. Science 229:869-871
20. Gardlund B, Sjolin J, Nilsson A et al (1995) Plasma levels of cytokines in primary septic
shock in humans: correlation with disease severity. J Infect Dis 172:296-301
21. Herrmann JL, Blanchard H, Brunengo P et al (1994) TNF alpha, IL-l beta and IL-6 plasma
levels in neutropenic patients after onset of fever and correlation with the C-reactive protein
(CRP) kinetic values. Infection 22:309-315
22. Fisher CJ Jr, Opal SM, Dhainaut JF et al (1993) Influence of an anti-tumor necrosis factor
monoclonal antibody on cytokine levels in patients with sepsis. The CB0006 Sepsis Syn-
drome Study Group. Crit Care Med 21:318-327
23. Cohen J, Carlet J, The INTERSEPT Study Group (1996) INTERSEPT: An international,
multicenter, placebo-controlled trial of monoclonal antibody to human tumor necrosis factor
alpha in patients with sepsis. Crit Care Med 24: 1431-1440
24. Reinhart K, Wiegend-Lohnert C, Grimminger F et al (1996) Assessment of the safety and ef-
ficacy of the monoclonal anti-tumor necrosis factor antibody-fragment, MAK 195F, in pa-
tients with sepsis and septic shock: A multicenter, randomized, placebo-controlled, dose-
ranging study. Crit Care Med 24:733-742
25. Abraham E, Glauser MP, Butler T et al (1997) Tumor necrosis factor receptor fusion protein
in the treatment of patients with severe sepsis and septic shock. JAMA 277:1531-1538
26. Fekade D, Knox K, Hussein K et al (1996) Prevention of Jarisch-Herxheimer reactions by
treatment with antibodies against tumor necrosis factor alpha. N Engl J Med 335:311-315
27. Dinarello CA (1991) Interelukin-l and interleukin-l antagonism. Blood 77:1627-1652
28. Cannon JG, Tompkins RG, Gelfand JA et al (1990) Circulating interleukin-l and tumor
necrosis factor in septic shock and experimental endotoxin fever. J Infect Dis 160:79-84
29. Girardin E, Grau GE, Dayer J-M et al (1988) Tumor necrosis factor and interleukin-l in the
serum of children with severe infectious purpura. N Engl J Med 319:397-400
30. Casey LC, Balk RA, Bone RC (1993) Plasma cytokine and endotoxin levels correlate with
survival in patients with the sepsis syndrome. Ann Intern Med 119:771-778
31. Calandra T, Baumgartner JD, Grau GE et al (1990) Prognostic values of tumor necrosis fac-
tor/cachectin, interleukin-l, interferon-alpha, and interferon-gamma in the serum of patients
with septic shock. J Infect Dis 161:98288-98987
32. Ohlsson K, Bjork P, Bergenfeldt M et al (1990) Interleukin-lreceptor antagonist reduces
mortality from endotoxin shock. Nature 348:550-552
Therapies Directed against TNF-a and IL-l during Sepsis 121
33. Fischer E, Marano MA, Van Zee KJ et al (1992) Interleukin-l receptor antagonist blockade
improves survival and hemodynamic performance in Escherichia coli septic shock, but fails
to alter host response to sublethal endotoxaemia. J Clin Invest 89: 1551-1557
34. Fisher CJ Jr, Slotman GJ, Opal SM et al (1994) Initial evaluation of human recombinant in-
terleukin-l receptor antagonist in the treatment of sepsis syndrome: a randomized, open-la-
bel, placebo-controlled multicenter trial. The IL-IRA Sepsis Syndrome Study Group. Crit
Care Med 22:12-21
35. Fisher CJ, Dhainaut JF, Opal SM et al (1994) Recombinant human interleukin 1 receptor an-
tagonist in the treatment of patients with sepsis syndrome. Results from a randomized, dou-
ble-blind, placebo-controlled trial. Phase III rhIL-lra Sepsis Syndrome Study Group. JAMA
271:1836-1843
36. Opal SM, Fisher CJ, and Dhainaut JF (1997) The confirmatory interleukin-l receptor antago-
nist trial in sever sepsis: a phase III randomized, double-blind, placebo-controlled, multicen-
ter trial. Crit Care Med 25: 1115-1124
37. Lukacs NW, Strieter RM, Chensue SW et al (1995) TNF-alpha mediates recruitment of neu-
trophils and eosinophils during airway inflarumation. J Immunol 154:5411-5417
38. Moore FD, Socher SH, Davis C (1991) Tumor necrosis factor and endotoxin can cause neu-
trophil activation through separate pathways. Arch Surg 126:70-73
39. Alexander HR, Sheppard BC, Jenson JC et al (1991) Treatment with recombinant human tu-
mor necrosis factor-alpha protects rats against lethality, hypotension, and hypothermia of
gram-negative sepsis. J Clin Invest 88:34-39
40. Nakane A, Minagawa T, Kato K (1988) Endogenous tumor necrosis factor (cachectin) is
essential to host resistance against Listeria monocytogenes infection. Infect Immun 56:
2563-2569
41. Eschtenacher B, Mannel DN, Hultner L (1996) Critical protective role of mast cells in a mod-
el of acute septic peritonitis. Nature 381:75-77
42. Chorinchath BB, Kong LY, Mao L et al (1996) Age-associated differences in TNF-alpha and
nitric oxide production in endotoxic mice. J ImmunoI156:1525-1530
43. Stuber F, Petersen M, Bokelmann F et al (1996) A genomic polymorphism within the tumor
necrosis factor locus influences plasma tumor necrosis factor-alpha concentrations and out-
come of patients with severe sepsis [see comments]. Crit Care Med 24:381-384
44. Assicot M, Gendrel D, Carsin H et al (1993) High serum procalcitonin concentrations in pa-
tients with sepsis and infection. Lancet 341:515-518
Effects of Adrenergic Agents on the
Hepato-Splanchnic Circulation: An Update
D. DE BACKER, J.-L. VINCENT
Recent studies [1, 2] have implicated the hepato-splanchnic area in the devel-
opment of multiple organ failure. Important alterations can take place in the
gut and the liver during sepsis, A decreased gastric intramucosal pH (pHi),
thought to reflect inadequate gut perfusion, is ofteIi encountered in sepsis and
is associated with increased mortality rates [3, 4]. Liver dysfunction is also
common in sepsis and alterations in liver blood flow may contribute to it. The
hepatic venous oxygen saturation (Sh0 2) is often decreased [5-9], reflecting an
imbalance between oxygen demand and supply in the hepato-splanchnic area.
Also, the liver could be a major source of cytokine release [10], possibly trig-
gered by hypoxia.
Several mechanisms have been implicated in these disturbances in the hepa-
to-splanchnic area. First, blood flow could be diverted away from the splanchnic
area in order to preserve blood flow to the heart and the brain. Second, the
splanchnic region contributes to a larger fraction of total body oxygen consump-
tion in septic patients [6, 11]. In particular, liver metabolism is increased out of
proportion to the increase in liver blood flow so that Sh02 is usually decreased
[5, 6, 12, 13]. Finally the gut mucosa could be more sensitive to a reduction in
blood flow. The countercurrent exchange mechanism promotes O2 diffusion be-
tween the artery and the vein at the base of the villus so that the top could be de-
prived of 02. Hence the hepato-splanchnic area could be more sensitive to blood
flow alterations in sepsis. Vasoactive agents could further interfere with the oxy-
gen balance in the hepato-splanchnic area, increasing oxygen demand, and alter-
ing the distribution of blood flow within or between organs.
We will briefly review recent data, both in animals and in humans, to better
understand the effects of various vasoactive agents on the hepato-splanchnic
area.
Effects of dopamine
Dopamine has been proposed to increase hepato-splanchnic blood flow through
activation of the dopaminergic receptors [14-17]. Several studies have con-
firmed that dopamine can increase hepato-splanchnic blood flow in critically ill
124 D. De Backer, J.-L. Vincent
patients. Ruokonen et al. [9] observed that dopamine administered to increase

mean blood pressure to 70 mmHg, significantly increased hepato-splanchnic
blood flow and oxygen consumption (V0 2). Nevertheless, there were substantial
interindividual variations. On the contrary, Maynard et aI. [18] did not observe
any significant change in hepato-splanchnic blood flow during dopamine ad-
ministration in critically ill patients. More recently, Meier-HeUmann et aI. [17]
studied the effects of low-dose dopamine (2.5-3.0 mcg/kg.min) in patients with
hyperdynamic septic shock already treated with norepinephrine. Dopamine se-
lectively increased hepato-splanchnic blood flow in patients in whom the ratio
between hepato-splanchnic blood flow and cardiac index was lower than 0.3,
but did not alter blood flow in the other patients.
Recent studies have questioned the role of dopamine on the microcirculation.
Segal et al. [19] observed that dopamine administration in pigs hastened the on-
set of gut hypoxia during reductions in blood flow. However, Germann et al.
[16] reported in a hyperdynamic porcine endotoxic shock model that dopamine,
administered at doses of 2.5, 5, 10 and 20 mcg/kg.min, increased gut mucosal
P02 in a dose dependent manner. The effects on mucosal blood flow, estimated
by pHi measurements, were also variable. In septic patients, Marik and Mo-
hedin [20] reported that dopamine, administered to restore arterial pressure, de-
creased pHi. Also Olson et al. [21] observed that low dose dopamine adminis-
tration did not change pHi. Finally, Neviere et al. [22] reported that 5
mcg/kg.min dopamine decreased gut mucosal blood flow measured by laser
Doppler technique but did not significantly alter pHi.
In summary, hepato-splanchnic blood flow remains unchanged or increases
during dopamine administration while mucosal blood flow and pHi decreases
or, at best, remains unchanged.
Effects of norepinephrine
The data concerning norepinephrine are scarce. Norepinephrine is a cate-
cholamine with predominant alpha but also beta adrenergic effects. In hypody-
namic septic animals, norepinephrine altered mesenteric perfusion, principally
through an alpha-adrenergic effect [23]. However, in a fully fluid resuscitated
endotoxic shock model in the dog, Zhang et al. [24] observed that norepineph-
rine increased mesenteric and hepatic blood flow. Furthemore, norepinephrine
increased oxygen extraction capabilities so that the liver critical D02 was de-
creased [24]. These effects seem to be due more to the beta- than alpha-effects
since phenylephrine, a pure alpha-agonist agent, did not alter oxygen extraction
capabilities and critical D02 [25].
In septic patients, Ruokonen et al. [9] reported that norepinephrine adminis-
tration increased hepato-splanchnic blood flow and V0 2• However, the effects
of norepinephrine on splanchnic blood flow were not uniform. Recently, Meier-
Hellmann [26] reported that hepato-splanchnic D02 and V0 2 was higher in pa-
Effects of Adrenergic Agents on the Hepato-Splanchnic Circulation: An Update 125
tients with septic shock treated with norepinephrine than in patients with severe
sepsis. However, the pHi was similar in both groups of patients. In a prospective
study, Marik and Mohedin [20] observed that norepinephrine, administered in
order to restore blood pressure in hypotensive septic patients, increased pHi.
These results await confIrmation, since to our knowledge, no other prospective
study has analyzed the effects of norepinephrine on pHi.
Effects of epinephrine
At low doses, epinephrine has a predominant beta-l and beta-2 adrenergic effect
while at moderate and high doses the alpha-l vasoconstrictor effect predomi-
nates.
In patients with septic shock, Meier-Hellmann et al. [7] demonstrated that
changing the catecholamine regimen from a combination of dobutamine and
norepinephrine to epinephrine alone increased the gradient between mixed ve-
nous and hepatic venous O2 saturations. These authors also reported that frac-
tional blood flow and pHi decreased while hepatic venous lactate increased. Us-
ing a similar study design, Levy et al. [27] reported that epinephrine increased
arterial lactate levels and decreased pHi in septic patients. However, these ef-
fects were transient. Thus the data, although scarce, consistently demonstrate a
deleterious effect of epinephrine on hepato-splanchnic blood flow and metabo-
lism, at least transiently.
Effects of dobutamine
In control conditions, beta adrenergic agents usually increase mesenteric and
hepatic blood flow [28, 29]. This effect is preserved in septic animals. Fink et al.
[30], observed that dobutamine increased mesenteric blood flow and prevented
alterations in ileal permeability and acidosis without influencing gut V0 2 . How-
ever, these effects were observed with high doses of dobutamine. In a model of
hyperdynamic endotoxic shock in the dog, De Backer et al. [31] observed that
dobutamine at doses of 5 and 10 mcg/kg.min increased mesenteric D02 and pre-
vented the increase in ileal PC0 2 gap. Interestingly the effects of dobutamine
were similar at the dose of 5 and 10 mcg/kg.min. Recently, in another model of
endotoxic shock in the pig, Neviere et al. [32] also observed that a 5
mcg/kg.min dobutamine infusion increased gastric mucosal blood flow and lim-
ited the increase in mucosal-arterial PC02 difference. De Backer et al. [33] also
observed that dobutamine at doses of 5 and 10 mcg/kg.min increased total liver
D02 , V0 2 and lactate consumption. This increase in liver blood flow and me-
tabolism was accompanied by an increase in hepatic venous oxygen saturation
suggesting an improved balance between oxygen supply and demand. Webb et
al. [34] and Tighe et al. [35] observed that dobutamine altered liver ultrastruc-
ture in a porcine model of fecal peritonitis but portal blood flow was not in-
creased by dobutamine in this model.
A number of investigators have studied the effects of dobutamine on
splanchnic perfusion in septic patients [13, 22, 33, 36-39]. Although pHi re-
mained stable in some studies [13, 39], most authors observed an improve-
ment in pHi during dobutamine administration in septic patients [22, 36-38].
Silverman et al. [37] reported that pHi increased in septic patients with a low
baseline pHi, but remained stable in septic patients with a normal pHi. Gutier-
rez et al. [36] reported increases in pHi and Neviere et al. [22] increases in
gastric mucosal blood flow, measured by laser Doppler. Hepato-splanchnic
blood flow, measured by the indocyanine green clearance technique, increased
in hemodynamically stable [33] as well as unstable [13] septic patients. This
effect was associated with significant increases in hepato-splanchnic D02 and
V02 [33].
In summary, experimental and clinical studies indicate that dobutamine ad-
ministration consistently increases hepato-splanchnic blood flow and usually in-
creases pHi.
Effects of dopexamine
Dopexamine combines beta-2, some beta-1 and dopaminergic adrenergic recep-
tor agonist activity without any alpha-agonist activity. Several experimental
studies have demonstrated that dopexamine can improve gut oxygenation. In
rabbits with septic shock, Lund et al. [40] reported that dopexamine increased
liver and gut mucosal tissue P02 . In another study [41], gut lactate production
could be reversed during dopexamine administration in endotoxemic dogs.
Moreover, Webb et al. [34] and Tighe et al. [35] observed that dopexamine pre-
served liver ultrastructure in a model of fecal peritonitis in pigs.
In critically ill patients, dopexamine increased hepato-splanchnic blood
flow determined by indocyanine green clearance and monoethylglycineexyli-
dide (MEGX) formation [18, 42]. However, the effects of dopexamine on pHi
have been inconsistent, since pHi increased in some studies [18,42] but not all
[43,44].
Hepato-splanchnic V02ID02 relationships

The use of hepatic vein catheterization offers the opportunity to investigate re-
gional V0 2ID0 2 relationships by measuring hepato-splanchnic blood flow and
hepatic venous oxygen saturation (Sh0 2). In 42 hemodynamically stable septic
patients, we [45] studied the effects of dobutamine at a rate of 5 and 10
mcg/kg.min on hepato-splanchnic D02 and V0 2 . Hepato-splanchnic V0 2 in-
HEPATO·SPLANCHNIC
V02 PATIENT II
50 Sv0272%
mLlmin.M2 Sh0245%
40
30
PATIENT I
20 Sv0270%
Sh0268%
10
O+---~----r---~---.----~--~
o 50 100 150
HEPATO.SPLANCHNIC 002 mLlmin.M2
Fig. 1. Example of hepato-splanchnic V0 2ID02 relationships in 2 patients
creased only in patients with an increased gradient between Sv02 and Sh02,
even though dobutamine similarly increased in hepato-splanchnic D0 2 in the 2
groups (Fig. 1). Also the slope of the hepato-splanchnic VOiD02 relationship
was inversely related to the difference between Sv02 and Sh02. On the contrary,
the response in global hemodynamic parameters was similar in both groups.
These results suggest that hepato-splanchnic V0 21D02 dependency could be ob-
served in some septic patients even when they appear hemodynamically stable.
Similarly Steffes et al. [12] observed that red blood cell transfusions increased
hepato-splanchnic V0 2 in some septic patients but they did not attribute these
changes to V0 21D02 dependency since splanchnic lactate consumption re-
mained stable. However, the interpretation of splanchnic lactate consumption is
very difficult in patients since it represents the balance between gut lactate pro-
duction and liver lactate consumption, and these parameters could be affected
differently by the various interventions. Recently, Reinelt et al. [46, 47] ob-
served that the change from norepinephrine to phenylephrine, depriving these
patients of beta adrenergic stimulation, reduced hepato-splanchnic D02 (from
182 ± 40 to 124 ± 21 ml/min.M2, p < 0.05) and V0 2 (from 80 ± 15 to 69 ± 11
ml/min.M2, p = ns), but also reduced lactate consumption by the liver directly
assessed by stable isotopes (from 603 ± 292 to 420 ± 164 mcM/min.M2). This
suggests that VOiD02 dependency occurred in these patients.
Conclusions
Hepato-splanchnic hemodynamics can be significantly altered in septic patients,
as indicated by decreased Sh02 and pHi, and, possibly in some cases, hepato-
splanchnic VOiD02 dependency. The effects of the different catecholamines on
hepato-splanchnic blood flow are variable and can not be inferred from the eval-
uation of global hemodynamic parameters.
References
1. Landow L, Andersen LW (1994) Splanchnic ischaemia and its role in multiple organ failure.
Acta Anaesthesiol Scand 38:626-639
2. Mythen MG, Webb AR (1994) The role of gut mucosal hypoperfusion in the pathogenesis of
post-operative organ dysfunction. Intensive Care Med 20:203-209
3. Maynard N, Bihari D, Beale R et al (1993) Assessment of splanchnic oxygenation by gastric
tonometry in patients with acute circulatory failure. JAMA 270:1203-1210
4. Friedman G, Berlot G, Kahn RJ, Vincent JL (1995) Combined measurements of blood lactate
concentrations and gastric intramucosal pH in patients with severe sepsis. Crit Care Med
23:1184-1193
5. Dahn MS, Lange MP, Jacobs LA (1988) Central mixed and splanchnic venous oxygen satu-
ration monitoring. Intensive Care Med 14:373-378
6. Dahn MS, Lange MP, Wilson RF et al (1990) Hepatic blood flow and splanchnic oxygen con-
sumption measurements in clinical sepsis. Surgery 107:295-301
7. Meier-Hellmann A, Hannemann L, Specht M et al (1994) The relationship between mixed
venous and hepatic venous O 2 saturation in patients with septic shock. In: Vaupel P (ed) Oxy-
gen transport to the tissues XV. Plenum Press, New York, pp 701-707
8. Ruokonen E, Takala J, Uusaro A (1991) Effect of vasoactive treatment on the relationship be-
tween mixed venous and regional oxygen saturation. Crit Care Med 19:1365-1369
9. Ruokonen E, Takala J, Kari A et al (1993) Regional blood flow and oxygen transport in sep-
tic shock. Crit Care Med 21:1296-1303
10. Douzinas EE, Tsidemiadou PD, Pitaridis MT et al (1997) The regional production of cy-
tokines and lactate in sepsis-related multiple organ failure. Am J Respir Crit Care Med
155:53-59
11. Wilmore DW, Goodwin CW, Aulick LH et al (1980) Effect of injury and infection on visceral
metabolism and circulation. Ann Surg 192:491-504
12. Steffes CP, Dahn MS, Lange MP (1994) Oxygen transport-dependent splanchnic metabolism
in the sepsis syndrome. Arch Surg 129:46-52
13. Reinelt H, Radermacher P, Fischer G et al (1997) Effects of a dobutamine-induced increase
in splanchnic blood flow ort hepatic metabolic activity in patients with septic shock. Anesthe-
siology 86:818-824
14. Breslow MJ, Miller CF, Parker SD et al (1987) Effect of vasopressors on organ blood flow
during endotoxin shock in pigs. Am J PhysioI252:H291-H300
15. Bersten AD, Hersch M, Cheung H et al (1992) The effect of various sympathomimetics on
the regional circulations in hyperdynamic sepsis. Surgery 112:549-561
16. Germann R, Haisjackl M, Schwarz B et al (1997) Inotropic treatment and intestinal mucosal
tissue oxygenation in a model of porcine endotoxemia. Crit Care Med 25:1191-1197
17. Meier-Hellmann A, Bredle DL, Specht Met al (1997) The effects oflow-dose dopamine on
splanchnic blood flow and oxygen uptake in patients with septic shock. Intensive Care Med
23:31-37
18. Maynard ND, Bihari DJ, Dalton RN et al (1995) Increasing splanchnic blood flow in the crit-
ically ill. Chest 108:1648-1654
19. Segal JM, Phang T, Walley KR (1992) Low-dose dopamine hastens onset of gut ischemia in a
porcine model of hemorrhagic shock. J Appl Physiol 73: 1159-1164
20. Marik PE, Mohedin J (1994) The contrasting effects of dopamine and norepinephrine on sys-
temic and splanchnic oxygen utilization in hyperdynamic sepsis. JAMA 272: 1354-1357
21. Olson D, Pohlman A, Hall JB (1996) Administration of low-dose dopamine to nonoliguric
patients with sepsis syndrome does not raise gastric intramucosal pH nor improve creatinine
clearance. Am J Respir Crit Care Med 154: 1664-1670
22. Neviere R, Mathieu D, Chagnon JL et al (1996) The contrasting effects of dobutamine and
dopamine on gastric mucosal perfusion in septic patients. Am J Respir Crit Care Med
154:1684-1688
23. Hussain SNA, Rutledge F, Roussos C, Magder S (1988) Effects of norepinephrine and fluid
administration on the selective blood flow distribution in endotoxic shock. J Crit Care 1:
32-42
24. Zhang H, Smail N, Cabral A (1997) Effects of norepinephrine on regional blood flow and
oxygen extraction capabilities during endotoxic shock. Am J Respir Crit Care Med 155:
1965-1971
25. Zhang H, de Jongh R, De Backer D et al (1997) Effects of adrenergic agonists on hepato-
splanchnic perfusion and oxygen extraction in endotoxic shock. Am J Respir Crit Care Med
155:A402 (Abstract)
26. Meier-Hellmann A, Reinhart K, Bredle DL et al (1997) Epinephrine impairs splanchnic per-
fusion in septic shock. Crit Care Med 25:399-404
27. Levy B, Bollaert PE, Charpentier C et al (1997) Comparison of norepinephrine and dobuta-
mine to epinephrine for hemodynamics, lactate metabolism, and gastric tonometric variables
in septic shock. Intensive Care Med 23:282-287
28. Kvietys PR, Granger DN (1982) Vasoactive agents and splanchnic oxygen uptake. Am J
Physiol 243:G I-G9
29. Braatveld GD, Stanners FA, Halliwell M, Corrall RJ (1993) Splanchnic blood flow in man:
Evidence for mediation via a ~-adrenergic mechanism. Clin Science 84:201-207
30. Fink MP, Kaups KL, Wang H, Rothschild HR (1991) Maintenance of superior mesenteric ar-
terial perfusion prevents increased intestinal mucosal permeability in endotoxic pigs. Surgery
110:154-161
31. De Backer D, Zhang H, Manikis P, Vincent JL (1996) Regional effects of dobutamine in en-
dotoxic shock. J Surg Res 65:93-100
32. Neviere R, Chagnon JL, Vallet Bet al (1997) Dobutamine improves gastrointestinal mucosal
blood flow in a porcine model of endotoxic shock. Crit Care Med 25: 1371-1377
33. De Backer D, Creteur J, Smail N et al (1996) Dobutamine increases hepatosplanchnic blood
flow in septic patients. Am J Respir Crit Care Med 153:A125 (Abstract)
34. Webb AR, Moss RF, Tighe D et al (1992) The effects of dobutamine, dopexamine and fluid
on hepatic histological responses to porcine faecal peritonitis. Intensive Care Med 17:
487-493
35. Tighe D, Moss R, Heywood G et al (1995) Goal-directed therapy with dopexamine, dobuta-
mine, and volume expansion: effects of systemic oxygen transport on hepatic ultrastructure in
porcine sepsis. Crit Care Med 23: 1997-2007
36. Gutierrez G, Clark C, Brown SD et al (1994) Effect of dobutamine on oxygen consumption
and gastric mucosal pH in septic patients. Am J Respir Crit Care Med 150:324-329
37. Silverman H, Tuma P (1992) Gastric tonometry in patients with sepsis: Effects of dobutamine
infusions and packed red blood cell transfusions. Chest 102: 184-188
38. Esen F, Telci L, <;akar N et al (1996) Evaluation of gastric intramucosal pH measurements
with tissue oxygenation indices in patients with severe sepsis. Clin Intens Care 7: 180-189
39. Radermacher P, Buhl R, Klein M et al (1995) The effects of prostacyclin on gastric intramu-
cosal pH in patients with septic shock. Intensive Care Med 21 :414-421
40. Lund N, de Asia RJ, Cladis F et al (1995) Dopexamine hydrochloride in septic shock: effects
on oxygen delivery and oxygenation of gut, liver, and muscle. J Trauma 38:767-775
41. Cain SM, Curtis SE (1991) Systemic and regional oxygen uptake and delivery and lactate
flux in endotoxic dogs infused with dopexamine. Crit Care Med 19:1552-1560
42. Smithies M, Yee TH, Jacson L et al (1994) Protecting the gut and the liver in the critically ill:
effects of dopexarnine. Crit Care Med 22:789-795
43. Trinder TJ, Lavery GG, Fee PH, Lowry KG (1995) Correction of splanchnic oxygen deficit
in the intensive care unit: dopexamine and colloid versus placebo. Anaesth Intens Care
23:178-182
44. Kuhly P, Oschrnann G, Hilpert J et al (1996) Dopexarnine does not change gastric and sig-
moid mucosal pH in critically ill patients. Clin Intens Care 7:58 (Abstract)
45. De Backer D, Creteur J, Smail N et al (1997) Hepato-splanchnic V0 2ID02 dependency in
septic patients: effects of dobutamine administration. Am J Respir Crit Care Med 155:A400
(Abstract)
46. Reinelt H, Kiefer P, Fischer G et al (1997) ~-receptor stimulation determines splanchnic per-
fusion in septic shock. Intensive Care Med 23:S80 (Abstract)
47. Reinelt H, Vogt J, Kiefer P et al (1997) Oz-supply determines liver metabolism in septic
shock. Intensive Care Med 23:S81 (Abstract)
When to Operate or to Stop Operating and to Plan
a Reoperation
A.E. BAUE
The indications for an operative procedure have remained much the same for a
number of years, but there are now several new indications for considering an
operation and particularly for c9nsidering a reoperation. The usual indications
for an elective operation must consider the risks versus the benefits. For emer-
gency operations the consideration must be whether or not the procedure is life-
saving and, again, risks versus benefits related to the procedure. What is the risk
for the patient, what are the potential benefits, and do they balance out as a win-
win situation for the patient and his or her family?
In recent years some of the usual indications for operation have been cloud-
ed because of end-of-life considerations and deterioration with the aging
process such as with Alzheimer's disease. In the United States this is particular-
ly evident for patients with Alzheimer's disease, severe senility, or total inca-
pacity who live in a nursing home, who do not recognize anyone, do not know
their families and are living a vegetative existence. If they develop a life-threat-
ening problem, which could be taken care of by an operation, is it ethical and is
it moral to carry out such a procedure and to make them suffer through it even
though they are going to go back to the same vegetative state that they were in
before? The fact that we can do an operation for an acute condition in such a
patient does not necessarily make it right or necessary. I have observed elderly
patients who were categorized as "Do Not Resuscitate" immediately after an
emergency or urgent operation. Can we propose "Do Not Resuscitate" preoper-
atively and avoid an operation that will not improve the patient's quality of
life? These are tremendous ethical dilemmas for surgeons taking care of such
patients.
Another consideration in risks versus benefits is with minimal operations.
Laparoscopy and thoracoscopic procedures now can be done at less risk for an
individual. For example, a post-traumatic hemothorax can be taken care of tho-
racoscopically, avoiding a full formal thoracotomy. The same is true for
empyema. The need for an abdominal exploration after trauma in stable pa-
tients with no specific indications for operation may be evaluated by a diag-
nostic laparoscopy. Staging of abdominal malignancy and the question of an
acute abdomen also could be determined with minimal surgical procedures.
132 A.E. Baue
They greatly reduce the risk for the operation and increase the potential benefit
for the patient.
There are now new indications for operation and/or reoperation which fall
into two categories. First, there are intensive care unit patients who mayor may
not have had a previous abdominal procedure because of trauma or an acute ab-
dominal problem. These patients, in the ICU, may be stable initially, but begin
to develop remote organ failure or worsening of organ function which could be
called MODS (multiple organ dysfunction syndrome). In such a situation there
may be very specific and real indications for operation or reoperation to try to
find a septic focus, an area of necrosis, or some other problems in the peritoneal
cavity which is causing the remote organ deterioration [1, 2].
Secondly, postoperative patients or patients with bowel obstruction who
have distention may have alterations in organ function based upon a great in-
crease in intraabdominal or intraperitoneal pressure. This has been known for
quite some time but it has not always been considered an important problem.
Now this has been called the abdominal compartment syndrome [3, 4]. It is a
very important consideration in patients with severe abdominal distension and
potential increase in pressure and a related decrease in organ function within
the peritoneal cavity.
There are now two major reasons for stopping an operation and planning on
a reoperation sometime later. The first reason is with abdominal trauma and ma-
jor organ injury with considerable blood loss. If the patient develops acidosis,
hypothermia and a coagulopathy (usually associated with blood loss and multi-
ple blood transfusions) there may be blood oozing from everywhere which can-
not be controlled. In such a situation it is now recognized that persistence in try-
ing to correct this situation during the operation in the operating room with the
abdomen and/or the chest open will not be successful. However, rapid conclu-
sion of the operation, bringing the abdominal wall together or using a silo, and
taking the patient to the intensive care unit, may allow rewarming the patient
and correction of the acidosis and coagulopathy. Then the patient can be taken
back to the operating room for a definitive conclusion of the procedure. Trauma
centers and trauma surgeons now recognize this possibility and it is a major rea-
son for stopping an operation and planning on a reoperation soon after the pa-
tient stabilizes in the intensive care unit [5].
The second reason for stopping an operation and planning on a reoperation is
for a patient with generalized peritonitis where reoperation is necessary to break
up foci of contamination, intraloop infection and the prevention of residual ab-
scesses. It is recognized that residual infection may not be preventable at the ini-
tial operation no matter how long one persists in working within the septic peri-
toneal cavity [1, 6].
These new approaches will be reviewed in detail as to the present standing,
indications, techniques and results. These approaches have led to improved pa-
When to Operate or to Stop Operating and to Plan a Reoperation 133
tient care and better survival for a number of patients after illness and/or opera-
tion or trauma.
Abdominal exploration for remote organ dysfunction or failure

When I first described multiple organ failure problems, I used several examples
of patients who had died with multiple organ difficulties [7]. One of these pa-
tients had a colon anastomosis which leaked and caused generalized peritonitis
- this is classic sepsis. Another patient had acute hemorrhagic pancreatitis, dete-
riorated and died rather quickly with what was a fairly pure inflammatory
process. A third patient had a low cardiac output after a double heart valve re-
placement and died, in essence, of a low output and organ ischemia. Thus we
know that these factors - sepsis, inflammation, ischemia, gangrene and necrosis
- can all contribute to multiple organ failure by mounting an inflammatory
process which then becomes dangerous.
Early on with multiple organ failure it was recognized that in many patients
sepsis, peritonitis or inflammation was a major cause of the difficulty. This is
particularly true with reports from institutions that had a high level of penetrat-
ing abdominal trauma and other problems of acute abdominal inflammatory or
infectious diseases. It was recognized then that if a patient with such a problem
had an immediate operation, all the recognized abnormalities were corrected,
and then the patient stabilized in the intensive care unit, all seemed to be going
well. If, however, following a period of stabilization, the patient then began to
experience difficulty with organ function, then there must be a recognized prob-
lem elsewhere. An example would be in a patient being supported by mechani-
cal ventilation who is stable, has not improved definitely but also whose pul-
monary function has not deteriorated. Such a patient then begins to develop
worsening respiratory function with a requirement for a higher Fi0 2 or other
problems attendant to ventilation, then one must think seriously about an infec-
tious process elsewhere. Remote organ dysfunction with sepsis elsewhere was
described by Polk et al. [2]. They recognized that, if this occurred, there should
be a "blind laparotomy" reoperation immediately and, in at least half of the pa-
tients, something would be found that could be corrected and the patient would
improve and survive. Now, with diagnostic possibilities, including ultrasound,
CT scans, and other studies of the chest and peritoneal cavity, such a blind reop-
eration should not be necessary.
The problems that could produce remote organ dysfunction include acute
aca1culous cholecystitis, bowel ischemia, abscesses in the pelvis, subphrenic re-
gion, subhepatic region, intraloop abscesses or retroperitoneum and finally bow-
el or gut perforation [1]. These and other problems, such as pancreatitis, can
produce an inflammatory process and mediator release which induces organ
damage and functional deterioration. These patients would have the setting of
elevated temperature and increased white count. They would look septic and
134 A.E. Baue
they would have other problems as well with a mass or collection problem iden-
tified. The most important concept, however, is to recognize remote organ dete-
rioration which may be caused by trouble within the peritoneal cavity [8].
A number of investigators have provided evidence for establishing the diag-
nosis of continuing abdominal sepsis by the presence or development of multi-
ple organ failure. Ferraris found that the most significant predictor of continuing
intraabdominal infection was unexplained single organ failure [9]. Others have
found that early operation in organ failure improved survival whereas delayed
operation furthered the peritonitis and worsened the prognosis [10, 11]. If, how-
ever, multiple organ failure has developed and then the focus of infection or
necrosis is removed, organ function may not improve or return [12, 13]. This
emphasizes the need for early identification of contributing factors.
The abdominal compartment syndrome

There is a long history of study in animals and people about the effects of in-
creased intraabdominal pressure on organ function and on the circulation. How-
ever, little was done about this clinically for many years. One of the earliest
studies was one by Thorington and Schmidt in 1923 on urinary output and
blood pressure changes resulting from experimental ascites [14]. They found,
for example, that a rise in intraabdominal pressure to 30 mmHg was uniformly
followed by complete suppression of urine output in dogs and an increase in ve-
na caval pressure. Thus, concern about intraabdominal pressure began with
studies in cirrhotic patients and animals with ascites [15, 16] and with the devel-
opment of the G suit for the prehospitalization treatment of shock by the United
States military. The effect of increased intraabdominal pressure on renal func-
tion in man was studied by Bradley and Bradley in 1947 [17]. They found that
effective renal plasma flow and glomerular filtration rate were reduced by in-
creased abdominal pressure along with reduced dye excretion. Massive ascites
could produce obstruction of the inferior vena cava. This was also implicated in
the hepatorenal syndrome with cirrhosis [16]. The first report of renal failure
from increased intraabdominal pressure in patients was by Richards et al., in
1983 [18]. They reported acute renal failure developing in four patients in asso-
ciation with increased intraabdominal pressure from postoperative hemorrhage.
Operative decompression of the abdomen resulted in polyuria and resolution of
the renal failure. This was followed in 1984 by the measurement of intraabdom-
inal pressure as a criterion for abdominal reexploration by Kron et al. [19]. Ab-
dominal pressure was measured by a manometer connected to a Foley catheter
in the urinary bladder. They found a correlation between intraabdominal pres-
sures above 25 mmHg and renal dysfunction and reported pressures above a 30
mmHg, resulting in oliguria in 11 postoperative patients. They recommend reex-
ploration and decompression of the abdomen if pressure rose to 25 mmHg. This
was first called the intraabdominal compartment syndrome in 1989 by Fietsam
et al. [3].
There have also been numerous studies in animals of the effects of increased
intraabdominal pressure on the circulation, on venous return, renal function, left
ventricular performance, regional blood flow, microcirculatory blood flow in the
gut, mucosal blood flow, and blood and tissue oxygen [20-25]. One of the
mechanisms of action of increased abdominal pressure may be upregulation of
plasma renin activity and aldosterone levels [26]. Increased intracranial pressure
has also been described with increased abdominal pressure as has pulmonary
dysfunction [27]. The physiological consequences of elevated intraabdominal
pressure were described by Schein et aI., in a collective review in 1995 [4].
These include an increase in heart rate, an increase in capillary wedge pressure,
an increase in peak airway pressure, central venous pressure, thoracic pleural
pressure, inferior vena cava pressure, renal vein pressure, and systemic vascular
resistance. There is decreased cardiac output and decreased venous return, vis-
ceral blood flow, renal blood flow, glomerular filtration rate, and abdominal
wall compliance.
The factors causing this problem could be spontaneous with peritonitis, a
ruptured abdominal aneurysm [3, 19], intestinal obstruction, etc. Postoperative
changes could be peritonitis, an abscess, acute gastric dilatation [28], or in-
traperitoneal hemorrhage [18]. After trauma, visceral edema could also be a fac-
tor and, of course, there are iatrogenic problems such as abdominal packing, re-
duction of massive hernias, etc. [29]. There are chronic causes such as large ab-
dominal tumors, ascites, etc. [30].
In order to recognize and treat the abdominal compartment syndrome in any-
one with abdominal distension, the bladder pressure should be measured [31].
Clinical observation may not be sufficient [32]. The answer, of course, with in-
creased pressure to 25 mmHg, is to take the patient back to the operating room
to operate or reoperate, open the abdomen and decompress it [33]. If the cause
is due to blood or fluid, and this can be corrected, then it may be possible to
close the abdomen. Frequently, however, it will be necessary to use some tech-
nique to leave the abdomen open for a brief time until the problem within the
peritoneal cavity is corrected. Various materials have been used for this purpose.
Marlex or PTFE is very expensive. Polypropylene sticks to the bowel and is no
longer used. Knitted Dexon will last only for three weeks. The best material
now seems to be poly glycolic acid. Woven vicryl mesh stays longer and seems
to be quite satisfactory. If the skin can be closed over the top of the mesh, even
though the fascia cannot be closed, this will help resolution of the intraabdomi-
nal process. Undermining of the skin in order to help close the skin or to close it
either with towel clips or with some other form of closure can be helpful in al-
lowing the process to resolve more quickly. As the pressure decreases, as the
bowel edema subsides and other factors lessen, staged reconstruction of the ab-
dominal wall may be necessary. It could be done in one stage if everything re-
solves quickly. Otherwise, it may take several stages.
A silastic membrane has also been used by many. The large wound is cov-
ered with a saline-soaked gauze dressing that is changed several times daily to
136 A.E. Baue
maintain the viability of exposed intraabdominal wall, subcutaneous tissue and

fascia. The impermeable nature of this silastic, however, may be a drawback in
not allowing drainage of fluid and other things, particularly if they are contami-
nated. There are many differences of opinion about materials and methods to be
used [34-37]. The important principle is to recognize the syndrome before irre-
versible organ damage occurs.
Trauma and coagulopathy - The staged or abbreviated laparotomy

Severe abdominal and thoracic trauma may produce major organ injury to the
liver, the spleen, the gut and the lungs, with massive blood loss and difficulty in
controlling the situation operatively. This concept began with hepatic injury and
the necessity to pack major liver injuries to control blood loss. For a time this
was condemned; however, it became necessary to do so in order to save pa-
tients' lives. One of the first definitive approaches for this was by Feliciano et
al., from the Ben Taub Hospital in Houston, Texas [38]. Patients with extensive
liver injuries survived by packing the injury and returning the patient to the op-
erating room after things were under better control. Shortly thereafter Svoboda
et aI., presented compelling information and experiences on severe liver trauma
in the face of coagulopathy and made the case for temporary packing and early
reexploration [39]. Since then there have been numerous reports of this ap-
proach. It has been called damage control [40], abbreviated laparotomy [41],
planned reoperation [42], and of course staged laparotomy which seems to be
the best term. Burch et al. [41] reported on 200 trauma patients with abbreviated
laparotomy in which 98 (49%) survived to undergo reoperation and 66 of the 98
(67%) survived to leave the hospital. The supposition is that all would have died
without this approach. Hirshberg et al. [42] found a mortality rate of 58% in
such patients. Moore recently reviewed these problems in great detail in the
Thomas Orr Memorial Lecture to the Southwestern Surgical Congress [5]. In
this presentation he and his group described what they called the "bloody vi-
cious cycle" (Fig. 1). They described this problem with massive torso trauma
frequently involving the liver and other abdominal or thoracic organs. Coagu-
lopathy develops with bleeding from everywhere. The patients frequently have
had massive rapid blood transfusions, evidence of persistent shock, progressive
metabolic acidosis with a low pH and refractory core hypothermia. The major
criteria for these phenomena as described by Moore are shown in Table 1. When
these criteria are met, or the bloody vicious cycle occurs with continued hemor-
rhage from everywhere, it is best to try to back off, conclude the operation very
quickly, remove the patient from the operating room to the intensive care unit
where the patient can be warmed, the coagulation problem is corrected, and oth-
er things brought into control. Then, at an appropriate and a more elective time,
the patient can be taken back to the operating room to do more definitive re-
pairs. The primary objectives of this approach are to arrest bleeding and then al-
"THE BLOODY VICIOUS CYCLE"
Fig. 1. The pathogenesis of the bloody vicious cycle following major torso injury is multifactorial,
but usually manifests as a triad of refractory coagulopathy, progressive hypothermia, and persis-
tent metabolic acidosis. (Reproduced with permission from [5])
Table 1. Major torso trauma
Coagulopathy - bleeding from everywhere

PT > 2 x normal
PIT > 2 x normal
Massive Rapid Transfusion
> 10 units/4 hours
Persistent Cellular Shock
V0 2 < 110 mlIminIM 2
Lactate> 5 mmollL
Progressive metabolic acidosis
pH < 7.20
Base deficit> 14 mEq/L
Refractory Core Hypothermia
<34°C
138 A.E. Baue
low correction of the inciting coagulopathy. Emergency arrest of bleeding

should consist of ligation, clamps which can be left in, intravascular balloons
for hepatic vein or caval injury, tamponade by packing and other means. Conta-
mination should be limited by excluding bowel injuries; there is no need to do
primary anastomoses at this time. Finally, it is necessary to enclose the abdomi-
nal contents to protect the viscera. This can be done with plastic sheeting, using
towel clips to bring the skin over the top of this. Usually the skin will be loose
enough to do this and not produce an abdominal compartment syndrome. Physi-
ologic restoration in the intensive care unit then consists of trying to increase the
oxygen delivery index, using inotropes and cardiovascular support, if needed.
Rewarming the patient is critical until the core temperature is above 35°C [43].
Blood components are given to raise the prothrombin time and the partial
thromboplastin time back toward normal with a platelet count above 100,000
and a fibrinogen above 100 mm %. Mechanical ventilation is continued during
this time. .
Careful observation for potential development of the abdominal compart-
ment syndrome is important. Timing of return to the operating room for the next
stage of reexploration depends upon the response of the patient. Angiography
with selective embolization may be necessary for major liver or other organ in-
jury or pelvic fractures. Usually patients are returned to the operating room
within 24 hours to complete cleaning of the peritoneal cavity, restoring bowel
continuity and other matters. If there is extensive bowel edema, then reexplo-
ration may be delayed for 48 to 72 hours until such edema subsides to some ex-
tent. After cleaning the peritoneal cavity, abdominal wall reconstruction will
have to be carried out and usually this will require a synthetic mesh template.
The Moores in Denver have encountered enteric fistulae with absorbable poly-
galactin and polyglycolic acid as well as polypropylene and nylon mesh when
they have been put over an exposed bowel. Polytetrafluoroethylene (PTFE)
avoids fistulae because it does not adhere but it is terribly expensive, as men-
tioned early. Latex rubber and silastic are all alternatives but, of course, they re-
quire removal. Moore recommends individualizing trying to cover the gut with
omentum if possible, then an absorbable mesh which can be sutured to the fas-
cia [5]. If there is inadequate omentum, then latex rubber is placed over the ex-
posed gut as the skin is gradually approximated. The ingenuity of the surgeon
may be tested in carrying out such reconstructions. Although the mortality re-
mains high (50%) with this approach, it is likely that the mortality would be
100% if the surgeon persisted in the operating room with the original operation.
Peritonitis and planned reoperation

The experts in the treatment of peritonitis now divide the problem into three
types of abnormalities. First, there is primary peritonitis in which bacteria gain
access to the peritoneal cavity either through lymphatics or via a blood-borne
mechanism. This occurs very rarely now and is primarily a disease of patients
with cirrhosis and ascites which becomes infected. Patients undergoing peri-
toneal dialysis also may have primary peritonitis. Secondary peritonitis is de-
fined as an infection occurring secondary to a disease process within the peri-
toneal cavity such as diverticulitis, appendicitis, perforated ulcer, etc. Tertiary
peritonitis represents a failure of the initial treatment for secondary peritonitis
with bacteria forming a chronic infectious state of multiple abscesses and other
problems.
The following discussion will deal primarily with secondary peritonitis
which has become generalized, rather than localized, and which is not amenable
to drainage of the peritoneal cavity. This may be considered a form of tertiary
peritonitis as well or may be considered to lead to tertiary peritonitis.
History of the development of treatment of peritonitis

For acute generalized suppurative peritonitis the results have never been totally
satisfactory. Therefore, surgeons have sought better ways to improve the results
and get more patients to survive. In the 1970's continuous peritoneal lavage was
tried through catheters inserted within the peritoneal cavity for fluid administra-
tion and drains to remove it. It was then recognized that this was not very worth-
while [44]. Some continued the practice, however, of continuous flow irrigation
[45]. In 1975 Hudspeth popularized the concept of radical peritoneal debride-
ment, an extensive operation removing material on the bowel wall, on the
mesentery, and so forth, debriding all the necrotic and/or inflammatory mem-
branes and tissue [46]. Polk et ai. in 1980 showed in a randomized trial that this
was not worthwhile either [47]. In 1979 the concept of leaving the abdomen
open was described by a number of French surgeons. One of the early reports in
the English literature was by Steinberg entitled "On leaving the peritoneal cavi-
ty open in acute generalized suppurative peritonitis" [48]. This approach has
been used in recent years in a number of different ways using packing, mesh,
plastic materials, skin closure with towel clips and other approaches. Along with
that, the concept of relaparotomy or another laparotomy on demand, when the
clinical circumstances demanded it, was also considered.
Pichlmayr et al. in 1983 developed the concept of dorsal ventral lavage with
a pallisade type of effect [49]. This concept has been pursued further by
Losanoff et aI., in which the pallisade of dorsal ventral lavage is applied after
completing abdominal exploration and eradicating all septic foci [50]. A screen
is fashioned from six to ten wide-bore silicone drains transfixed with stainless
wire which keeps them together. The intestines are covered in this modification
of Pichlmayr's technique by a large polyethylene foil with multiple perfora-
tions. It is placed far under the fascial edges. The screen is positioned over the
polyethylene foil and secured in place with retention sutures to avoid eviscera-
tion. The abdomen is thus left open with a screen and underlying polyethylene
140 A.E. Baue
foil covering the viscera. Wide-bore inflow drains enter the abdomen depen-
dently and laterally and three or four small-bore drains enter the space between
the screen and the foil and then the laparostomy is closed with gauze. These
drains are then used for continuous irrigation. This is a complicated technique
which was modified from the original Pichlmayr technique which had led to the
development of intestinal fistulae. The authors believe that this approach is sat-
isfactory.
Planned relaparotomy was promulgated in 1986 by Teichman et al., with a
technique of ettappenlavage, leaving the abdomen open with planned daily rela-
parotomy and irrigation [51]. More recently, the staged abdominal repair or
STAR approach has been popularized. In a recent review of leaving the ab-
domen open Wittmann identified and used a new term called "open abdominos-
tomy" in which there is no reapproximation of the abdominal fascia or wall. A
second method is a mesh abdominostomy in which the open abdomen is cov-
ered with mesh [52]. One'such technique is an improved zipper closure of the
abdominal wall called the Ethi-Zip. The third approach, described in detail by
Wittmann, is the STAR abdominostomy in which multiple planned laparotomies
with staged reapproximation and final suture closure of the abdominal fascia is
carried out utilizing a mesh with a zipper or suture, or other such approaches.
The problems with these approaches are that they are used in terribly sick pa-
tients and it is difficult to know whether survival is improved by it or not.
Wittmann et al. have provided for us the management principles of peritoni-
tis which included supportive measures and then operative treatment [52]. Prin-
ciple 1: Repair by control of the source of infection, if this can be done. Princi-
ple 2: Evacuation of bacterial purulence and thorough irrigation. If the septic
process has been controlled and the peritoneal cavity is clean, then that is all
that is necessary to be done and many patients will do very well following this.
Principle 3: With severe suppurative generalized peritonitis, there may be a. need
to decompress and treat the potential abdominal compartment syndrome. Princi-
ple 4: Control is to prevent or treat persistent and recurrent infection in which
other procedures are necessary. Wittmann believes that the indications for a
staged abdominal repair are: 1) a patient in critical condition with hemodynam-
ic, ventilatory instability, 2) excessive peritoneal edema with swelling of the
bowel which could lead to,an abdominal compartment syndrome, 3) massive ab-
dominal wall loss due to infection or gangrene, 4) impossibility to eliminate or
control the source of infection, 5) incomplete debridement of necrotic tissue, 6)
uncertain viability of remaining bowel, and 7) uncontrolled bleeding and the
need for packing.
Wittman and Wallace reviewed 60 publications with a total of 1983 patients
with peritonitis who met their study criteria [52]. They classified them into three
approaches: an open abdominostomy (OPA), with a mortality of 42%, a cov-
ered/mesh abdominostomy, with a mortality of 39%, and a staged abdominal re-
pair abdominostomy (STAR), with a mortality of 28%. Thus, the results tend to
be better with the STAR approach. The complications of OPA are hernias [5]
and a large number of fistulae. With the second approach (covered mesh) her-
nias were frequent but fistulae were less frequent. With the STAR approach,
hernias were infrequent as were fistulae. Thus, they believe that this is a best ap-
proach to go when it is necessary to do this.
Hau et al., in a case control study by the peritonitis group of the Surgical In-
fection Society of Europe, found that there was no significant difference in mor-
tality between patients treated with planned relaparotomy versus relaparotomy
on demand [53]. Postoperative multiple organ failure, as defined by the Goris
score was more frequent in patients undergoing planned relaparotomy. Thus, the
evidence for a major benefit from planned relaparotomy was lacking. A recent
nonrandomized series failed to demonstrate the benefits of planned relaparo-
tomies [54]. The complications were high [55]. However, another prospective
study showed the staged abdominal repair approach to be superior to conven-
tional operative therapy when the patient's mortality risks were compared by
prognostic factors [56]. A 24-hour interval was recommended for the relaparoto-
my and reexploration and irrigation of the peritoneal cavity.
In 1993 Pusaj6, along with Bumaschny et aI., found that the use of an ab-
dominal reoperation predictive index for peritonitis patients, as compared with
just clinical judgment, was most useful in reducing mortality, the time elapsing
between the first operation and relaparotomy, and the length of stay in the inten-
sive care unit [57]. The abdominal reoperation predictive score was made up of
whether or not the procedure was an emergency operation, whether or not there
was respiratory failure, renal failure, ileus from 72 hours after operation, ab-
dominal pain from 48 hours after operation, wound infection, consciousness al-
terations and symptoms appearing from the fourth day of operation. These are
in essence the same variables that would be included in documenting the pro-
gressive development of organ dysfunction and potential multiple organ failure.
We are left with a dilemma. Is a planned relaparotomy and abdominal repair
a futile procedure? Is it necessary? Is it helpful? When is it necessary? It is very
difficult to randomize and come to a clear answer. What techniques are best?
Some advocate a zipper technique [58]. Others strongly advocate a burr-like de-
vice [59]. It seems to me that there are patients who benefit from the planned re-
laparotomy with an open abdomen and intermittent irrigation and cleaning out
the peritoneal cavity and there are those who are jeopardized by that approach.
Andrus et ai. also found the same mortality (38%) in patients treated by planned
reoperation and those treated by expectant therapy - reoperation only if neces-
sary [60]. Kinney and Polk provided an argument against the open treatment of
peritonitis [61]. Now Wickel et ai. have provided evidence that secondary peri-
toneal infection frequently resolved 77/105 patients with recurrent intra-abdom-
inal infection in only 15 other patients [62]. Thus, disease acuity and organ fail-
ure were the causes of mortality. Thus, the disease process may be changing.
Thus, there is no final answer. We must strive to do what is best for each indi-
vidual patient to be sure that we are doing what seems like the logical best ap-
142 A.E. Baue
proach for them at that time. This alone will allow good judgment and better pa-
tient survival.
Conclusion
I have described two indications for operation and/or reoperation which are: 1)
deterioration of remote organ function which may be due to an intra-abdominal
focus; 2) treatment of the abdominal compartment syndrome. There are two rea-
sons to stop operation and plan a reoperation: 1) the abdominal injury-hemor-
rhage, transfusion, hypothermia coagulopathy acidosis cycle; 2) the general sup-
purative peritonitis problem with planned reoperation and an open abdomen.
All of these approaches maybe necessary and/or worthwhile in certain pa-
tients. Our job is to further learn when each should be used and how it is best
used. The risks for the patients with these problems are high but the alternatives
without these daring procedures are worse.
References
1. Baue AE (1990) Chapter 25, the particular problem of peritonitis and increased abdominal
pressure. In: Baue AE (ed) Multiple Organ Failure. Mosby Year Book, St. Louis, pp. 452-470
2. Polk HC Jr, Shields CL (1977) Remote organ failure: a valid sign of occult intra-abdominal
infection. Surgery 81 :310-317
3. Fietsam R Jr, Villalba M, Glover JL, Clark K (1989) Intra-abdominal compartment syndrome
as a complication of ruptured abdominal aortic aneurysm repair. Am Surg 55:396-402
4. Schein M, Wittmann DH, Apraharnian CC, Condon RE (1995) Collective reviews: the ab-
dominal compartment syndrome: the physiological and clinical consequences of elevated in-
tra-abdominal pressure. JAm Coll Surg 180:745-753 '
5. Moore EE (1996) Staged laparotomy for the hypothermia, acidosis, and coagulopathy syn-
drome. Am J Surg 405-410
6. Wittmann DH, Schein M, Condon RE (1996) Management of secondary peritonitis. Ann
Surg 224:10-18
7. Baue AE (1975) Multiple progressive or sequential system failure: a syndrome of the 1970s.
Arch Surg 110:779-781
8. Fry DE et al (1980) Determinants of death in patients with intra-abdominal abscess. Surgery
88:517-523
9. Ferraris VA (1983) Exploratory laparotomy for potential abdominal sepsis in patients with
multiple-organ failure. Arch Surg 118:1130-1133
10. Sinanan M, Maier RV, Carrico CJ (1984) Laparotomy for intra-abdominal sepsis in patients
in an intensive care unit. Arch Surg 119:652-658
11. Hoogewoud HM et al (1986) The role of computerized tomography in fever, septicemia and
multiple system organ failure after laparotomy. Surg Gynecol Obstet 162:539-545
12. Norton LW (1985) Does drainage of intra-abdominal pus reverse multiple organ failure? Am
J Surg 149:347-350
13. Baue AE (1985) Recovery from multiple organ failure. Am J Surg 149:420-421
14. Thorington JM, Schmidt CF (1923) A study of urinary output and blood-pressure changes re-
sulting in experimental ascites. Am J Med Sci 165:880-886
15. Ranniger K, Switz DM (1965) Local obstruction of the inferior vena cava by massive ascites.
J Rad 93:935-939
16. Mullane JF, Gliedman ML (1966) Elevation of the pressure in the abdominal inferior vena
cava as a cause of a hepatorenal syndrome in cirrhosis. Surgery 59: 1135-1146
17. Bradley SE, Bradley GP (1947) The effect of increased intra-abdominal pressure on renal
function in man. J Clin Invest 26:1010-1022
18. Richards WO, Scovill W, Shin B, Reed W (1983) Acute renal failure associated with in-
creased intra-abdominal pressure. Ann Surg 197: 183-187
19. Kron IL, Harman PK, Nolan SP (1984) The measurement of intra-abdominal pressure as a
criterion for abdominal re-exploration. Ann Surg 199:28-30
20. Richardson JD, Trinkle JK (1976) Hemodynamic and respiratory alterations with increased
intra-abdominal pressure. J Surg Res 20:401-404
21. Barnes GE, Laine GA, Giam PY et al (1985) Cardiovascular responses to elevation of intra-
abdominal hydrostatic pressure. Am J Physio 17:R21O-R213
22. Robotham JL, Wise RA, Bromberger-Barnea B (1985) Effects of changes in abdominal pres-
sure on left ventricular performance and regional blood flow. Crit Care Med 13:803-809
23. Harman PK, Kron IL, McLachlan HD (1982) Elevated intra-abdominal pressure and renal
function. Ann Surg 196(5):594-597
24. Diebel LN, Dulchavsky SA, Wilson RF (1992) Effect of increased intra-abdominal pressure
on mesenteric arterial and intestinal mucosal blood flow. J Trauma 33(1):45-48
25. Diebel LN, Wilson RF, Dulchavsky SA, Saxe J (1992) Effect of increased intra-abdominal
pressure on hepatic arterial, portal venous, and hepatic microcirculatory blood flow. J Trauma
33(2):279-282
26. Bloomfield GL, Blocher CR, Fakhry IF (1997) Elevated intra-abdominal pressure increases
plasma renin activity and aldosterone levels. J Trauma 42(6):997-1005
27. Bloomfield GL, Dalton JM, Sugerman HJ et al (1995) Treatment of increasing intracranial
pressure secondary to the acute abdominal compartment syndrome in a patient with com-
bined abdominal and head trauma. J Trauma 39(6):1168-1170
28. Pusajo JF, Bumaschny E, Agurrola A et al (1994) Postoperative intra-abdominal pressure: Its
relation to splanchnic perfusion, sepsis, multiple organ failure and surgical reintervention. Int
Crit Care Dig 13(1):2-4
29. Bendahan J, Coetzee CJ, Papagianopoulos C, Muller R (1995) Abdominal compartment syn-
drome. J Trauma 38(1):152-153
30. Celoria G, Steingrub J, Dawson JA, Teres D (1987) Oliguria from high intra-abdominal pres-.
sure secondary to ovarian mass. Crit Care Med 15(1):78-79
31. Therti TJ, Kelly KM, Gentili DR et al (1987) A simple technique to accurately determine in-
tra-abdominal pressure. Crit Care Med 15(12):1140-1142
32. Rapanos T, Boulanger B, McLean R, Brenneman F (1997) Is clinical examination an accu-
rate indicator of intra-abdominal pressure in ventilated trauma patients? Crit Care Med 25
(1):A133
33. Daly RC, Mucha P Jr, Farnell MB (1989) Abdominal reexploration for increased intra-ab-
dominal pressure and acute oliguric renal failure. Contemp Surg 35: 11-18
34. Howdieshell TR, Yeh KA, Hawkins ML, Cue (1995) Temporary abdominal wall closure in
trauma patients: Indications, technique, and results. World J Surg 19:154-158
35. Sherck J, Oakes D, Seiver A et al (1996) The open abdomen in intensive care: A novel ap-
proach for coverage. Crit Care Med 24:A73
36. Smith PC, Tweddell JS, Bessey PQ (1992) Alternative approaches to abdominal wound clo-
sure in severely injured patients with massive visceral edema. J Trauma 32(1):16-20
37. Sleeman D, Sosa JL, Gonzalez A et al (1995) Reclosure of the open abdomen. J Am ColI
Surg 180:200-204
38. Feliciano DV, Mattox KL, Jordan GL (1981) Intra-abdominal packing for control of hepatic
hemorrhage - a reappraisal. J Trauma 21 :285-290
39. Svoboda JA, Peter ET, Dang CV et al (1982) Severe liver trauma in the face of coagulopathy
- a case for temporary packing and early reexploration. Am J Surg 144:717-721
144 A.E. Baue
40. Rotondo MF, Schwab CW, McGonigal MD et al (1993) "Damage control": An approach for
improved survival in exsanguinating penetrating abdominal injury. J Trauma 35(3):375-383
41. Burch lM, Ortiz VB, Richardson RJ et al (1992) Abbreviated laparotomy and planned reop-
eration for critically injured patients. Ann Surg 215:476-484
42. Hirshberg A, Wall MJ Jr, Mattox KL (1994) Planned reoperation for trauma: A two year ex-
perience with 124 consecutive patients. J Trauma 37(3):365-369
43. Bernabei AF, Levison MA, Bender JS (1992) The effects of hypothermia and injury severity
on blood loss during trauma laparotomy. J Trauma 33(6):835-839
44. Schein M, Saadia R, Decker G (1988) Intraoperative peritoneal lavage. Surg Gynecol Obstet
166:187-195
45. Silenas R et al (1983) Mechanical effectiveness of closed peritoneal irrigation in peritonitis.
Am J Surg 145:371-373
46. Hudspeth AS (1975) Radical surgical debridement in the treatment of advanced generalized
bacterial peritonitis. Arch Surg 110: 1233-1236
47. Polk HC Jr, Fry DE (1980) Radical peritoneal debridement for established peritonitis: the re-
sults of a prospective randomized clinical trial. Ann Surg 192:350-355
48. Steinberg D (1979) On leaving the peritoneal cavity open in acute generalized suppurative
peritonitis. Am J Surg 137:216-220
49. Pichlmayr R et al (1983) Postoperativ Kontinuierliche offene dorso-ventrale Bauchspulung
bei schweren formen der Peritonitis. Der Chirug 54:299-305
50. Losanoff J, Kjossev K (1997) Palisade dorsoventral lavage for neglected peritonitis. Am J
Surg 173:134-135
51. Teichman W, Wittman DH, Andreone PA (1986) Scheduled reoperations (Ettappenlavage) for
diffuse peritonitis. Arch Surg 121:147-152
52. Wittmann DH, Wallace JR The various leaving-the-abdomen-open techniques for advanced
severe peritonitis: Mesh, Zippers, Artificial Burr, and Other Methods (in press)
53. Hau T, Ohmann C, Wolmershauser A et al (1995) Planned relaparotomy vs relaparotomy on
demand in the treatment of intra-abdominal infections. Arch Surg 130: 1193-1197
54. Neijenhuis P, Muckart D (1995) Planned relaparotomy: A futile procedure? Surgical Infec-
tion Society Europe 8th Annual Meeting S28, 26
55. van Goor H, Hulsebos R, Bleichrodt R (1995) Complications of planned relaparotomies in
patients with severe generalized peritonitis. Surgical Infection Society Europe 8th Annual
Meeting S28, 25
56. Wittmann DH, Bansal N, Bergstein JM et al (1994) Staged abdominal repair compares ~avor
ably with conventional operative therapy for intra-abdominal infections when adjusting for
prognostic factors with a logistic model. Theor Surg 25:273-284
57. Pusaj6 JF, Bumaschny E, Doglio GR et al (1993) Postoperative intra-abdominal sepsis re-
quiring reoperation. Arch Surg 128:218-223
58. Mizrahi S, Deutsch M, Hayes D et al (1993) Improved zipper closure of the abdominal wall
in patients requiring multiple intra-abdominal operations. Am J Surg 166:62-63
59. Wittmann DH, Aphraharnian C, Bergstein JM et al (1993) A bun·-like device to facilitate tem-
porary abdominal closure in planned multiple laparotomies. Eur J Surg 159:75-79
60. Andrus C, Doering M, Hemuann VM, Kaminski DL (1986) Planned reoperation for general-
ized intraabdominal infection. Am J Surg 152:682-686
61. Kinney EV, Polk HC (1987) Open treatment of peritonitis: an argument against. Adv Surg 21:
19-28
62. Wickel DJ, Cheadle WG, Mercer-Jones MA, Garrison RN (1997) Poor outcome from peri-
tonitis is caused by disease acuity and organ failure, not recurrent peritoneal infection. Ann
Surg 225:744-756
Multiple Therapeutic Agents - Are We Making
Progress?
A.E. BAUE, H. REDL
Immunotherapy in sepsis
is the Bermuda Triangle
of the Biotech Industry.
Phillip Dellinger, M.D.
Lecture to the Shock Society,
June 1997
It is not unusual to develop therapeutic agents in animal models which seem

protective or therapeutic in the experimental animal but never proven to be
worthwhile in patients. Over the years we have studied many such promising
agents which never made a difference clinically. These include low molecular
weight dextran, which is an anti-sludging agent, Dibenzyline (phenoxybenza-
mine), an alpha adrenergic blocking agent to decrease the intense vasoconstric-
tion of shock, 2-3 diphosphoglycerate, which helps red cells unload oxygen in
the peripheral circulation, polarizing solutions with homeopathic doses of Mg,
K, insulin, glucose, and steroids, white blood cells which led to excess Ringer's
lactate solution being given, buffers for extra-cellular acidosis, THAM
tris(hydroxymethyl)amino-methane, an intracellular buffering agent, excess lac-'
tate and the LIP ratio, which would indicate anaerobiosis and steroids for septic
shock [1-3]. In spite of many positive effects in the experimental laboratory,
none of these substances ever came into clinical use for very long. When they
were subjected to randomized clinical trials, they failed to improve survival or
help patients. Recently, studies of injury, infection and inflammation have
shown many mediators or agents which contribute to illness from such insults.
This led to the development of agents which could block the harmful effects of
such mediators. Many of these agents demonstrated excellent results in experi-
mental animals and suggested great promise for clinical use. However, so far
none of these agents (Table 1), when used individually, has had a positive effect
in decreasing mortality in prospective randomized placebo-controlled trials in
sick patients.
There are many reasons why such trials have been negative (Table 2) [4].
One important reason is that injury, infection and inflammation bring out com-
plex changes and responses in the host. There are multiple pro-inflammatory
mediators with overlap, redundancy and cross-stimulation. This is followed by
146 A.E. Baue, H. Redl
Table 1. Previous clinical trials of agents to control infection or inflammation
No improvement in 28-day mortality with

HA-IA MAb to endotoxin Pentoxifylline
J-5 MAb to endotoxin Polyglobin
Tauralidine Ibuprofen
IL-l ra PAP antagonists
Hydrocortisone in hyperdynamic septic shock Bradykinin antagonist
Anti-TNF MAb Anti-thrombin III concentrate
sTNFr Interferon-gamma
Table 2. Problems with magic bullets
"Attempts to control the proinflammatory response were too crude"

Results would have been better if
- we had intervened at the proper time
- we had not overlooked the antiinflammatory response
- we had restored homeostasis through compensatory antiinflammatory response syndrome (CARS)
- we had had a mediator profile
- we had had a rapid endotoxin assay
- we had had a rapid culture technique
- we had known where the patient was in the inflammatory cascade process
- we had only entered patients at a high risk of dying
- we had been able to resolve the paradoxes and conundrums of many biologic activities
(such as CD 11118 MAb protects the lungs but worsens survival)
- sepsis had not been so heterogeneous
- we had known what we were doing
- we had treated diseases, not intellectual constructs
- the agents had been effective
- we had known that IL-6 was above 1000 picogram or LBP was high or whatever
- we had been able to recognize the heterogeneity in cytokine activities in patients with sepsis
- patient selection had been more appropriate
- we had been able to restore monocyte function before infection developed
- the end-point had not been 28-day mortality
- the end-point had been the reduction of the MODS score
an anti-inflammatory response to try to control the process before it gets out of

hand. The timing and variability of these processes are inconsistent [5, 6]. Even
with similar diseases there is great variability. This has led to consideration of
multiple therapeutic agents for patients with diseases or injuries which stimulate
an inflammatory response.
Multiple therapeutic agents in other diseases

There are many human diseases in which multiple agents are required for appro-
priate therapy. This includes antituberculous therapy for tuberculosis, immuno-
suppression for transplanted organs, ionotropes and diuretics for heart failure,
Multiple Therapeutic Agents - Are We Making Progress? 147
multiple antibiotics for polymicrobial peritonitis, cancer chemotherapy and sup-

port of the gastrointestinal tract. A review of several of these disease processes
will illustrate the difficulties and the evolution that occurred in therapy with the
addition of multiple agents.
The development of chemotherapy for tuberculosis and its evolution over the
years will serve as an example of the problems even when dealing with a specif-
ic disease process and one organism which may be typical, atypical or may
develop resistance to antibiotics. In 1944 Dr. Selman Waksman and colleagues
isolated streptomycin [7]. It was found to be effective against tuberculosis in a
small trial in 1945 [8], and this was followed by a large national trial in 1947,
demonstrating impressive clinical results [9]. It was immediately apparent that
there was a high incidence of relapse and the development of resistant organ-
isms [10]. To counteract this effect, an agent p-aminosalicylicacid (PAS), a drug
which had mild tuberculostatic activity, was used in combination with strepto-
mycin in a trial in 1948-49 [11]. PAS extended the time during which strepto-
mycin could be administered without developing resistance. In 1950 a specific
program by industry led to the development of an antituberculous agent which
was synthesized called isonicotinic acid hydrazide (INH) or isoniazid. This was
found to be very effective in vitro and was strikingly successful in patients in
1952 [12]. This ushered in the modem era of chemotherapy. Other drugs were
then developed. Presently recommended basic treatment for previously untreat-
ed patients with pulmonary tuberculosis in an initial phase are isoniazid,
rifampin, and pyrazinamide given daily for two months followed by four
months of isoniazid and rifampin [13, 14]. Ethambutol can be added in the ini-
tial two months if there is any suspicion of resistance or if the patient is thought
to be HIV infected. Thus, there has been a steady and continuing evolution of
appropriate multiagent chemotherapy for tuberculosis. We are reminded, howev-
er, that tuberculosis is a single disease even though there are variations in the
organism, typical, atypical and resistant, etc., which primarily involve the lungs
initially. Much of the development of successful treatment of tuberculosis was
done by in vitro studies of the organism in culture and then trial and error clini-
cally [15]. Also, each of the agents used in combination was effective for some
time when used singly.
The development of cancer chemotherapy is another example of the com-
plexities and difficulty in treating the manifestations and causes of human dis-
ease. Cancer chemotherapy was initially modeled after the multi-agent treat-
ment of tuberculosis. Paul Ehrlich is said to have coined the word "chemothera-
py" at the tum of the century. He used rodent models of infectious diseases to
develop antibiotics. This led Clowes at Roswell Park Memorial Institute in the
early 1900s to develop inbred rodent lines to carry transplanted tumors to screen
for potential anticancer drugs [16].
The first modem chemotherapeutic agents were a product of a secret war gas
program in both world wars. There was an explosion in Bari Harbor during
World War II. Seamen were exposed to mustard gas which caused bone marrow
and lymphoid suppression [17]. This led to trials in patients with hematopoietic
neoplasms such as Hodgkin's disease and lymphocytic lymphomas. This was

first attempted at the Yale Cancer Center in 1943. Because of the secret nature
of the wartime gas program, these results were not published until 1946 [18].
Initially, there was great excitement because of regression of these neoplasms
but this was followed by discouragement because the tumors always grew back.
This was followed by Farber's observation that folic acid accelerated leukemia
and folic acid antagonists were developed [19]. Early therapy for childhood
leukemias and Hodgkin's disease was with combination chemotherapy. There
was then a long period of trial and error, observation of chemotherapy failure,
and many other problems. DeVita states that, with some exceptions (choriocar-
cinoma and Burkitts lymphoma) single drugs and standard doses do not cure
cancer [20]. In the early years of chemotherapy, drug combinations were devel-
oped based upon known biochemical actions of available anticancer drugs rather
than on their clinical effectiveness. These were largely ineffective. DeVita states
that the era of effective combination chemotherapy began when an array of ac-
tive drugs from different classes became available for use in combination in the
treatment of leukemias and lymphomas [21]. He concludes that, for multiagent
cancer chemotherapy, only drugs known to be partially effective against the
same tumor when used alone should be selected for use in combination. The
least toxic drug should be used and given in an optimal dose and schedule. The
principle of cancer chemotherapy has been clinical trial designed and dominated
by the use of alternating cycles of combination chemotherapy [19]. Of course,
the response to chemotherapy is affected by the biology of tumor growth. Thus,
all cancers are different. They respond to very different agents. What is effective
for one malignancy may do nothing for another. Malignancy is not a common
denominator for therapy. Some tumors are hormone dependent, some respond to
radiation therapy, some respond to chemotherapy and various combinations,
some respond to both, some respond to operation with or without adjuvants.
Staging and grade also have a lot to do with this. It is apparent also now that
cure of malignancy is unusual and the malignant setting in patients is very im-
portant in terms of oncogene influence and genetic mutations and other factors.
Thus, the lessons learned from the treatment of tuberculosis and cancer indi-
cate that specific diseases must be treated by a combination of agents, each of
which has been shown to be individually effective in some way, shape or form.
In addition, these processes of infection and neoplasia are chronic processes,
they are not immediate, acute and life-threatening problems. Treatment can be
carried out over many weeks. Thus, there are many dissimilarities between the
use of multiple chemotherapy for these diseases and the possibility of using
agents for the control of acute inflammation and of SIRS, MODS, and MOE
Experimental studies of multiple agents for inflammation

Therapy for excess inflammation could require control or replenishment of a
number of agents shown in Table 3. Several years ago one of us (H. Redl), along
with G. Schlag, hosted a shock conference in Vienna during which a number of
Table 3.
Therapy for excess inflammation may require control of:

Endotoxin Coagulation activation
Pro-inflammatory cytokines Adhesion molecule activation
Bradykinin Complement activation
Proteinases Cyclo-. and lipo-oxygenase activation
Oxygen radicals Histamine stimulation
Therapy may also require replenishment of:
Antiinflammatory mediators
Antioxidants
Immunostimulators
speakers presented models for the use of multiple agents or mUltiple component
therapy for the treatment of sepsis and septic shock in critically ill surgical
patients (May 7-11, 1995). Aasen and colleagues from Oslo, Norway, gave a
presentation based upon a study in a pig model receiving endotoxin. They used
a combination of three protease inhibitors (Cl inhibitor, antithrombin III and
aprotinin) together with methylprednisolone, naloxone, ketanserin, and promet-
hazine, which was found to counteract endotoxin-induced hyperdynamic
changes [22]. This protected the animal against endotoxin-induced changes in
the plasma enzyme cascade systems. At the same meeting Opal and colleagues
from Brown University used an established infection model of pseudomonas
sepsis and treated the animals with a combination of J-5 antisera, opsonophago-
cytic MAb, and anti-TNF MAb [23]. They found that this provided significantly
greater protection than the single component therapy.
Faist presented a hypothesis for a combined therapeutic strategy which
included 1) a global short-term « 72 hours) downregulation of inflammatory
monocyte activity via drugs like pentoxyfylline and IL-lO or IL-13, 2) the pre-
vention of excessive monocyte stimulation by neutralization of circulating endo-
toxins with high-dose polyvalent immunoglobulins, BPI-bacterial permeability
increasing protein, and soluble complement receptors, and 3) the cell mediated
specific immune performance should be upregulated to overcome post-traumatic
paralysis by administration of substances like thymomimetic hormones, gamma
interferon, and granulocyte-colony stimulating factor [24]. At that conference
Charles Fischer suggested that a combination of agents could be helpful and
should be evaluated [25]. He listed BPIP for anti-endotoxin effects, IL-iRra for
anticytokine effects, antithrombin III to protect against the coagulation cascade,
and a complement inhibitor to decrease the complement cascade.
There have been recent studies which change some of these hypotheses and
proposals. For example, Mannick et al. found that a monoclonal antibody to IL-
10 restored resistance to a septic challenge in an animal model [26]. Dalton et
al. found that combined administration of interleukin-I receptor antagonist (IL-
Ira) and soluble tumor necrosis factor receptor (sTNF-r) decreased mortality
and organ dysfunction in animals after hemorrhagic shock [27].
Demling et aI., in Boston, used a combined chemotherapeutic regime in bum

patients [28]. They gave antioxidants which include vitamins C, E and gluta-
mine, with an endotoxin binder (parenteral polymyxin B), a cyclo and lipoxyge-
nase inhibitor, ibuprofen, and reconstituted human growth hormone. They
believe that this improves mortality but it is based on historical controls.
Civetta and Kirton and colleagues in Miami used a multi agent approach for
patients after trauma [29]. Kilbourn, Szabo, and Traber suggest that a combina-
tion of approaches that treat vasodilatation, multi organ damage, metabolic dys-
function and coagulation abnormalities may be needed to treat septic shock
[30]. They believe that there is still something worthwhile about nitric oxide
synthase inhibitors in circulatory shock, like it was shown in recent nonhuman
primates [31].
Ideal combinations of agents

Because of the many mediators, each of which seems to have a role in the
pathogenesis of excessive inflammation, it makes scientific sense to use multi-
ple agents. If we tried to put together an ideal combination of agents for excess
inflammation, what would be the components? Certainly early on in the disease
process some attempts to block pro-inflammatory mediators (Table 4) should be
Table 4.
Scavenging of inducers
Endotoxin - rBPI21
Pro-inflammatory mediator blockade
lL-lra
sTNFr
anti-TNF Mab - to restore function
Supplementation of anti-inflammatory agents
IL-IO - to reduce inflammation
IL-12, IL-13
Anti IL-IO Mab - to restore immune function
rHDL
Antioxidants
Protease inhibitors
Tissue factor pathway inhibitor
Cascade control
Coagulation - ATIII
Complement inhibitor
Cyclo- and lipo-oxygenase inhibition - ibuprofen
Histamine antagonist
Bradykinin antagonist
Control of other factors
PAF antagonist
Immunomodulators - drugs, diet
Anti-adhesion agents
worthwhile. Soon thereafter, supplemental antiinflammatory mediators would

seem necessary.
Included should be control of the many enzyme cascades which are activated
by shock, trauma or infection (Table 4). How many of these are necessary or
important or possible is not known? How do we begin to formulate such an
approach. What is the timing? What will be the cost? If the multi-agent cocktail
becomes beneficial, what ingredients are critical, some may be ineffective. What
is the model on which to test such approaches? One model is the sheep model
developed by Regel et al. [32]. These models were reviewed by Redl et al. [33].
A baboon model which is in the final development stage could be helpful for
multi-agent testing [31]. Would that fill the bill? Perhaps so, or do we also need
new "multiple models" to cope with a "two or multiple hit" theory as suggested
[34, 35]. In any case, it will be difficult to prepare a sufficient multidimensional
protocol for such a study. We ar,e told that the Food and Drug Administration in
the United States would probably not approve a multi-agent approach. Perhaps a
trial in Europe would help. In the meanwhile, we may learn more from multiple
agents in animal studies.
References
1. Baue AE (1971) The treatment of shock - New concepts and old Shibboleths. In: WT Irvine
(ed) Modern trends in surgery, Butterworths, London 1: 1-13
2. Baue AE (1972) Facts, fads and fallacies of activism and apathy. J Trauma 12: 179 (Editorial)
3. Baue AE (1996) Casualty of disease - The problems of animal studies and clinical results. In:
E Faist, AE Baue, FW Schildberg (eds) The immune consequences of trauma, shock and sep-
sis II-I :612-622
4. BaueAE (1997) MOF, MODS, and SIRS - Why no magic bullets? ArchSurg 132:1-5
5. Friedland JS, Porter JC, Daryanani Set al (1996) Plasma proinfiammatory cytokine concentra-
tions, acute physiology and chronic health evaluation (APACHE III) scores and survival in
patients in an intensive care unit. Crit Care Med 24: 1775-781
6. Damas P, Carnivet JL, DeGroote D et al (1997) Sepsis and serum cytokine measurements. Crit
Care Med 25:405-412
7. Schatz A, Bugie E, Waksman SA (1944) Streptomycin, a substance exhibiting antibiotic activ-
ity against gram-positive and gram-negative bacteria. Proc Soc Exper BioI and Med 4:66-69
8. Hinshaw HC, Feldman WH (1945) Streptomycin in treatment of clinical tuberculosis: A pre-
liminary report. Proc Staff Meet, Mayo Clinic, 20:313-318
9. Hinshaw HC, Pyle MM, Feldman WH (1947) Streptomycin in tuberculosis. Am Jr Med 2:
429-435
10. McDermott W, Muschenheim C, Hadley SJ et al (1947) Streptomycin in the treatment of
tuberculosis in humans. Ann Intern Med 27:769-822
11. Medical Research Council (1950) Treatment of pulmonary tuberculosis with streptomycin and
para-aminosalicylic acid. Br J Med 2: 1073-1085
12. Robitzek EH, Selikoff IJ (1952) Hydrazine derivative of isonicotinic acid (Rimifon, Marsalid)
in the treatment of acute progressive caseous-pneumonic tuberculosis. A preliminary report.
Am Rev Tuberc 65:402-428
13. Committee on Treatment, International Union Against Tuberculosis and Lung Disease (1988)
Antituberculosis regimens of chemotherapy. Bull Int Union Tuberc Lung Dis 63:60-64
14. Tuberculosis Unit, Division of communicable Diseases, World Health Organization (1991)
Guidelines for tuberculosis treatment in adults and children in national tuberculosis programs.
World Health Organization. WHOffB 91:161
15. MacGregor RR (1993) Treatment of mycobacterial disease of the lungs caused by mycobac-
terium tuberculosis. In: A Fislunan (ed) Pulmonary diseases and disorders. McGraw Hill, New
York, chapt. 118, pp 1869-1882
16. Marchall EK Jr (1964) Historical perspectives in chemotherapy. In: Goldin A, Hawking IF
(eds) Advances in chemotherapy, vol. 1. Academic Press, New York, pi
17. Alexander SF (1944) Final report of Bari mustard casualties. Allied Force Headquarters,
Office of the Surgeon. APO 512, June 20
18. DeVita VT (1978) The evolution of therapeutic research in cancer. N Engl J Med 298:807
19. DeVita VT Jr (1997) Principles of cancer management: Chemotherapy. In: DeVita VT Jr, Hell-
man S, Rosenberg SA (eds) Cancer principles & practice of oncology, 5th edn, vol 1, Lippin-
cott-Raven, Philadelphia, pp 333-339
20. DeVita VT Jr (1978) The evolution of therapeutic research in cancer. Sounding Boards, 298;
16:907-910
21. DeVita VT Jr, Schein PS (1973) Medical progress - The use of drugs in combination for the
treatment of cancer, rationale 'and results. N Engl J Med 288:998-1006
22. AasenAO, Naess E, Carlsen H et al (1995) Shock 3[SuppIJ:65 (abstr)
23. Opal S, Cross AS, Sadoff JC, Fisher CJ Jr (1995) Combined immunotherapy in the treatment
of septic shock. Shock 3 [Suppl]:65 (abstr)
24. Faist E (1995) Immunomodulatory approaches in critically ill surgical patients. Shock 3
[Suppl]:65-66 (abstr)
25. Fischer C (1995) Unpublished discussion, Fifth Vienna Shock Forum, May 7-11
26. Mannick JA, Lyons A, Kelly J et al. Major injury induces increased production of IL-1O
by cells of the immune system with a negative impact on resistance to infection. Ann Surg
(in press)
27. Dalton JM, Gore DC, DeMaria EJ et al. Combined administration of interleukin-l receptor
antagonist (IL-IRA) and soluble tumor necrosis factor receptor (STNF-R) decreases mortality
and organ dysfunction following hemorrhagic shock. J Trauma (in press)
28. Knox J, Demling R, Wilmore D et al (1995) Increased survival after major thermal injury: The
effect of growth hormone therapy in adults. J Trauma 39:526-530
29. Kirton 0, Windsor J, Civetta et al (1996) Persistent uncorrected intramucosal pH in the criti-
cally injured: The impact of splanchnic and antioxidant therapy. Crit Care Med 24:A82 (abstr)
30. Kilbourn RG, Szabo S, Traber DL (1997) Beneficial versus detrimental effects of nitric oxide
synthase inhibitors in circulatory shock: Lessons learned from experimental and clinical stud-
ies. Shock, 7:235-246
31. Dwenger A, Remmers D, Gratz M et al (1996) A protinin prevents the development of the trau-
ma-induced multiple organ failure in a chronic sheep model. Eur J Clin Chern Clin Biochem
30:207-214
32. Dwenger A, Remmers D, Gratz M et al (1996) A protinin pJ;'events the development of the
trauma-induced multiple organ failure in a chronic sheep model. Eur J Clin Chern Clin
Biochem 30:207-214
33. Redl H, Schlag G, Bahrami S, Yao YM (1996) Animal models as the basis of pharmacologic
intervention in trauma and sepsis patients. World J Surg 20:487-492
34. Moore E, Moore F, Franciose R et al (1994) The postischemic gut serves as a priming bed for
circulating neutrophils that provoke multiple organ failure. J Trauma 37:881
35. Ayala A, Perrin MM, Wagner MA, Chaudry IH (1990) Enhanced susceptibility to sepsis after
simple hemorrhage. Depression of Fe and C3b receptor mediated phagocytosis. Arch Surg
125:70-75
I NEUROMUSCULAR DYSFUNCTION I
SIRS and Sepsis-Induced Neuromuscular
Dysfunctions
C.P. BOLTON
Sepsis [1] has been increasingly associated with a severe systemic response to
infection, usually resulting in early death. However, sepsis may be evoked in the
absence of infection (e.g. trauma and burns), thus the term "systemic inflamma-
tory response syndrome" or SIRS [2] is now used. Considering the profound
changes induced by SIRS, it is not surprising that the nervous system is affected
(Fig. O. The chief manifestations are septic encephalopathy and critical illness
polyneuropathy, each occurring in approximately 70% of septic patients [3].
Investigating neuromuscular conditions in the Intensive Care Unit

(ICU)
One should first exclude conditions which may not have SIRS as an underlying
factor and begin before admission to the critical care unit. Acute infective, trau-
matic or neoplastic spinal cord compression, Guillain-Barre syndrome, myas-
thenia gravis, muscular dystrophy, and so forth, are usually obvious before
endotracheal intubation and placement on a ventilator. However, occasionally,
the conditions worsen so rapidly that an early diagnosis is not possible. These
conditions must then be investigated while the patient is in the critical care unit.
Then consider patients who have been admitted to a critical care unit for a vari-
ety of severe, primary illnesses or injury, develop SIRS, are observed to have
difficulty in weaning from the ventilator and limb weakness. An underlying neu-
romuscular condition can be suspected if, after lung or cardiac causes of respira-
tory insufficiency have been eliminated, on attempted weaning, voluntary respi-
rations are rapid and weak and accompanied by a rising blood CO2, Neurologi-
cal signs of neuropathy or myopathy mayor may not be present. The main fea-
tures of these conditions are summarized in Table 1.
Involvement of the high cervical spinal cord, peripheral nerves, neuromuscu-
lar junctions and muscles should be systematically investigated. It may be nec-
essary to perform one or more of MRI scanning of the cervical spinal cord;
motor and sensory nerve conduction; repetitive electromyography of muscle,
tests of the respiratory system by phrenic nerve conduction studies and needle
156 C.P. Bolton
Table 1. Neuromuscular conditions in the critical care unit associated with SIRS
Creatine
Clinical Electro- Muscle
Conditions Incidence myography Phospho- biopsy
features
kinase
A. Polyneuropathy
Critical Common Flaccid Axonal Near Denervation
illness limbs, and degeneration normal atrophy
polyneuropathy respiratory of motor
weakness and sensory
fibres
Motor Common with Flaccid Axonal Near Denervation
neuropathy neuromuscular limbs, and degeneration normal atrophy
blocking respiratory of motor
agents weakness fibres
B. Neuromuscular Transmission Defect
Transient Common with Flaccid Abnormal Normal Normal
neuromuscular neuromuscular limbs, and repetitive
blockade blocking respiratory nerve
agents weakness stimulation
studies
C. Myopathy
Thick filament Common with Flaccid Abnormal Elevated Central loss
myopathy steroids, limbs, and spontaneous of thick
neuromuscular respiratory activity filaments
blocking weakness
agents and
asthma
Disuse Common(?) Muscle Normal Normal Normal or
(cachectic wasting type 2 fibre
myopathy) atrophy
Necrotizing Rare Flaccid Abnormal Markedly Panfascicular
myopathy of weakness, spontaneous elevated muscle fibre
intensive care Myoglo- activity necrosIs
binuria in muscle
(With pennission from [4])
electromyography of the diaphragm; measurements of serum creatine phospho-

kinase; and biopsy of muscle [3-5, 6].
Critical illness polyneuropathy

After the development of SIRS the earliest nervous system manifestation is sep-
tic encephalopathy. Within hours of the appearance of a positive blood culture,
careful testing may reveal impaired attention, concentration, orientation and
writing [7]. If the SIRS continues, the patient gradually slips into deep coma,
SIRS and Sepsis-Induced Neuromuscular Dysfunctions 157
Infection Trauma
Bacteria. Viruses. Fungi Physical. Surgical. Bums. Chemicals
1
Sepsis
Systemic Inflammatory Response Syndrome (SIRS)
Single or Muttiple Neuromuscular Blocking Agents
1
Organ Failure
Steroids
Septic
Encephalopathy
IMyoLv'
Fig. 1. The various factors associated with the development of the systemic inflammatory response
syndrome (SIRS) and its nervous system complications. (Adapted with permission from [4])
usually without the development of focal signs, seizures, myoclonus or asterixis.

The electroencephalogram (EEG) is a sensitive indicator of the presence and
severity of septic encephalopathy. Head scans and cerebrospinal fluid examina-
tions are usually unremarkable [8].
If SIRS can be treated by antibiotics, surgical drainage of an infected focus,
inotropic drugs and fluid replacement, the encephalopathy usually improves
rapidly but, at this time, it will be noticed that there is a difficulty in weaning
from the ventilator. Studies in our unit indicate that the commonest neuromus-
cular cause for this, after cardiac and pulmonary causes have been excluded, is
critical illness polyneuropathy [9-11]. However, clinical signs of neuropathy,
including depressed deep tendon reflexes, are present in only half of these
patients. Hence, electrophysiological studies are necessary to establish the diag-
158 C.F. Bolton
nosis. A more severe polyneuropathy can, however, be suspected when, on deep

painful stimulation of the distal extremities, it will be noted that limb move-
ments seem clearly weak despite strong grimacing of facial musculature.
In the ICU, central respiratory drive may be assess by temporarily decreasing
ventilatory support to 5 to 8 cm H 20 of pressure support or continuous positive
airway pressure (to overcome airway/ventilator resistance), for a maximum of
15 minutes. Mechanical ventilation is restored if there is evidence of respiratory
distress, arterial oxygen saturation < 90% (based on a pulse oximeter reading)
or a significant rise in heart rate or blood pressure. We found it is advantageous
to do this at the same time that we perfonn needle electromyography of the
diaphragm. The presence and types of disorders of central drive can be more
accurately assessed.
The earliest electrophysiological sign is a reduction of compound muscle
action potential amplitudes, with minor change in latency. This is typical of
axonal damage and occurs with one week. Fibrillation potentials and positive
sharp waves may not appear in muscle until three weeks. Motor unit potentials,
if they can be voluntarily activated by the patient (and may not be due to seda-
tion or septic encephalopathy) will often appear nonnal or somewhat low ampli-
tude and polyphasic, suggesting an associated primary involvement of muscle
by sepsis. These electrophysiological changes could also be due to a primary
myopathy. Hence, it is important to demonstrate depression of sensory com-
pound action potential amplitudes before a finn electrophysiological diagnosis
of polyneuropathy can be made. Repetitive nerve stimulation studies to demon-
strate a defect in neuromuscular transmission should also be perfonned. We
have shown that this does not occur in sepsis but will be present if neuromuscu-
lar blocking agents have been used. Their effects may persist beyond several
hours, to a number of days if the patient is in renal or liver failure [12]. It is also
important to do phrenic nerve conduction studies and needle electromyography
of the chest wall and diaphragm, to establish that the difficulty in weaning from
the ventilator is, in fact, due to critical illness polyneuropathy [6].
In CIP [9] there is a primary axonal degeneration of peripheral nerve motor
and sensory fibers, but no evidence of inflammation, as may be seen in Guillain-
Barre syndrome. Muscle shows scattered atrophic fibers in acute denervation,
and grouped atrophy in chronic denervation. There are occasional necrotic mus-
cle fibers, suggesting an associated primary myopathy. The only central nervous
system manifestation is central chromatolysis of anterior hom cells and loss of
dorsal root ganglion cells, secondary to the peripheral nerve axonal damage. No
changes appear distinctive of critical illness polyneuropathy.
Knowledge of the presence of critical illness polyneuropathy aids manage-
ment on the ventilator and, in particular, indicates that the patient has a neuro-
muscular problem, which may prolong care in the critical care unit. If it is a
mild polyneuropathy, recovery is expected to occur within a matter of weeks,
but if it is severe it may take months. In physiotherapy a rehabilitation tailored
to polyneuropathy should be instituted. Critical illness polyneuropathy is associ-

ated with increased mortality and rehabilitation problems [13]. A few instances
of critical illness polyneuropathy in children are now being observed [14-16].
To date, there has been no specific treatment for CIP. An open trial of high
dose intravenous immunoglobulins was not effective [17].
Other neuromuscular complications of SIRS

Acute motor neuropathy associated with competitive neuromuscular
blocking agents
In this condition [18-24], the patient will have been in the ICU for several days,
or possibly weeks, and competitive neuromuscular blocking agents, such as
pancuronium bromide, or the shorter-acting vecuronium, will have been given
to ease mechanical ventilation. These will have been used for longer than 48
hours, occasionally days or weeks. When these agents are discontinued, difficul-
ty weaning from the ventilator and limb paralysis are noted. The serum creatine
phosphokinase is mild or moderately elevated. Electrophysiological testing may
or may not reveal a defect in neuromuscular transmission. If present, it will be
demonstrated on slower rates of stimulation, to be expected in a post-synaptic
defect. There is evidence of a severe, primary axonal degeneration, of predomi-
nantly motor fibers, on nerve conduction and needle electromyographic studies.
Muscle biopsy shows varying degrees of denervation atrophy and muscle necro-
sis. I believe that SIRS is an important underlying factor in most, if not all, of
these patients [25]. If the various systemic complications can be treated success-
fully, the neuromuscular condition, itself, improves spontaneously and good
recovery may occur, sometimes quite rapidly. However, the neuromuscular
blocking agent likely has an additional toxic effect on nerve and muscle, and its
use should be avoided, if possible.
Transient neuromuscular blockage

Competitive neuromuscular blocking agents are metabolized or cleared by the
liver and kidney. Hence, in the presence of failure of these organs, the effect of
the neuromuscular blocking agent may be quite prolonged, for a number of
days, after it has been discontinued [12]. Repetitive nerve stimulation will cor-
rectly identify the defect in neuromuscular transmission. However, by the time
of testing, many of these patients will already have developed an underlying
critical illness polyneuropathy, in addition to a neuromuscular transmission
defect, each disclosed by electrophysiological studies. Here, recovery will not
occur in a short period of time but may be prolonged for several weeks, or even
months in severe cases.
160 C.F. Bolton
Thick filament myopathy

A distinctive syndrome [13,22,26-28] occurs in children or adults in the setting
of sudden, severe asthma or in the post-transplant state. Endotracheal intubation
and placement on a ventilator is necessary. High dose steroids and neuromuscu-
lar blocking agents to ease ventilation are given, often for a number of days.
Again, on attempted weaning from the ventilator, it will be noted that the patient
has severe neuromuscular respiratory insufficiency and limb weakness. Ophthal-
moplegia may be present [29]. Creatine kinase levels are often considerably ele-
vated. Repetitive nerve stimulation studies are usually normal. Sensory conduc-
tion is normal, as is motor conduction, except for a low amplitude compound
muscle action potential. On needle electromyography, motor unit potentials tend
to be low amplitude, short duration and polyphasic, indicating a primary
myopathy. Muscle biopsy shows a loss of structure centrally in muscle fibers.
Under the electron microscope, this has been shown to be due to destruction of
the thick myosin filaments [27]. Denervation of muscle, secondary to either crit-
ical illness polyneuropathy or the neuromuscular blocking agent, likely predis-
poses to this distinctive pathological change [25, 30]. Recovery occurs quite
rapidly. The clinical and electrophysiological features are usually so distinctive
in this syndrome that muscle biopsy is often not necessary, a worthwhile consid-
eration in children because of the disfiguring scar. Thus, neuromuscular block-
ing agents should be used to ease ventilation only when there are clear-cut indi-
cations, and in as Iowa dosage and for as short a period as possible.
Cachectic myopathy
Cachectic myopathy, disuse atrophy and catabolic myopathy [31] are often cited
as complications of critical illness. However, even though they cause muscle
weakness and wasting, all are ill-defined in clinical terms [25]. Motor and sen-
sory nerve conduction studies, needle electromyography of muscle and creatine
kinase levels are all normal. Muscle biopsy may be normal or show Type 2 mus-
cle fibre atrophy, a nonspecific finding.
Acute necrotizing myopathy of intensive care

It may be precipitated by a wide variety of infective, chemical, and other insults,
basically involving the differential diagnosis of acute myoglobinuria [32]. It
would be expected to occur with increased frequency in critical care units, in
which there is a high incidence of trauma, infection and the use of various med-
ications. There is severe weakness with high levels of CPK and often myoglo-
binuria. Electrophysiological studies were consistent with a severe myopathy,
and muscle biopsy showed widespread necrosis of muscle fibers. Rapid and
spontaneous recovery is expected to occur in milder cases, but in more severe
cases the prognosis may be poor [32-35].
Recent studies on myopathy in critical illness have been quite interesting.

Latronico et al. [36] observed on combined biopsy of superficial peroneal nerve
and peroneal brevis muscle evidence of muscle pathology, and in only 8 was it
clearly that of denervation atrophy. The nerve biopsy indicated an axonal neu-
ropathy in 8 of 22 patients. However, these biopsies were performed very early
in the course of the critical illness perhaps before structural changes had time to
take place, and electrophysiological studies clearly pointed to an axonal motor
and sensory polyneuropathy in most of these patients. Rich et al. [37] in
Philadelphia developed and applied the technique of direct muscle stimulation
to 14 critically ill patients. In eleven the amplitude ratio of nerve/muscle stimu-
lation indicated a primary myopathy and in five a neuropathy.
Lacomis et al. [38], studied 14 critically ill patients prospectively. They were
either post transplant, or being treated for severe pulmonary disorders and sep-
sis. They all received neuromuscular blocking agents and steroids. A severe
necrotic myopathy resulted in 79% in which there was a selective loss of thick
myosin filaments. Creatine phosphokinase values were normal or only mildly
elevated. However, sensory conduction studies were abnormal in approximately
50%, indicating there was likely an associated critical illness polyneuropathy in
a significant number of these patients. Recovery from the myopathy occurred
within 3-4 months in patients who survived. However, in recent prospective
studies by Berek et al. [39], and Leijten and De Weerd [40], there was no corre-
lation with the use of these agents and neuromuscular disease. The results in the
present study were similar. Thus, the precise role of NM blocking agents and
steroids in causing a critical illness myopathy is still in doubt.
Pathophysiology
Retrospective [9] and prospective [41] studies have failed to incriminate a vari-
ety of potential causes of critical illness polyneuropathy, including types of pri-
mary illness or injury, Guillain-Barre syndrome, medications including amino
glycoside antibiotics and neuromuscular blocking agents and specific nutritional
deficiencies. We have speculated that sepsis is the cause [9,41].
The micro circulation is disturbed in sepsis (Fig. 2). Blood vessels supplying
peripheral nerve lack autoregulation [42]. Cytokines that are secreted in sepsis
have histamine-like properties which may increase micro vascular permeability
[9]. The resulting endoneurial edema could induce hypoxia. Severe energy
deficits would result and induce a primary axonal degeneration, most likely dis-
tally, if highly energy-dependent systems involving axonal transport of structur-
al proteins are involved. The predominantly distal involvement may explain why
recovery in some patients may be surprisingly short, conforming to the short
length of nerve through which axonal regeneration takes place.
Through increased capillary permeability induced by the sepsis, neuromus-
cular blocking agents, notably vecuronium or its metabolite, 3 desacetyl-vecuro-
162 c.F. Bolton
Denervatlon
Atrophy of
Muscle
Fig. 2. Theoretical mechanisms of neuromuscular complications of SIRS. 1. Sepsis induces a

release of cytokines which cause increased capillary permeability. This, and other microvascular
mechanisms, induce a critical illness polyneuropathy, with distal axonal degeneration of nerve and
denervation atrophy of muscle. 2.Neuromuscular (N-M) blocking agents in the presence of SIRS
traverse the hyperpermeable capillary membrane and have a direct toxic effect on nerve, or cause
"functional denervation" to increase denervation of muscle. 3. Steroids gain access to muscle by
this mechanism and, in the presence of denervation due to 1 and 2, induce a thick filament myopa-
thy and varying degrees of necrosis. Combinations of 1, 2, and 3 may occur in the same patient.
(Adapted with permission from [3])
nium [12], could have a direct toxic effect on peripheral nerve axons. These
neuromuscular blocking agents may also cause functional denervation through
their prolonged neuromuscular blocking action [43]. Hund et al. [44] have iden-
tified in the serum of CIP patients a factor which is toxic to cultured neurons
from fetal rat spinal cord.
In the acute myopathy which develops when patients are trea~ed with neuro-
muscular blocking agents and steroids, we suspect infection is often a precipitat-
ing event. Animal experiments by Karpati et al. [30] have shown that if the mus-
cle is first denervated by nerve transection and then steroids are given, a thick
filament myopathy similar to that seen in humans can be induced. Thus, in the
human condition, critical illness polyneuropathy and the additional effects of
neuromuscular blocking agents would denervate muscle and then steroids would
induce the typical myopathic changes. Corticosteroids may activate an ATP -
ubiquitin dependent proteolytic system [45], or proteolysis may be initiated
through calpain expression, which alters calcium homeostasis [46] .
References
1. Critchley M (ed) (1978) Butterworths Medical Dictionary. 2nd ed. Butterworths, Boston: 1415
2. American College of Chest Physicians/Society of Critical Care Medicine Consensus Confer-
therapies in sepsis. Crit Care Med 20;6:864-874
3. Bolton CF (1993) Neuromuscular complications of sepsis. Intens Care Med 19:558-S63
4. Bolton CF (1996) Sepsis and the systemic inflammatory response syndrome: Neuromuscular
manifestations. Crit Care Med 24: 1408-1416
5. Bolton CF (1987) Electrophysiologic studies of critically ill patients. Muscle & Nerve 10:
129-135
6. Bolton CF (1993) Clinical neurophysiology ofthe respiratory system. AAEM Minimonograph
#40. Muscle & Nerve 16:809-818
7. Young GB, Bolton CF, Austin TW et al (1990) The encephalopathy associated with septic ill-
ness. Clin Invest Med 13:297-304
8. Young GB, Bolton CF, Austin TW et al (1991) The electroencephalogram in sepsis-associated
encephalopathy. J Clin Neurophysiol 9: 145-152
9. Zochodne DW, Bolton CF, Wells GA et al (1987) Critical illness polyneuropathy: A complica-
tion of sepsis and multiple organ failure. Brain 110:819-842
10. Maher J, Rutledge F, Remtulla H et al (1993) Neurophysiological assessment of failure to
wean from a ventilator. Can J Neurol Sci 20:S28
11. Maher J, Rutledge F, Remtulla H et al (1995) Neuromuscular disorders associated with failure
to wean from the ventilator. Intens Care Med 21:737-743
12. Segredo V, Caldwell JE, Matthay MA et al (1992) Persistent paralysis in critically ill patients
after long-term administration of vecuronium. New Engl J Med 327:524-528
13. Douglass JA, Tuxen DV, Home M et al (1991) Myopathy in severe asthma. Amer Rev Respir
Dis 146:517-519
14. Sheth RD, Pryse-Philips WEM (1994) Post-ventilatory quadriplegia: Critical illness polyneu-
ropathy in childhood. Neurology 44:2:AI69
15. Dimachkie M, Austin SG, Slopis JM et al (1994) Critical illness polyneuropathy in childhood.
J Child Neurol 9:207
16. Bolton CF (1997) EMG in the paediatric critical care unit. In: Jones HR, Bolton CF, Harper M
(eds) Paediatric electromyography. Raven Press (in press)
17. Widjicks EFM, Fulgham JR (1994) Failure of high dose intravenous immunoglobulins to alter
the clinical course of critical illness polyneuropathy. Muscle & Nerve 17: 1494-1495
18. Subramony SH, Carpenter DE, Raju S et al (1991) Myopathy and prolonged neuromuscular
blockage after lung transplant. Crit Care Med 19; 12: 1580-1582
19. Rossiter A, Souney PF, McGowan Set al (1991) Pancuronium-induced prolonged neuromus-
cular blockade. Crit Care Med 12: 1583-1587
20. Gooch JL, Suchyta MR, Balbierz JM (1991) Prolonged paralysis after treatment with neuro-
muscular junction blocking agents. Crit Care Med 9: 1125-1131
21. KupferY, Namba T, Kaldawi E et al (1991) Prolonged weakness after long-term infusion of
vecuronium-bromide. Ann Intern Med 117 :484-486
22. Barohn RJ, Jackson CE, Rogers SJ et al (1994) Prolonged paralysis due to nondepolarizing
neuromuscular blocking agents and corticosteroids. Muscle & Nerve 17:647-654
23. Giostra E, Magistris Mri, Pizzolato G et al (1994) Neuromuscular disorder in intensive care
unit patients treated with pancuronium bromide. Chest 106; 1:210-220
24. Op de Coul AAW, Verheul GAM, Ley ten ACM et al (1991) Critical illness polyneuromyopa-
thy after artificial respiration. Clin Neurol Neurosurg 93:27-33
25. Bolton CF (1994) Muscle weakness and difficulty in weaning from the ventilator in the critical
care unit (editorial). Chest 106:1-2
26. Lacomis D, Smith TW, Chad DA (1993) Acute myopathy and neuropathy in status asthmati-
cus: Case report and literature review. Muscle & Nerve 16:84-90
164 C.F. Bolton
27. Danon MJ, Carpenter S (1991) Myopathy and thick filament (myosin) loss following pro-
longed paralysis with vecuronium during steroid treatment. Muscle & Nerve 14:1131-1139
28. Hirano M, Ott BR, Raps EC (1992) Acute quadriplegic myopathy: complication of treatment
with steroids, nondepolarizing blocking agents, or both. Neurology 42:2082-2087
29. Sitwell LD, Weinshenker BG, Monpetit Vet al (1991) Complete ophthalmoplegia as a compli-
cation of acute corticosteroid and pancuronium-associated myopathy. Neurology 41:921-922
30. Karpati G, Carpenter S, Eisen AA (1972) Experimental core-like lesions and nemaline rods: A
correlative morphological and physiological study. Arch Neuro127:247-266
31. Clowes GHA, George BC, Villee CA (1983) Muscle proteolysis induced by a circulating pep-
tide in patients with sepsis or trauma. New Engl J Med 308:545-552
32. Penn AS (1986) Myoglobinuria. In: Engel AG, Banker BQ (eds) Myology. Vol 2. McGraw-
Hill 1792
33. Ramsay DA, Zochodne DW, Robertson DM et al (1993) A syndrome of acute severe muscle
necrosis in intensive care unit patients. J Neuropath Exper Neuro152:4:387-398
34. Zochodne DW, Ramsay DA, Saly V et al (1994) Acute necrotizing myopathy of intensive care:
electrophysiological studies. Muscle & Nerve 17:285-292
35. Helliwell TR, Coakley JR, Wagemnakers AJM et al (1991) Necrotizing myopathy in critical-
ly-ill patients. J PathoI164:307-314
36. Latronico N, Flavio F, Daniela R et al (1996) Critical illness myopathy and neuropathy. Lancet
347: 1579-1582
37. Rich MM, Bird SJ, Raps EC et al (1997) Direct muscle stimulation in acute quadriplegic
myopathy. Muscle & Nerve 20:665-673
38. Lacomis D, Giuliani M, Van Cott A et al (1996) Acute myopathy of intensive care: clinical,
electromyographic, and pathological aspects. Am Neurol Association 40:645-654
39. Berek K, Margreiter J, Willeit J et al (1996) Polyneuropathies in critically ill patients: A
prospective evaluation. Intensive Care Med 22:849-855
40. Leijten FSS, De Weerd AW et al (1996) Critical illness polyneuropathy in multiple organ dys-
function syndrome and weaning from the ventilator. Intensive Care Med 22: 856-861
41. Witt NJ, Zochodne DW, Bolton CF et al (1991) Peripheral nerve function in sepsis and multi-
ple organ failure. Chest 99: 176-184
42. Low PA, Tuck RR, Takeuchi M (1987) Nerve microenvironment in diabetic neuropathy. In:
Dyck PJ, Thomas PK, Asbury AK et al (eds) Diabetic neuropathy. WB Saunders 268-277
43. Wemig A, Pecot-Dechavassine M, Stover H (1980) Sprouting and regression of the nerve at
the frog neuromuscular junction in normal conditions and after prolonged paralysi~ with
curare. J NeurocytoI9:277-303
44. Hund E, Wolfgang F, Drieger D et al (1996) Critical illness polyneuropathy: Clinical findings
and outcomes of a frequent cause of neuromuscular weaning failure. Crit Care Med Vol 24,
N08
45. Minetti C, Hirano M, Morreale G (1996) Ubiquitin expression in acute steroid myopathy with
loss of myosin thick filaments. Muscle & Nerve 19:94-96
46. Showalter C, Engel A (1997) Acute quadriplegic myopathy: Analysis of myosin isoforms and
evidence for calpain-mediated proteolysis. Muscle & Nerve 20:316-322
PREVENTION AND MANAGEMENT
OF SEPSIS AND MODS
Lights and Shadows in Sepsis and Multiple Organ
Dysfunction Syndrome (MODS)
G. BERLOT, L. SILVESTRI, F. ISCRA, G. SGANGA, A. GULLO
In the last decade, understanding of the basic pathophysiologic mechanism un-

derlying the development of sepsis and septic shock has made impressive
progress. New discoveries have, challenged previously accepted concepts of in-
fection and sepsis, and profound revisions and new therapies have been pro-
posed. This scientific revolution led to the implementation of several large clini-
cal trials employing new agents aimed at preventing or blocking the release of
mediators involved in the development of the systemic inflammatory response
characteristic of sepsis and related disorders, ultimately leading to Multiple Or-
gan Dysfunction Syndrome (MODS). Unfortunately, the overall results indicate
that these novel approaches have not been associated with a substantially im-
proved outcome of septic patients, and in many cases both the design and the re-
sults of these trials have been criticized [1].
On the basis of these considerations, it appears that the clinical fall-out of
these discoveries lies far away from the expectations derived from the more in-
depth understanding of the biology of sepsis, and that, in many cases, areas of
bright light are interrupted by spots of deep shadow. In this chapter, we shall fo-
cus on some of the many controversial points existing in the area of sepsis and
related consequences.
Lights and shadows on definitions

A 4I-year old man was admitted to the hospital with acute abdomen and shock.
The patient required artificial ventilation and cardiovascular pharmacological
support. An explorative laparotomy revealed severe ischemia of the intestinal
loops caused by thrombosis of the superior mesenteric artery. An extensive in-
testinal resection was performed. After the intervention, the patient's conditions
remained unchanged, and were characterized by shock associated with coagula-
tive abnormalities (fibrinogen = 50 mg/dl, platelets = I5.000/ml, PT = 18 secs,
PDP > 70 mcgr/ml, ATIII < 40%). All cultures remained sterile, and blind thera-
py with antibiotics was instituted. Despite the aggressive hemodynamic and res-
piratory support and the repeated transfusions of fresh frozen plasma, the patient
ultimately died. The autopsy revealed diffuse peritoneal inflammation and dif-
168 G. Berlot et aI.
fuse microvascular thromboses associated with the hemorrhage of the adrenal

glands.
The time-honored term septicemia almost disappeared from current medical
literature when the critical care community realized that a more precise defini-
tion of certain clinical settings was urgently needed. Two main considerations
led to the redefinition of these conditions. Firstly, just as patients enrolled in the
clinical trials had to be comparable, also the diagnostic definitions had to be
standardized. Indeed, although every experienced intensive care specialist is
able to recognize, or define, a patient in septic shock, confusion could arise in
the description of patients with less severe, infection-related systemic disorders.
Secondly, both patients with severe infection and those following non-infective
events (e.g. ruptured aortic aneurysms, acute pancreatitis, etc.) may share the
very same symptoms of fever, leukocytosis and hypotension, since the same bi-
ological mechanisms are involved in the pathogenesis of the disturbances. As a
consequence, it appeared useful to separate patients in whom an infection had
triggered the systemic symptoms, from those in whom an infection was unlikely
or, at least, not demonstrated. Thirdly, it might be useful, in a retrospective
analysis, to identify some easily recognizable symptom(s) heralding a worsen-
ing of patients' conditions.
A number of leading authorities in the field of critical care then proposed a
new classification system to be adopted in the diagnosis of patients with severe
systemic disturbances [2]. According to these definitions, it thus became possi-
ble to identify the following clinical entities:
systemic inflammatory response syndrome (SIRS), which is characterized by
two or more of the following conditions: temperature> 38°C or < 36°C, heart
rate > 90 bpm, respiratory rate > 20 breaths/minute or PaC02 < 20 mmHg,
leukocytosis (> 12.000 WBC/ml) or leukopenia « 4.000 WBC/ml), in the ab-
sence of an infection. Acute pancreatitis or the postoperative state following the
repair of a ruptured aortic aneurysm are typical clinical examples of SIRS,
which is supposed to be caused by the production and release of a number of
mediators activated by a non-infective stimulus;
sepsis, which includes the very same alterations described above for SIRS,
but in the presence of an infective focus;
severe sepsis and septic shock, which are characterized by the occurrence of
hypotension and signs of organ malfunction, requiring the infusion of large
amount of fluids or the administration of vasoactive drugs.
mUltiple organ dysfunction syndrome (MODS), characterized by the simulta-
neous dysfunction of multiple organs. This definition offers a more accurate de-
scription of the situation previously indicated as multiple organ failure (MOF),
because the passage from a normal functioning to a totally failing organ occurs
along a continuum of ever-decreasing function and is not an "all-or-nothing"
phenomenon. At the same time, the term "dysfunction" conceptually implies a
certain degree of reversibility, which is almost totally lost in the term "failure".
Lights and Shadows in Sepsis and Multiple Organ Dysfunction Syndrome (MODS) 169
For instance, in many critically ill patients a mild increase of azotemia and cre-
atinemia is common, and their values return to the baseline with a more aggres-
sive volume therapy and/or a judicious use of diuretics. In this particular situa-
tion, the diagnosis of acute renal failure would be inappropriate.
Since these definitions are a compromise deriving from different points of
view, they have not been universally accepted [3, 4]. Heavily criticized is the
unspecificity of the definition of SIRS in that it yields poor prognostic informa-
tion, as its symptoms can be present in patients with disorders associated with
an extremely different outcome. However, if the clinical evolution of patients in
SIRS is taken into account, it appears that in many instances SIRS may be con-
sidered a precursor of sepsis, and a more comprehensive diagnostic approach
should be implemented as soon as this condition is recognized [5]. Actually, in
patients admitted to a surgical critical care unit, Pittet et al. [6] demonstrated
that a) SIRS was almost universally present (93%) in the patients enrolled, and
b) more than half (53%) of these patients subsequently developed sepsis. Fur-
thermore, all patients who developed severe sepsis had previously been diag-
nosed to be in sepsis. The authors concluded that there was a progression from
SIRS to sepsis and severe sepsis, even if the diagnostic criteria of SIRS were so
wide that almost every patient fulfilled them, and thus that the definition of
SIRS was sensitive but not specific. The same group demonstrated that this pro-
gression was also present in a larger group of medical, surgical and respiratory
patients, and that there was an inverse correlation between the number of crite-
ria used to diagnose SIRS and the time interval (in days) between the diagnosis
of SIRS and the onset of sepsis [7]. In patients enrolled in a large Italian multi-
center study, also Salvo et al. [8] observed that prognosis was progressively
worse in patients diagnosed to be in sepsis, severe sepsis and septic shock; how-
ever, these authors failed to demonstrate any prognostic difference between pa-
tients in SIRS and those who did not fulfill its diagnostic criteria, concluding
that SIRS yields little or no diagnostic and/or prognostic information.
Lights and shadows on anti-mediator agents

A 58-year old man was admitted to the leu due to a peritonitis-associated sep-
tic shock which occurred 5 days after a right hemicolectomy for cancer. The pa-
tient was operated again, a purulent collection was drained, and antibiotic treat-
ment was instituted. Despite this approach, the hemodynamic and respiratory
conditions further deteriorated, requiring continuous infusion of norepinephrine
and dobutarnine at ever-increasing dosages. As the entry criteria had been ful-
filled, the patient was enrolled in a double-blind, placebo-controlled, multicen-
tre trial of anti-TNF antibodies. Despite an aggressive treatment which included
repeated percutaneous drainages of abdominal collections, the clinical condi-
tions worsened and, 30 days after the enrolment in the trial, the attending physi-
cians and the family agreed to discontinue the treatment, whereupon the patient
170 G. Berlot et al.
died a few hours later. The pharmaceutical company responsible for the trial
recorded him as a survivor.
The wide array of cardiorespiratory and metabolic derangements commonly
observed in sepsis is caused by the production and the release, by immunocom-
petent and endothelial cells, of a number of mediators resulting from the inter-
action between the infecting agent and the host [9, 10], including the tumour
necrosis factor (TNF), the platelet activating factor (PAF), an ever-increasing
number of Interleukins (IL), the endothelin, arachidonic acid derivatives, and
many others. The same substances are also implicated in the development of
SIRS [11]. As in other biological systems (e.g. the coagulative cascade), the se-
cretion of many, if not all of these mediators is accompanied by the simultane-
ous release of substances with inhibitory properties, teleologically aimed at
down-regulating the inflammatory process [12]. Basically, these agents act by
binding and inactivating the circulating septic mediators or by occupying their
receptors on the surface of the target cells, thus making them unavailable for the
active mediators [12].
Since the late 1980s, greater in-depth knowledge of the basic pathophysio-
logic mechanisms underlying the septic process, and the impressive progress of
genetic engineering techniques have led to the development of many substances
able to blunt the effects of the septic mediators. This goal can be accomplished
in different ways [13]. First, circulating endotoxin and septic mediators can be
inactivated by specific antibodies. Secondly, cell receptors can be selectively
blocked by specific antagonists. Finally, bloodborne mediators can be blocked
by binding them to soluble receptors identical to those present on the cell sur-
face. It is evident that, in many cases, these strategies reflect what is already oc-
curring spontaneously in septic organisms. Unfortunately, despite the sound
pathophysiologic basis and the promising experimental data, the overall results
deriving from many large double-blind, placebo-controlled clinical trials using
different anti-mediator strategies have so far been disappointing or far below the
expectations, since some beneficial effects could only be demonstrated in cer-
tain subgroups of patients [1]. Moreover, some trials had to be suspended when
an interim analysis demonstrated excess mortality in the treatment group [1].
A number of anti-endotoxin antibodies were the first agents to be studied. At
the beginning of the 1980s, Ziegler et al. [14] showed that increased survival
was associated with the administration of human polyclonal antibodies against
the lipid A [15]. These results were confirmed by two other clinical studies
which utilized polyclonal antiendotoxin sera [15, 16]. The major drawbacks of
this treatment consisted in the instability of the solution, in the difficulty of
measuring the real protective effect of a polyclonal preparation, and in the risk
of transmissible diseases [17]. Thanks to the recent impressive development of
genetic techniques, it was possible to overcome these disadvantages. As a re-
sult, ten years later, two large, multicenter, double-blind, placebo-controlled
studies demonstrated a better outcome in septic patients who had been given
antiendotoxin antibodies. The first trial involved 486 patients with suspected
Gram- sepsis who received E5, a murine monoclonal IgM anti lipid A antibody
or placebo [19]. Although the overall mortality rate was comparable in the two
groups, in the subset of patients with confirmed Gram- septic shock, mortality
was slightly higher in the placebo group, whereas it was significantly lower in
treated patients with sepsis but not in shock. During the same period, another
trial was performed using human monoclonal IgM anti endotoxin antibodies
(HA-IA) (given as a single bolus of 100 mg i.v.) [20]. Contrarily to what had
been reported with E5, a better survival rate was observed in patients with con-
firmed Gram- bacteremia and septic shock. In non-bacteriemic, non-shocked
patients with Gram- infection, mortality was higher in the treatment than in the
placebo group. Subsequent data analysis demonstrated that HA-IA was particu-
larly effective in patients with higher blood endotoxin levels [17]. In addition to
these main studies, other investigators demonstrated that in a limited number of
patients the administration of either E5 or HA-IA was well tolerated, that it was
associated with reduced mortality and was not accompanied by major side ef-
fects [21, 22]. However, a more recent randomized, double-blind, placebo-con-
trolled study with E5 failed to demonstrate any effect on survival, although it
was associated with greater resolution of organ failure in the treated group [23].
After its publication, some doubts have been raised about the results of the HA-
lA study [24]. The areas of major concern were the randomization and the con-
comitant treatment (the placebo group was older, had a higher APACHE II
score and a higher rate of inadequate therapy with antibiotics). To test its effec-
tiveness conclusively, another multicenter, double-blind study was started, using
HA-IA. However, interim data analysis revealed slight and non-significant ex-
cess of mortality in non-bacteriemic patients given HA-IA, and the study was
suspended.
Other clinical trials which involved anti-mediator agents shared the same
destiny. As TNF is supposed to be released early in sepsis and exert multiple-
systemic detrimental effects [9, 10], several investigations focussed on the pos-
sible therapeutic effects of its antagonism in septic patients. Experimentally, an-
tibodies directed against TNF exert a protective cardiovascular effect in animal
models of septic shock [25]. Early clinical experiences were encouraging, some-
times indicating a rise in arterial pressure [26] and an improvement in left ven-
tricular function [27]. A preliminary clinical trial involving only 42 patients
demonstrated that the administration of anti-TNF antibodies was free from
harmful side effects and associated with a substantial decrease of circulating
TNF molecules [28]. Unfortunately, two large, double-blind, placebo-controlled
trials with different anti-TNF antibodies failed to demonstrate any effect on the
mortality of septic patients [29, 30]. However, in one of these studies, when
treated patients were retrospectively divided into subgroups, an improved out-
come could be demonstrated in patients with elevated baseline plasma levels of
IL-6 [30]. A confirmatory trial was then launched, in which only septic patients
with plasma IL-6 levels> 1000 pg/ml were enrolled. The study was suspended
by the regulatory authorities when an interim analysis demonstrated that the re-
sults were inconclusive.
As stated above, inactivation of septic mediators can also be accomplished in

other ways, for instance by the administration of agents able to selectively block
the receptors on the cell surface, making them unavailable for the active sub-
stance, or by the interaction with specific circulating receptors which bind the
mediator and so prevent the contact with the target cell. IL-lra has also been
isolated in healthy volunteers given endotoxin [31]. So far, a number of trials
have been performed using the receptor-blocking agent of IL-l (IL-lra), the sol-
uble receptor of TNF and the PAP receptor antagonist. The potentially benefi-
cial effects of these substances have been experimentally demonstrated in many
experimental studies. Rabbits given E. coli intravenously and treated with IL-
Ira show only a transient early hypotension, suggesting that TNF could be re-
sponsible for the early phase of septic shock, whereas IL-l is implicated at a lat-
er stage [32]. Experimentally, the administration of ILl-ra has been associated
with an improvement of cardiovascular function and a reduction of mortality
[33-35]. On the basis of these encouraging results, some clinical trials of IL-lra
in septic patients have been implemented. A randomized, open-label multicenter
Phase II study was performed in the United States. It involved 99 septic patients,
who were given an initial loading dose of ILl-ra or placebo, followed by a 72-
hours continuous infusion of one of three different doses [17, 67 or 133 mg/hr,
respectively) or placebo [36]. The endpoint of the study was survival at 28 days.
In the treatment group, a dose-related increase in survival was observed, mortal-
ity being 44% in the placebo group, 32% in patients receiving the lowest
dosage, 25% in patients receiving the intermediate dosage, and 16% in patients
receiving the highest dosage. At the end of the infusion, the severity of the dis-
ease, as expressed by the APACHE II score, was reduced in the treatment group.
On the basis of these results, a larger, randomized, double-blind, placebo-con-
trolled Phase III trial was launched. The trial involved 893 patient receiving ei-
ther placebo or ILl-ra, which was administered as an intravenous loading ,dose
of 100 mg followed by a 3-day infusion at two different dosages (l or 2
mg/kg/hr, respectively) [37]. Unfortunately, the results of this study failed to
confirm those of the previous one. Actually, no significant reduction in mortality
was observed in the treated patients as compared with the controls (29% vs.
35%, respectively). Some secondary retrospective analyses demonstrated a bet-
ter outcome in patients with dysfunction of one or more organs, and in patients
with a predicted risk of mortality;?: 24% [37, 38]. The results of these studies
were so different that a second Phase III study was then performed, on the as-
sumption that if the trend toward a better outcome could be demonstrated by a
more sophisticated statistical analysis, then the ILl-ra could be clinically valu-
able, at least in selected subgroups of patients. However, this trial had to be sus-
pended when an interim analysis demonstrated the futility of the results [1].
Other investigations focussed on the effects of the TNF soluble receptors. In
septic organisms, the effects of TNF are mediated through 55-kd (TNFR I) and
75-kd (TNFR II) receptors, whose extracellular domain is shed from the cells
during sepsis, and, once released into the interstitial space and in the blood-
stream, they bind the circulating TNF molecules, thus blocking their effects
[39]. Experimentally, the administration of soluble TNF receptors is associated
with the attenuation of the increased pulmonary permeability and the neutrophil
sequestration induced by an intestinal ischemia-reperfusion injury [40]. Howev-
er, as indicated by Van Zee et al. [41], the possible clinical utility of the soluble
TNF receptors could be severely limited by their short half-life (minutes), and
by the fact that the active TNF-a circulates as a trimeric molecule and it is nec-
essary to block at least two of its components to inactivate it. To overcome these
shortcomings, recombinant TNF-a receptors have been linked to the FC and
hinge regions of a human IgG molecule (TNFR:Fc) [42]. Two clinical trials test-
ed the hypothesis that these fusion molecules could be useful in septic patients.
In the first trial, the TNFR II:Fc was administered to patients with septic shock
in three different dosages vs. placebo [43]. Even if a dose-response relation be-
tween treatment and mortality was observed, no difference of mortality could be
demonstrated between the control and the three treatment groups. Moreover, an
excess mortality was observed in patients with Gram+ infections receiving the
highest dosage. Another clinical trial was then performed, using the fusion mol-
ecule TNF I:Fc, again at three different dosages vs. placebo [44]. The study in-
volved 498 patients with severe sepsis and septic shock. The patients treated
with the lowest dosage presented an excess mortality as compared with the oth-
er groups, and this arm of the study had to be discontinued at an interim analy-
sis. Overall, there was a non-significant trend toward reduced 28-day mortality
in all treatment groups. However, a preplanned logistic regression analysis to as-
sess the effects of treatment on 28-day mortality by means of predicted mortali-
ty and plasma IL-6 levels, revealed significantly improved survival in patients
with severe sepsis who received TNF I:Fc at the highest dosage as compared
with the placebo group.
Other investigations involved the antagonism of the PAF, which is a low mol-
ecular weight phospholipid produced by macrophages under the influence of en-
dotoxin [45]. Experimentally, the administration of PAF is associated with the
hemodynamic and metabolic features of septic shock [46], including hypoten-
sion, tachycardia, the increase of microvascular permeability, a negative inotrop-
ic effect and the margination and aggregation of leukocytes and platelets. A
number of natural as well as synthetic PAP inhibitors have so far been identi-
fied. Some of them underwent both experimental and clinical evaluation, shed-
ding some light on the effects of PAP in various target tissues. In rats given en-
dotoxin, the appearance of patchy necrotic bowel lesions is associated with a
TNF-induced increase in PAF secretion [47]; these lesions were prevented by
pretreatment with a specific PAF antagonist [48]. In healthy volunteers treated
with a PAP antagonist 18 hours before the Lv. administration of endotoxin, a re-
duced cardiovascular, metabolic and hormonal response was observed, in the
absence of significant changes of sepsis mediators in the serum [49]. A large
randomized, multicenter placebo-controlled, double-blind clinical trial has re-
cently been performed using the PAF receptor antagonist BN 52021, which was
administered at a dosage of 120 mg every 12 hours for 4 days [50]. There was a
non-significant improvement in the 28-day survival rate in the treatment group
as compared with the placebo group. However, when subgroups of patients were
analyzed separately, a significant reduction of mortality was observed in pa-
tients with Gram- infections, either shocked or not. Conversely, no difference in
the outcome was demonstrated between the placebo and the treatment groups in
the absence of a Gram- sepsis.
Lights and shadows of blood purification techniques

A 34-year old man was admitted to the ICU after emergency surgery for recur-
rent rupture of pulmonary emphysema bullae. An upper right lobectomy was
performed, but the patient was not able to resume spontaneous ventilation due to
the occurrence of respiratory distress. Blood cultures were positive for Serratia
marcescens. Sepsis was further complicated by the occurrence of acute renal
failure (creatinine = 8.5 mg/dl). Hemodialysis was then started, which was poor-
ly tolerated and had to be discontinued, and continuous arterio-venous hemofil-
tration (CAVH) was implemented, allowing the liberal administration of fluids,
medication and total parenteral nutrition. Despite maximal support, the patient
died from MODS after 93 days in the ICU.
There is experimental and clinical evidence suggesting that septic mediators
can to a certain degree be removed by some extracorporeal depuration tech-
niques originally developed for the treatment of acute renal failure - such as
continuous arterio-venous and veno-venous hemofiltration (CAVH and CVVH),
and continuous arterio-venous and veno-venous hemodiafiltration (CAVHD and
CVVHD) - and immunologic and other heterogeneous disorders, such as plas-
ma exchange (PE).
Several experimental studies were devised to evaluate the effects of these
techniques on sepsis-induced cardiorespiratory derangements.
On the whole, these studies demonstrated that in septic animals hemofiltra-
tion was associated with an improvement of cardiovascular function and a re-
duction of extravascular lung water, regardless of a change in blood volume [51-
53]. Interestingly, the fluid means of hemofiltration removed from the septic an-
imals, but not that drawn from normal controls, was able to reduce myocardial
contractility in other healthy animals [54,55]. In another model of Gram+ sep-
sis, treated with CAVH at three different levels of plasma filtration (5.5%,
16.6% and 33.4%, respectively), an increased survival time paralleled the in-
crease of depuration efficiency [56].
With the aim of investigating a more effective depuration technique, several
studies focussed on PE. Results have again been controversial. Experimentally,
Natanson et al. [57] observed a detrimental hemodynamic effect of machine-dri-
ven PE as compared with the controls, treated with sham PE. On the contrary, in
another animal model of sepsis, Busund et al. [58] reported hemodynamic im-
provement and a higher survival rate in animals treated with PE and fresh frozen
plasma transfusions (FFP) , as compared with the control group; furthermore,
these authors observed a decrease of both TNF and IL-l in the PE group, where-
as in the group treated with FFP only a decrease of the TNF was observed. In
another experiment, the same authors attributed the increased mortality rate as-
sociated with PE or FFP transfusions to the negative inotropism associated with
the depletion of ionized calcium [59]. The clinical application of hemopurifica-
tion techniques paralleled the experimental investigations, as several authors ob-
served both an improvement of cardiorespiratory function and an increased sur-
vival in patients with different cardiovascular disorders treated with these tech-
niques. Gotloib et al. [60] demonstrated an improvement of hemodynamics and
gas exchanges in patients with septic ARDS treated with HF, which occurred in-
dependently from the removal of fluid. They also demonstrated that some sub-
stances involved in the septic process, including endorphines and arachidonic
acid derivatives, had a high sieving coefficient through the filter membrane, and
related the beneficial cardiorespiratory effects associated with HF to their re-
moval. Better hemodynamic cardiovascular performance was observed by
COl·aim et al. [61] in patients with low cardiac output after cardiac surgery treat-
ed with CAVH, and attributed this effect to the removal of myocardial depres-
sant factors generated during the cardio-pulmonary bypass (CPB). In septic pa-
tients, Berlot et al. [62] demonstrated that PE was associated with a significant
improvement of the increase of LVSWI, CI, D0 2 and \10 2 and that these
changes were more marked in patients whose cardiac function was more de-
pressed before treatment; however, these changes were not associated with an
improvement of mortality. These beneficial effects have been mainly ascribed to
the removal of septic mediators. This hypothesis is indirectly also justified by
some studies in which the reduction of mortality of septic patients was associat-
ed with increased depuration effectiveness. Storck et al. [63] demonstrated a re-
duced mortality rate in patients with postoperative ARF treated with pump-dri-
ven CVVH as compared with the group treated with CAVH, and hypothesized
that this result could be ascribed to the removal of larger amounts of mediators
obtained with CVVH. Also Barzilay et al. [64] evaluated the effects of depura-
tion techniques in four groups of septic patients treated conservatively (i.e. with-
out any depuration treatment) or with CAVH, CAVHD and CAVHD associated
with PE, respectively. In this latter group, the mortality rate of 36% was signifi-
cantly lower than in patients treated with CAVH, with CAVHD (respectively
71 % and 50%) and in patients who were not treated with any blood purification
technique (87%).
In recent years it became possible to assess cytokines in biological fluids,
and this prompted some authors to measure these substances both in the blood
and in the ultrafiltrate. However, the available studies carry conflicting results.
In a group of septic patients with ARF treated with CVVH, Bellomo et al. [65]
demonstrated that in 12 out of 18 patients (66%) the serum of both TNF and IL-
l decreased not significantly 4 and 24 hours after the treatment had been started,
respectively. Other studies, however, failed to confirm these findings. In septic

patients treated with CVVH, Heering et al. [66] did not observe any significant
decrease of circulating TNF-a, even if the treated patients presented an im-
provement of hemodynamics. The same results have been reported in another
study, in which hemofiltration performed in patients with early-onset SIRS was
associated with an increased clearance of IL-6, even if both blood IL-6 and
TNF-a levels remained unchanged [67]. These studies indicate that in septic pa-
tients a) blood depuration techniques are usually associated with an improve-
ment of cardiorespiratory performance, which has been largely attributed to the
removal of septic mediators, but b) measurement of these substances in patients
treated did not provide conclusive evidence, resulting diminished in some inves-
tigations and unaltered in others; moreover, c) their real impact on the outcome
remains to be established. Although these apparently mutually exclusive find-
ings are difficult to explain, it is nonetheless possible to formulate some hy-
potheses. Firstly, some mediators could be produced during blood-filter interac-
tion [68]. Actually, Byrick failed to demonstrate a drop in serum TNF levels in
patients undergoing CAVHD due to rhabdomyolysis-induced ARF [69]. On the
contrary, TNF increased both in the blood leaving the filter and in systemic cir-
culation (from 646 to 765 pglml) after 18 hours since the treatment was started,
but was undetectable in the ultrafiltrate. Secondly, some mediators circulate
bound to other plasma protein; the resulting molecular weight being higher,
trans-membrane passage is impaired [70, 71]. Thirdly, the release of septic me-
diators may not be constant; actually, blood levels of endotoxin TNF fluctuate
during the clinical course of septic patients and can be influenced by several
factors, including the administration of antibiotics [72, 73] or the occurrence of
hypotension [74]; moreover, at least in some studies, it may be that TNF and
other cytokines have been measured in different phases, leading to conflicting
results. Finally, excessive attention may have been paid to the "wrong" m~dia
tors: Hoffmann et al. [75] have recently demonstrated that in septic patients
treated with hemofiltration improved cardiovascular function was associated
with a reduction of C3a and C5a, in the absence of any significant change of
TNF-a, IL-l, IL-6 and IL-8. On the basis of current evidence, it can be conclud-
ed that CVVH does not remove a substantial amount of TNF, but can be associ-
ated with a substantial decrease of other more distal mediators involved in the
pathogenesis of SIRS and MODS [76]. More powerful techniques, such as PE
or derived techniques, may effectively remove all mediators, due to the higher
cut-off value of the membranes. However, the real effect of these techniques on
the prognosis of septic patients without acute renal failure is not yet clear.
Lights and shadows on blood cytokine measurement

A 65-year old woman was admitted to the hospital after 7 days of fever and
vomit, with the diagnosis of renal colic. Eight hours after the admission the pa-
tient became hypotensive (90/50 mmHg). Coagulative abnormalities were also
present. Abdominal US and CT scans revealed right pyonephrosis. The patient

was operated on, and a nephrectomy was performed. In the postoperative period
the patient developed severe hypoxemia despite mechanical ventilation at 60%
Fi02 and with a PEEP of 10 cm H20 for 24 hours. The initial blood lactate level
was high (2,5 MmollL) and remained elevated for 6 days. Blood TNF and IL-6
values were also elevated, suggesting a poor prognosis. However, after draining
an abdominal purulent collection, these values returned to normal and the clini-
cal conditions improved. The patient was then discharged to the surgical ward
after 10 days in the ICU.
Many investigations have concentrated on the correlation between the time
course of the blood concentrations of sepsis mediators and the outcome. Once
again, the results have not been unequivocal. Some authors demonstrated that
blood levels of TNF remained elevated in septic patients who ultimately devel-
oped MODS and died, as compared with patients whose conditions improved
[77, 78]. However, other studies did not confirm these findings [79].
It would therefore seem that the most appropriate method to monitor the evo-
lution of SIRS and sepsis is far from been elucidated, as most of the currently
used indicators of the inflammatory response, such as body temperature, white
cell count, erythrocyte sedimentation rate or C-protein concentration, are unspe-
cific parameters with changing reliability. Recent studies have shown that IL-6
has a role in the activation of the acute phase response in the liver. IL-6 is pro-
duced by various cells, including monocytes, macrophages, and lymphocytes.
IL-6 is thought to be the most important mediator of the inflammatory response,
and it is also of prognostic value in sepsis and bums [80]. Tissue injury leads to
local production of IL-6 and other cytokines that mediate most of the systemic
aspects of inflammation [81]. IL-6 is a candidate as a laboratory test that rapidly
normalizes after uncomplicated surgery or trauma. In patients undergoing
scheduled surgery, the IL-6 serum value rose within 2-4 hours after the interven-
tion, and its peak value varied in the various surgical groups, since the response
was related to the duration of surgery and the degree of tissue trauma [82]. Oth-
er authors noted a concentration of IL-6 peaks between 4 and 6 hours after
surgery and/or trauma, but there is a poor correlation with blood loss, fever,
white cell count or duration of surgery [83]. Conversely, its value is well related
to the occurrence of infections in the postoperative period [84]. In patients with
severe infections, initial high serum levels of IL-6 have been related to the
severity of sepsis and the mortality [85].
Procalcitonin (PCT) is an innovative diagnostic variable with features differ-
ent from other presently available indicators of the inflammatory response. The
amino acid sequence (116 aa) of PCT is identical to the prohormone of calci-
tonin. In the normal and healthy individual, hormonal active calcitonin is pro-
duced and secreted by C-ce1ls of the thyroid gland after specific intracellular
proteolytic cleavage of the prohormone. PCT is selectively induced during bac-
terial inflammation, and also in sepsis and MODS. PCT is only induced as a re-
sponse to systemic infection. Bacterial colonization, capsuled abscesses and
limited local infections in fact do not induce PCT. The amount of PCT induced
and the increase of plasma levels are correlated with the extent of the inflamma-
tory reaction. At the end of the acute inflammatory reaction, PCT concentra-
tions immediately decrease according to their plasma half-life times. It has been
demonstrated that endotoxin injection into normal subjects results in PCT se-
cretion in the absence of an increase in calcitonin [86]. The PCT produced un-
der an infective stimulation is most likely not produced by C-cells of the thy-
roid, and neuroendocrine cells of the lung or of the gut are indicated as the pos-
sible sources of PCT. In humans PCT is detectable after endotoxin injection at
4h, with peaks at 6h [86]. PCT rises heavily during sepsis in children, and was
within or slightly above the normal range in patients with peripheral, local or
viral infection [87]. The PCT serum value is clinically related to pulmonary in-
jury [88].
In our study we evaluated eleven patients (7 males, 4 females, age = 68.9 y,
52-82 y) undergoing major surgery, and twelve patients (7 males, 5 females, age
= 58.16 y, 16-80 y) with SIRS according to ACCP/SCCM criteria [2]. In the pa-
tients undergoing surgery, the blood samples were collected before anesthesia
(basal) and after surgery. In the SIRS patients blood samples were collected
every 24 h after admission to the Intensive Care Unit.
The data are summarized and expressed as mean and range in Table 1.
Table 1.
peT (nglml) IL-6 (pglml)

Basal 0.136 (0.025-0.53) 11.84 (0-44.59) a, g
After surgery 0.252 (0.038-0.341) b 252.787 (119.74-1517.766) a
SIRS 24h 9.58 (0-336) b 293.503 (76.6-15000) g
SIRS 48 h 4.54 (0-420) 215.715 (21.8-11139)
SIRS 72h 3.755 (0-304) 139.61 (7.45-2168)
a Wilcoxon test: p = .0033

b Mann-Whitney U test: p = .0007
g Mann-Whitney U test: p = .0001
Our data demonstrate that the IL-6 level in postoperative patients with SIRS
was significantly higher than the basal value. The patient groups are not similar,
but the group undergoing major surgery may be considered a control group:
there was no occurrence of fever, white cell increase or other signs of inflam-
mation. The IL-6 value decreased rapidly in this group, as opposed to the trend
observed in the SIRS group. PCT increased heavily in SIRS, but not after un-
complicated surgery. Much higher cytokine concentrations were found in the
peritoneal fluid than in the plasma, which suggests that the postoperative cy-
tokine response may to a large extent originate from the peritoneal cavity [89].
A local cytokine response may stimulate mesothelial cells and later induce
chemotaxis of neutrophils and a further production of cytokines. Peritoneal cy-
tokines may be absorbed into the portal and systemic bloodstream and stimulate
the production of acute phase protein by hepatocytes. In SIRS, IL-6 is higher
than in the normal control group and may be related to the prognosis [90]. Re-
cently, some authors have demonstrated high serum levels of PCT in severe ac-
cidental injury as a sign of a serious inflammatory state [91]. In our SIRS
group, the PCT is 90 times higher than normal. The high value of IL-6 in SIRS
is similar to that in severe accidental injury. The strongly associated high values
of PCT and IL-6 (r = .844, p = .0001) indicate a heavy reactive inflammatory
condition, which may be related to respiratory failure. PCT does not increase
postoperatively because of the absence of inflammatory states, while there is a
release of cytokines which represents the physiological response to injury [5]. A
recent paper indicates PCT in association with the measurement of nitrite/ni-
trate ratio as the most suitable test for defining patient with septic shock. PCT
did not increase in patients with such inflammatory conditions as cardiogenic
shock or bacterial pneumonia. Compared with TNF or IL-6, PCT had better
sensitivity, specificity and predictive value for positive tests, and predictive val-
ue for negative tests [92].
Lights and shadows on the selective decontamination of the digestive

tract
What is SDD?
The term systemic digestive decontamination (SDD) has been extended to dif-
ferent protocols aiming at preventing infections due to potentially pathogenic
micro-organisms (PPMs) in critically ill patients. Several authors who were dis-
appointed with SDD did not use an appropriate SDD protocol. The full SDD
protocol comprises four components: a combination of selected non-absorbable
antimicrobials, parenteral antibiotics, a high standard of hygiene, and surveil-
lance cultures [93-95].
Secondary endogenous infections caused by PPMs not carried in the throat
and/or gut on ICU admission, but acquired during ICU stay, can be prevented by
non-absorbable antimicrobials administered topically in the oropharynx and gut.
Polymyxin E, tobramycin and amphotericin B (PTA) is the most frequent asso-
ciation: a 2% paste or gel of PTA is applied to the oropharyngeal mucosa four
times a day, and 9 m1 of suspension containing 100 mg of polymyxin E, 80 mg
of tobramycin and 500 mg of amphotericin B is administered in the digestive
tract four times a day. In studies in which lower doses of polymyxin E or other
aminoglycosides, quinolones, macrolides, or polyenes were used, or where su-
cralfate was employed, PPMs were difficult to eradicate [96-98]. This regimen
has been shown to eradicate aerobic Gram-negative bacilli (AGNB) and to re-
duce fecal endotoxin concentration in healthy human volunteers [99].
Primary endogenous infections, i.e. infections caused by PPMs carried in the
throat andlor gut on leU admission, will be prevented if, for the first four days,
such parenteral antimicrobials as cefotaxime or ceftazidime (if patients are sus-
pected of carrying Pseudomonas) are added at a daily dosage of 100 mg/kg.
Both parenteral and topical antimicrobials cannot control exogenous infec-
tions, because these infections are caused by micro-organisms introduced direct-
ly into the internal organs without previous carriage. High standards of hygiene
are necessary to prevent these infections.
Surveillance samples of throat and rectum represent the fourth component of
the full SDD protocol. In general, the surveillance set is comprised of throat and
rectal swabs taken on admission and afterwards twice weekly. They are
processed semiquantitatively to determine the carriage level of PPMs. They
monitor the effectiveness of the PTA protocol, assess the level of hygiene of the
unit, allow the identification of the type of infection, and determine the intrinsic
pathogenicity index for a micro-organism [100].
What SDD cannot do

The aim of the full SDD protocol is to prevent and/or eradicate oropharyngeal
and intestinal carriage of both "community" and "hospital" PPMs, leaving the
indigenous flora undisturbed. Methicillin-resistant Staphylococcus aureus (MR-
SA) is intrinsically resistant to the antibiotics commonly used in SDD. Surveil-
lance cultures can detect early MRSA carriage so that topical 2% vancomycin
may be used together with skin disinfection with 2% aqueous chlorhexidine.
Moreover, infections due to low pathogens, such as viridans streptococci, co-
agulase-negative staphylococci (eNS), enterococci and anaerobes, are not pre-
vented by SDD. An increase in eNS and enterococcal carriage may be common
is SDD treated patients [101].
Which patients may benefit from SDD?

SDD is one of the few maneuvers in the leU subjected to scientific and statisti-
cal appraisal. It has been evaluated in approximately 50 controlled studies, and
its effects have been assessed in several reviews and meta-analyses.
All meta-analyses showed a significant reduction in respiratory tract infec-
tions in the SDD group, with a marked effect in patients treated with topical
plus systemic antimicrobials (up to 60%) [102-104]. Urinary tract infections,
bacteremia and wound infections were also reduced in SDD patients [105, 106].
Full SDD protocol significantly reduced mortality by around 20% [102-104].
No trial included in the most complete meta-analyses showed a significant
harmful effect of SDD [102-104].
The full SDD protocol will be useful in critically ill patients admitted to the
ICU for (surgical) trauma, burns, acute pancreatitis or liver failure, or acute de-
terioration of the underlying disease (e.g. chronic obstructive pulmonary dis-
ease, cardiac failure requiring more than 3 days of mechanical ventilation) [93-
95]. SDD without parenteral antibiotics may be useful in patients submitted to
elective major abdominal or thoracic surgery [93-95]. Two days before surgery
the patient receives PTA; at the induction of anesthesia, a proper parenteral an-
tibiotic prophylaxis will be administered. This prophylaxis has been shown to
reduce morbidity and mortality in esophageal, cardiac, and gastric surgery and
liver transplantation [107-109].
Lights and shadows on the surgical approach to the septic patient

A 37 year-old man was admitted to the surgical department for acute pancreati-
tis. The radiological examination revealed the presence of two cysts in the pan-
creas. The patient presented a progressive respiratory insufficiency (Pa02 < 40
mmHg) associated with hypotension and leukocytosis (WBC 17,500/ml). An
aggressive treatment was instituted, which included mechanical ventilation, the
insertion of a Swan-Ganz catheter and percutaneous drainage of the pancreatic
collections. Despite this approach, the patient developed ARDS associated with
oliguria. Blood cultures were positive for Ps. aeruginosa. Continuous arterio-ve-
nous hemofiltration was then implemented, and the clinical conditions improved
in two weeks, with the full recovery of the respiratory, renal and pancreatic
functions.
The conventional surgical approach to intra-abdominal infections includes
laparotomy, drainage of fluid collections, peritoneal lavage, removal of septic
foci and closure of the abdominal walL The source of sepsis must be accurately
researched, and involved organs should, if possible, be removed (cholecystecto-
my, appendectomy, etc.) or sutured. Non-viable tissues should be eliminated,
and the hemostasis must be extremely accurate to avoid hematomas which could
become foci of secondary infections. Intestinal anastomoses should be avoided
due to the high risk of delayed leaking of sutures. Exteriorization of the ansae is
highly recommended. The main limitation of conventional treatment is the fail-
ure to clear completely all septic foci, which ultimately leads to the occurrence
of postoperative intra-abdominal abscesses, possibly causing septic shock and
MODS. It is therefore clear that the early diagnosis of postoperative complica-
tions is of paramount importance in patients with previously treated intra-ab-
dominal infections to prevent the deleterious effects of poorly drained collec-
tions. Re-Iaparotomy should be considered in three different clinical settings.
The first includes peritonitis and firmly documented abdominal abscesses, in
which percutaneous drainage is not feasible. Under these circumstances, re-Ia-
parotomy is directed toward the definitive eradication of the septic focus. The
second indication for re-Iaparotomy includes postoperative MODS which can-
182 G. Berlot et aI.
not be attributed to other causes. In such cases, re-laparotomy has a diagnostic

and a therapeutic role. The third indication for re-laparotomy includes the per-
sistence of clinical signs of sepsis despite previous surgical exploration. In these
patients, abdominal procedures are scheduled every 24-72 hours, depending on
the clinical evolution. Abdominal wall closure can be accomplished in the con-
ventional way or by using loose sutures (bridge) or zippers, which reduce the in-
tra-abdominal pressure and facilitate further interventions. When the abdomen
will be free of pus and/or necrotic tissues, the re-laparotomies can be discontin-
ued, even in the presence of small fluid collections.
Another option consists of leaving the abdominal surgical wound open [11 0].
The main drawback of this technique is the need for prolonged mechanical ven-
tilation [111]. Clinical studies report conflicting results: Mughal et aL [112]
demonstrated good survival in patients treated with this technique, whereas An-
derson et aL [113] did not observe any major effect on the outcome. According
to Wittman et al. [114] the indications for the open approach are:
- diffuse peritonitis associated with conditions which impede the complete
eradication of the septic focus or the surgical removal of the involved
organ(s);
- diffuse peritonitis lasting> 48 hours;
- diffuse peritonitis of multiple abdominal abscesses in patients with MODS or
rapidly worsening clinical conditions.
The open approach has several advantages, including daily surveillance
which requires only minimal sedation and prevents increased intra-abdominal
pressure, which can have deleterious effects on the renal function. Some limita-
tions do however exist, including the need for a prolonged stay in the Intensive
Care Unit, the fluid loss, the ileus pain, and the risk of such post-procedural
complications as fistulas, laparocele, bleeding and the perforation of abdo~nal
viscera.
The open approach can be performed with different techniques. The simplest
consists in leaving the wound open, which is covered with a sterile dress. The
more widely used technique utilizes a non-absorbable biocompatible net, which
is sewed to the edges of the surgical wound. The open approach may be limited
to some intra-abdominal compartment, such as the lesser sac, during severe
acute infected pancreatitis.
Conclusions
Despite major progress in the understanding of the pathophysiologic mecha-
nisms linking an initially circumscribed infection (i.e. pneumonia) to the devel-
opment of sepsis and MODS, the survival of patients with these systemic distur-
bances is still poor. This could be ascribed to several factors, including a) a rela-
tively late diagnosis, caused by the time gap existing between the triggering of
the mediator network and the occurrence of tissue damage and the onset of
symptoms; b) the side effects of commonly adopted therapeutic measures (i.e.
certain antibiotics), which may in many cases be considered a double-edged
sword, and c) still insufficient recognition of individual functional reserves,
which even the most advanced treatments tend to leave untouched. In our opin-
ion, the results of the large clinical trials with anti-mediator agents only added
further to the confusion, as mortality rate in highly complicated patients, such as
those with sepsis and MODS, is a rather rough endpoint to verify the effective-
ness of a treatment, and clinical and underlying conditions of the patients en-
rolled in these trials were too different.
In the treatment of septic patients, the shadows stretching out between the ar-
eas of light are still far from being illuminated - unless more appropriate end-
points are chosen, and really comparable patients are enrolled in clinical trials.
References
1. Bernard GR (1995) Sepsis trials. Am J Resp Crit Care Med 102:4-10
2. Members of the American College of Chest Physician/Society of Critical Care Medicine Con-
sensus Committee (1992) Definitions for sepsis and organ failure and guidelines for the use of
innovative therapies in sepsis. Crit Care Med 20:864-890
3. Vincent JL, Bihari D (1992) Sepsis, severe sepsis or sepsis syndrome: need for clarification.
Intens Care Med 18:255-257
4. Vincent JL (1997) Dear SIRS, I'm sorry to say I that I don't like you ... Crit Care Med 25:
372-374
5. Bone RC, Grodzin CJ, Balk RA (1997) Sepsis: a new hypothesis for pathogenesis of the dis-
ease process. Chest 112:235-243
6. Pittet D, Rangel-Frausto M, Li N et al (1995) Systemic inflammatory response, sepsis, severe
sepsis and septic shock: incidence morbidities and outcomes in ICU patients. Intens Care Med
21:32-309
7. Rangel-Frausto M, Pittet D, Costigan M et al (1995) The natural history of the systemic
inflammatory response syndrome (SIRS). JAMA 273:117-123
8. Salvo I, de Cian W, Musicco M et al (1995) The Italian SEPSIS study; preliminary results on
the incidence and evolution of SIRS, sepsis, severe sepsis and septic shock. Intens Care Med
21:S244-S249
9. Rackow EC, Astiz ME (1991) Pathophysiology and treatment of septic shock. JAMA 266:
548-554
10. Berlot G, Vincent JL (1992) Cardiovascular effects of cytokines. Clin Intens Care 3:199-205
11. Davies MG, Hagen PO (1997) Systemic inflammatory response syndrome. Br J Surg 84:
920-935
12. Moldawer LL (1994) Biology of proinflammatory cytokines and their antagonist. Crit Care
Med 22:S3-S7
13. Christman JW, Holden EP, Blackwell TS (1995) Strategies for blocking the systemic effects of
cytokines in the sepsis syndrome. Crit Care Med 23:955-963
14. Ziegler EJ, McCuchan JA, Fierer Jet al (1982) Treatment of gram-bacteremia and shock with
human antiserum to a mutant Escherichia coli. New Engl J Med 307: 1225-1230
15. Lachman E, Pitsoe SB, Gaffin SL (1984) Antilypolisaccharide immunotherapy in the manage-
ment of septic shock of obstetrical and gynecological origin. Lancet 1:981-983
184 G. BerIot et al.
16. Fomsgaard A, Baek L, Foomsgard JS et al (1988) Preliminary study in treatment of septic

shock patients with antilypolisaccharide IgG from blood donors. Scand J Infect Dis 21:
697-708
17. Talan DA (1993) Recent developments in our understanding of sepsis: evaluation of antiendo-
toxin antibodies and biological response modifiers. Ann Emer Med 22: 1871-1990
19. Greenman RL, Schein RMH, Martin MA et al (1991) A controlled clinical trial of E5
murine monoclonal IgM antibody to endotoxin in the treatment of Gram- sepsis. JAMA 266:
1097-1102
20. Ziegler EJ, McCutchan JA, Fierer J et al (1991) Treatment of Gram- bacteremia and septic
shock with HA-IA human monoclonal antibody against endotoxin. A randomized, double
blind, placebo controlled trial. N Engl J Med 324:429-436
21. Fisher CJ, Zimmerman J, Khazaeli MB et al (1990) Initial evaluation of human monoclonal
anti-lipid A antibody (HA-IA) in patients with sepsis syndrome. Crit Care Med 18:1311-1315
22. Greenberg RN, Wilson KM, Kunz AY et al (1992) Observations using antiendotoxin antibody
(E5) as adjuvant therapy in humans with suspected, serious Gram- sepsis. Crit Care Med 20:
730-735
23. Bone RC, Balk RA, Fein AM et al (1995) A second large controlled clinical study of E5, a
monoclonal antibody to enaotoxin: results of a prospective, multicenter, randomized, con-
trolled study. Crit Care Med 23 :994-1 006
24. Cunnion RE (1992) Clinical trials of immunotherapy for sepsis. Crit Care Med 20:721-723
25. Silva AT, Bayston KF, Cohen J (1990) Prophylactic and therapeutic effects of a monoclonal
antibody to tumour necrosis factor-alfa in experimental Gram- shock. J Inf Dis 162:421-427
26. Exley AR, Cohen J, Buurman WA et al (1990) Monoclonal antibody to TNF in severe septic
shock. Lancet 335:1275-1277
27. Vincent JL, Bakker J, Marecaux G et al (1992) Administration of anti TNF antibodies
improves left ventricular function in septic shock patients: results of a pilot study. Chest 101:
810-815
28. Dhainaut JFA, Vincent JL, Richard C et al (1995) DP 571, a humanized antibody to tumour
necrosis factor-alpha: safety, pharmacokinetics, immune response, and influence of the anti-
body on cytokine concentrations in patients with septic shock. Crit Care Med 23:1461-1469
29. Cohen J, Carlet J (1996) INTERSEPT: an international, multicenter, placebo-controlled trial
of monoclonal antibody to human tumor necrosis factor-a in patients with sepsis. Crit Care
Med 24:1431-1440
30. Rheinhart K, Wiegand-Lohnert C, Grimminger F et al (1996) Assessment of the safety and
effectiveness of the monoclonal anti-tumour necrosis factor antibody fragment, MAK 195F, in
patients with sepsis and septic shock: a multicenter, randomized, placebo-controlled, dose
ranging study. Crit Care Med 24:733-742
31. Granowitz EV, Santos AA, Poutsaka DD et al (1991) Production of Interleukin-l receptor
antagonist during experimental endotoxemia. Lancet 1338: 1423-1424
32. Dinarello CA (1991) The proinflammatory cytokines Interleukin-l and tumor necrosis factor
and treatment of septic shock syndrome. J Inf Dis 163: 1177-1184
33. Ohlsson K, Bjork P, Bergenfeldt M et al (1990) Interleukin-l receptor antagonist reduces mor-
tality from endotoxin shock. Nature 343:550-552
34. Arend WP (1991) Interleukin 1 receptor antagonist: a new member of Interleukin 1 family. J
Clin Invest 88: 1445-1451
35. Fisher E, Marana MA, Van Zee KJ et al (1992) Interleukin-l receptor antagonist improves sur-
vival and hemodynamic performance in Escherichia coli septic shock, but fails to alter host
responses to sublethal endotoxemia. J Clin Invest 89: 1551-1557
36. Fisher CJ, Slotman GJ, Opal SM et al (1994) Initial evaluation of human recombinant Inter-
leukin 1 receptor antagonist in the treatment of sepsis syndrome: a randomized, open label,
placebo-controlled multicentre trial. Crit Care Med 22: 12-21
37. Fisher CJ, Dhainaut JF, Opal SM et al (1994) Recombinant human interleukin 1 receptor
antagonist in the treatment of patients with sepsis syndrome. Results from a randomized, dou-
ble blind, placebo-controlled trial. JAMA 271:1836-1843
38. Knaus WA, Harrell FE, LeBreque JF et al (1996) Use of predicted risk of mortality to evaluate
the effectiveness of anticytokine therapy in sepsis. Crit Care Med 24:46-56
39. Bazzoni F, Beutler B (1996) Seminars in Medicine at the Beth Israel Hospital, Boston: The
tumour necrosis factor ligand and receptor families. New Engl J Med 34: 1717 -1725
40. Sorkine P, Setton A, Halpern P et al (1995) Soluble tumour necrosis factor receptors reduce
bowel ischemia-induced lung permeability and neutrophil sequestration. Crit Care Med 23:
1377-1381
41. Van Zee KJ, Kohno T, Fisher E et al (1992) Thmour necrosis factor soluble receptors circulate
during experimental and clinical inflammation and can protect against excessive tumour
necrosis factor alpha in vitro and in vivo. Proc Natl Acad Sci USA 89:4845-4849
42. Mohler KM, Torrance DS, Smith CA et al (1993) Soluble tumour necrosis factor (TNF) recep-
tors are effective therapeutic agents in lethal endotoxemia and function simultaneously as both
TNF carriers and TNF antagonists. J Immunol151:1548-1561
43. Fisher CJ, Agosti JA, Opal SM et al (1996) Treatment of septic shock with the tumor necrosis
factor receptor:Fc fusion protein. New Engl J Med 334: 1697-1702
44. Abraham E, Glauser MP, Butler T et al (1997) Tumor necrosis factor receptor fusion protein in
the treatment of patients with severe sepsis and septic shock. A randomized controlled multi-
center trial. Jama 277: 1531-1538
45. Bone RC (1992) Phospholipids and their inhibitors: a critical evaluation of their role in the
treatment of sepsis. Crit Care Med 20:884-890
46. Lefer A (1989) Significance of lipid mediators in shock states. Circ Shock 27:3-12
47. Sun X, Hsueh W (1988) Bowel necrosis induced by tumour necrosis factor in rats is mediated
by platelet activating factor. J Clin Invest 81:1328-1331
48. Sun X, Hsueh W, Torre-Amione G (1990) Effects of in vivo priming on endotoxin induced
hypotension and tissue injury: the role of PAF and tumour necrosis factor. Am J Pathol 136:
949-956
49. Thompson WA, Coyle S, Van Zee K et al (1994) The metabolic effects of platelet activating
factor in endotoxemic man. Arch Surg 129:72-79
50. Dhainaut JFA, Tenaillon A, Le Tulzo Y et al (1994) Platelet-activating factor antagonist BN
52021 in the treatment of severe sepsis: a randomized, double-blind, placebo-controlled, mul-
ticenter clinical trial. Crit Care Med 22: 1720-1728
51. Gomez A, Wang R, Unruh H et al (1990) Hemofiltration reverses left ventricular dysfunction
during sepsis in dogs. Anesthesiology 73:671-785
52. Stein B, Pfenninger, Grunert A et al (1990) Influence of continuous haemofiltration on haemo-
dynamics and central blood volume in experimental endotoxic shock. Intens Care Med 16:
494-499
53. Stein B, Pfenninger, Grunert A et al (1991) The consequences of continuous haemofiltration
on lung mechanics and extravascular lung water in a porcine endotoxic shock model. Intens
Care Med 17:293-298
54. Grootendorst AF, van Bommel EFH, van der Hoven B et al (1992) High volume hemofiltration
improves hemodynamics of endotoxin-induced shock in the pig. J Crit Care 7:67-75
55. GrootendorstAF, van Bommel EFH, van der Hoven Bet al (1992) High volume hemofiltration
improves right ventricular function endotoxic shock in the pig. Intensive Care Med 18:
235-240
56. Lee PA, Matson JR, Pryor RW, Hinshaw LB (1993) Continuous arteriovenous hemofiltration
therapy for staphylococcus aureus-induced septicemia in pigs. Crit Care Med 21 :914-924
57. Natanson C, Hoffman WD, Danner RL et al (1993) Plasma exchange does not improve sur-
vival in a canine model of human septic shock. Transfusion 33:243-248
58. Busund R, Lindsetmo RO, Rasmussen LT et al (1991) Thmour necrosis factor and Interleukin
1 appearance in experimental Gram-septic shock: the effects of plasma exchange with albumin
and plasma infusion. Arch Surg 126:591-597
59. Busund R, Lindsetmo RO, Balteskard L et al (1993) Repeated plasma therapy induces fatal
shock in experimental septicemia. Circ Shock 40:268-275
60. Gotloib L, Barzilay E, ShustakA et al (1986) Hemofiltration in septic ARDS: the artificial kid-
ney as an artificial endocrine lung. Resuscitation 13:123-132
61. Coraim FJ, Coraim HP, Ebermann R, Stellwag PM (1986) Acute respiratory failure after car-
diac surgery: clinical experience with the application of continuous arteriovenous hemofiltra-
tion. Crit Care Med 14:714-717
62. Berlot G, Gullo A, Fasiolo S et al (1997) Hemodynamic effects of plasma exchange in septic
patients: preliminary report. Blood Purification 15:45-53
63. Storck M, Hartl WH, Zimmerer E, Inthom D (1991) Comparison of pump-driven and sponta-
neous continuous hemofiltration in postoperative acute renal failure. Lancet 337:452-455
64. Barzilay E, Kessler D, Berlot G et al (1989) The use of extracorporeal supportive techniques as
additional treatment for sepsis-induced MOF patients. Crit Care Med 17:634-637
65. Bellomo R, Tipping P, Boyce N (1993) Continuous veno-venous hemofiltration with dialysis
removes cytokines from the circulation of septic patients. Crit Care Med 21 :522-526
66. Heering P, Morgera S, Schmitz FJ et al (1997) Cytokine removal and cardiovascular hemody-
namics in septic patients with continuous venovenous hemofiltration. Intens Care Med 23:
288-296
67. Sander A, Armbruster W, Sander B et a1 (1997) Hemofiltration increases IL-6 clearance in ear-
ly systemic inflammatory response syndrome but does not alter IL-6 and TNF-a plasma con-
centrations. Intens Care Med 23:878-884
68. Peetre C, Thysell H, Grubb A et al (1988) A tumour necrosis factor binding protein is present
in human biological fluids. Eur J Haematol41 :414-419
69. Byrick RJ, Goldstein MB, Wong PJ (1992) Increased plasma tumour necrosis factor concen-
tration in severe rhabdomyolysis is not reduced by continuous arteriovenous hemodialysis.
Crit Care Med 20: 1483-1486
70. Nagaki M, Hughes RD, Lau JIN, Williams R (1991) Removal of endotoxin and cytokines by
adsorbents and the effect of plasma protein binding. Int J Artif Org 14:43-50
71. Bellomo R, Parkin G, Love J, Boyce N (1993) A prospective comparative study of continuous
arteriovenous hemodiafiltration and continuous venovenous hemodiafiltration in critically ill
patients. Am J Kidney Dis 21 :400-404
72. Dofferhoff ASM (1991) Release of endotoxin and other mediators during the treatment of
Gram- sepsis. An introduction. Rijksuniversity Groeningen 1-32
73. Shenep J, Flynn PM, Barrett FF et al (1988) Serial quantitation of endotoxemia and bac-
teremia during therapy for Gram- bacterial sepsis. J InfDis 157:565-568
74. Endo S, Inada K, Yamada Y et al (1994) Plasma endotoxin and cytokine concentration in
patients with hemorrhagic shock. Crit Care Med 22:994-1001 '
75. Hoffman IN, Hartl WH, Deppish R et al (1996) Effect ofhemofiltration on hemodynamics and
systemic concentrations of anaphylotoxins and cytokines in human sepsis. Intens Care Med
22:1360-1367
76. Schetz M, Ferdinande P, Van den Berghe G, Lauwers P (1995) Removal of pro-inflammatory
cytokines with renal replacement therapy: sense or nonsense? Intens Care Med 21:169-176
77. Pinsky MR, Vincent JL, Deviere J et al (1993) Serum cytokine levels in human septic shock:
relation to mUltiple system organ failure and mortality. Chest 103:565-575
78. Calandra T, Baumgartner JD, Grau GE et al (1990) Prognostic values of tumour necrosis fac-
tor!cachectin, interleukin-l, interferon-alpha, and interferon gamma in the serum of patients
with septic shock. J Infect Dis 161 :982-987
79. Calandra T, Gerain J, Heumann D et al (1991) High circulating levels of interleukin-6 in
patients with septic shock evolution during sepsis, prognostic value and interplay with other
cytokines. AmJ Med 91:23-29
80. Bone RC (1996) Toward a theory regarding the pathogenesis of the systemic inflammatory
response syndrome: What we do and do not know about cytokine regulation. Crit Care Med
24:163-172
81. Galley HF, Webster NR (1996) The immuno-inflammatory cascade. Br J Anaesth 77:11-16
82. Cruickshank AM, Fraser WD, Bums HJ et al (1990) Response of serum IL-6 in patients under-
going elective surgery of varying severity. Clinical Science 79: 161-165
83. Pullicino EA, Carli F, Poole S et al (1990) The relationship between the circulating concen-
trations of IL-6, TNF and the acute phase response to elective surgery and accidental injury.
Lymphokine Res 9:231-238
84. Fassbender K, Pargger H, Muller W, Zimmerli W (1993) IL-6 and acute phase protein con-
centrations in surgical intensive care unit patients: diagnostic sign in nosocomial infection.
85. Waage A, Aasen AO (1992) Different role of cytokine mediators in septic shock related to
meningococcal disease and surgery. Immunology 127:221-230
86. Dandona P, Nix 0, Wilson MF et iiI (1994) Procalcitonin increase after endotoxin injection in
normal subjects. J Clin Endocrinol Metab 79: 1605-1608
87. Assicot M, Gendrel 0, Carsin H et al (1993) High serum procalcitonin concentrations in
patients with sepsis and infection. Lancet 341:515-518
88. Nylen ES, O'Neill W, Jordan MH et al (1992). Serum procalcitonin as an index of inhalation
injury in bums. Horm Met Res 24:439-442
89. Badia JM, Whawell SA, Scott-Coombes OM et al (1996) Peritoneal and systemic cytokine
response to laparotomy. Br J Surg 83:347-348
90. Iscra F, Antonaglia V, Guglielmi 0, Gullo A (1997) Prognostic value of IL-6, TNF and CRP
in SEPSIS/SIRS patients. Int Care Med 23 [Suppl 1]:75
91. Wildling E, Pusch F, Aichelburg A et al (1997) Procalcitonin is elevated in patients after
severe injury. Int Care Med 23[Suppll]:62
92. de Werra I, Jaccard C, Corradin SB et al (1997) Cytokines, nitrite/nitrate, S TNFr, and PCT
concentration: comparisons in patients with septic shock, cardiogenic shock and bacterial
pneumonia. Crit Care Med 25 :607-613
93. Baxby 0, van Saene HKF, Stoutenbeek CP, Zandstra OF (1996) Selective decontamination of
the digestive tract: 13 years on, what it is and what it is not. Intensive Care Med 22:699-706
94. van Saene HKF, Stoutenbeek CP, Zandstra OF, Fox MA (1998) Prevention oflCU infections.
In: van Saene HKF, Silvestri L, de la Cal MA (eds) Infection control in the intensive care
unit. Springer-Verlag Italia, Milano, pp 225-238
95. Silvestri L, van Saene HKF, Gullo A (1997) Selective decontamination of the digestive tract
in critically ill patients: a pathogenesis-based infection control method. Care Crit III 13:
112-117
96. Brun-Buisson C, Legrand P, Rauss A et al (1989) Intestinal decontamination for control of
nosocomial multi-resistant Gram-negative bacilli. Ann Intern Med 110:873-881
97. Silvestri L (1991) Selective decontamination of the digestive tract. Is it a strategy? In: Gullo
A (ed) Recent Advances in Anaesthesia, Pain, Intensive Care and Emergency, APICE, Tri-
este, vol 6, pp 581-595
98. Potgieter PO, Hammond JM (1995) Prophylactic use of the new quinolones for the preven-
tion of nosocomial infection in the intensive care unit. Drugs 49 [Suppl 2]:86-91
99. van Saene JJM, Stoutenbeek CP, van Saene HKF et al (1996) Reduction of the intestinal
endotoxin pool by three different SOD regimens in human volunteers. J Endotox Res 3:
337-343
100. Damjanovic V, van Saene HKF (1998) The value of surveillance cultures in neonatal inten-
sive care units. In: van Saene HKF, Silvestri L, de la Cal MA (eds) Infection control in the
Intensive Care Unit. In: Infection control in the Intensive Care Unit. Springer-Verlag Italia,
Milano, pp 316-238
101. van Saene HKF, Baxby 0, Stoutenbeek CP, Zandstra OF (1993) Selective decontamination of
the digestive tract (SOD). Does it improve outcome? Does it lead to resistance? In: Gullo A
(ed) Recent Advances in Anaesthesia, Pain, Intensive Care and Emergency, APICE, Trieste,
pp 395-405
102. Selective decontamination of the digestive tract trialists collaborative group (1993) Meta-
analysis of randomised controlled trials of selective decontamination of the digestive tract. Br
Med J 307:525-532
103. Heyland OK, Cook OJ, Jaeschke R et al (1994) Selective decontamination of the digestive
tract: an overview. Chest 105: 1221-1229
104. Liberati A, D' Amico R, Pifferi S et al (1997) Antibiotic prophylaxis in adult patients treated
in Intensive Care Units. The Cochrane Library, issue n 3. Abstract available from: http://
www.hcn.net.aulcochrane/abstracts/abOOOOOO.htm
105. Unertl KE, Lenhart F-P, Holzel C, Ruckdeschel G (1992) Selective digestive decontamina-
tion in ICU patients. Clinical results in trauma and general ICU patients. Rean Urg 1[3 bis]:
516-520
106. Hurley JC (1995) Prophylaxis with enteral antibiotics in ventilated patients: selective decont-
amination or selective cross-infection? Antimicrob Agents Chemother 39:941-947
107. Tetteroo GWM, Wagenvoort JHT, Mulder PGH et al (1993) Decreased mortality rate and
length of hospital stay in surgical intensive care unit patients with successful selective decon-
tamination of the gut. Crit Care Med 21: 1692-1698
108. Bion JF, Badger I, Crosby HA et al (1994) Selective decontamination of the digestive tract
reduces Gram-negative pulmonary colonization but not systemic endotoxaemia in patients
undergoing elective liver transplantation. Crit Care Med 22:40-49
109. Schardey HM, Joosten U, Finke U et al (1997) The prevention of anastomotic leakage after
gastrectomy with local decontamination. A prospective, randomized, double-blind placebo-
controlled multicenter trial. Ann Surg 225: 172-180
110. Steiberg D (1979) On leaving the peritoneal cavity open in acute generalized suppurative
peritonitis. Am J Surg 137-216-220
111. Duff J, Moffat J (1981) Abdominal sepsis managed by leaving abdomen open. Surgery 90:
774-778
112. Mughal MM, Bancewicz J, Irving MH (1986) "Laparostomy": a technique for the manage-
ment of intra-abdominal sepsis. Br J Surg 73:253-259
113. Anderson ED, Mandelbaum DM, Ellison EC et al (1983) Open packing of the peritoneal cav-
ity in generalized bacterial peritonitis. Am J Surg 145: 131-135
114. Wittman DH, Aprahamian C, Bergstein JM (1990) Etappenlavage: advanced diffuse peritoni-
tis managed by planned multiple laparotomies utilizing zippers, slide fastener, and velcro
analogue for temporary abdominal closure. World J Surg 14:218-226
INDEX
Ilz-antiplasmin 105 hemofiltration 174
abdominal heparin 106
-pain 683 hepatic 54,60,372
-X-ray 687 - system 58
acidosis hepatorenal syndrome 134
hepatic - 372
metabolic - 382, 388, 702 hydroxyethlystarch 86
acute necrotizing myopathy 160 hypertension 17,47,61,
angiography 138 ibuprofen 81
antibiotics 147 immunosuppression 146
APACHEll 35 infection 5,17,26,115,116,145,167
bacteremia 15,17,19 inflammation 26, 116, 133, 145
beta 2 integrins 75 injury 26, 145
bradycardia 61 interleukin-l beta 73
Brussels' score 46 interleukin-8 74
cachectic myopathy 160 interleukin-lO 74
cardiac output 91 intracranial pressure monitoring 66
cardiovascular 54, 60 intubation 66
- system 58 ionotropes 146
catecholamines 128 ischemia 133
ceramide 78 laparoscopy 131
coagulation 99, 100 laparotomy 136
disseminated intravascular - 99 leukocytes 74
creatinine 61 lipopolysaccharide 28
cytokines 115,116,161,175 logistic organ dysfunction (LOD) system 54
dibenzyline 145 lymphopenia 78
diuretics 146
mediators 27, 118, 167
dobutamine 93,125
metabolic 382, 388, 702
dopamine 94, 123
mortality 19
dopexamine 94, 126
mUltiple organ dysfunction 62, 167
E selectin 75 - syndrome 17
elastase 28 multiple organ failure 20, 68
endotoxin 27,28,81,100,101 necrosis 133
epidemiology 15 neurologic 54, 60
epinephrine 94 - system 58
fibrinolysis 99,101 neuropathy 157
foal-directed therapy 93 norepinephrine 94, 124
fram-negative bacteria 18 oliguria 61
fram-positive 18 operation 26
gangrene 133 osmotic agents 66
Glasgow coma score 46, 68 oxidants 28
Guillain-Barre syndrome 158 oxygen
hematologic 54 - consumption 91
- system 58 - delivery 91
192 Index
- delivery 91 septic
- flux test 91 - encephalopathy 156
P se1ectin 75 - shock 17,91, 167
pentoxifylline 102 splanchnic circulation 123
peritonitis 138 syndrome
Guillain-Barre - 158
platelet activating factor 116
hepatorenal - 134
pneumonia 19 multiple organ dysfunction - 17
procalcitonin 119 systemic inflammatory response - 16,
prostaglandin 86 167
protein C 100 system
pulmonary 54 cardiovascular - 58
- system 58 hematologic - 58
hepatic - 58
purpura fulminans 99 logistic organ dysfunction - 54
relaparotomy 140 neurologic - 58
renal 54,60 pulmonary - 58
- system 58 renal- 58
reperfusion 28 systemic inflammatory response syndrome
score 65,68 16, 167
selectin tachycardia 61
E-75 thick filament myopathy 160
P-75 tumor necrosis factor a 73
selective decontamination of the digestive .trauma and coagulopathy 136
tract 179
triage 66
sepsis 15, 17,23,26,81,91,99, 106, 115,
133, 155, 167 tuberculosis 146
- severity score 53 ventilation 66

Mods Score

Uploaded by

Mods Score

Uploaded by

Sepsis and Organ Dysfunction

Epidemiology and Scoring Systems

ORGAN FAILURE ACADEMY

O.F.A. - ORGAN FAILURE ACADEMY, VIA BATTISTI, 1 - 34125 TRIESTE (ITALY)

© Springer-Verlag Italia, Milano 1998

ISBN 978-88-470-0297-5 ISBN 978-88-470-2271-3 (eBook)

Trieste. 19th November 1997

EPIDEMIOLOGY AND CLINICAL COURSE OF SEPSIS AND MODS

Epidemiology and Clinical Course of Sepsis

The Complexities of Sepsis and Organ Dysfunction

An Overview to Introduce Prognostic Indexes in MODS

Multiple Organ Dysfunction (MOD) Score

The Logistic Organ Dysfunction (LOD) System

MEDIATORS AND PROSTAGLANDINS

Modulating Prostaglandin Metabolism in Sepsis

Oxygen Delivery and Consumption Relationships in Sepsis

Disturbances and Modulators of Coagulation and Fibrinolysis in Sepsis

EXPERIMENTAL AND CLINICAL APPROACH

Therapies Directed against TNF-a and IL-t during Sepsis

Effects of Adrenergic Agents on the Hepato-Splanchnic Circulation: An Update

When to Operate or to Stop Operating and to Plan a Reoperation

Multiple Therapeutic Agents - Are We Making Progress?

SIRS and Sepsis-Induced Neuromuscular Dysfunctions

PREVENTION AND MANAGEMENT OF SEPSIS AND MODS

Index ........................................................................................................................................... 189

AIS, abbreviated injury score HTI, hospital trauma index

CRAMS, circulation, respiration, abdomen, MOF, multiple organ failure

GM-CSF, granulocyte macrophage-colony PODS, probability of death score

PTS, poly trauma score TAT, thrombin-antithrombin

Sepsis can be broadly defined as the systemic inflammatory response to infec-

Table 1. Examples of various definitions or criteria for diagnosis of sepsis

For standardizing the terminology, a North American consensus conference

Table 2. The definitions of ACCP/SCCM consensus conference

Infection Microbial phenomenon characterized by an inflammatory response to the

used in cases of an evident systemic response in the absence of a verified infec-

Clinical course of sepsis

Table 3. Attack rates for end-organ dysfunction

In summary, sepsis is an important reason for morbidity and mortality in

Inflammation in itself is not to be considered as a

Table 1. Complexities (and confusion) about...

Sepsis - What is it? And how do you treat it?

Table 2. Early treatment of SIRS

Table 3. MOF, MODS, SIRS and now ...

CARS Compensatory, antiinflammatory response syndrome

under some generalized description is not going to allow us to treat patients

Basic antiinflammatory response (BARS)

Biologic conundrums - Puzzles or paradoxes?

Causes of organ dysfunctionlfailure

Table 5. Biologic conundmms

Anti-ll-6 mab given to endotoxemic mice yields increased IL-6

Table 6. Cause of organ dysfunction/failure

Ischemia - Dysoxia (Hypoxia)

Predictions are trickly

Our purpose in this symposium is to seek a consensus on scoring or to try to de-

Table 1. History of scoring

Game scores MOFscores

Table 2. Scoring systems

Health Status Evaluation

Table 3. Sepsis scoring

Sepsis Severity Score (SSS) - Stevens

Table 4. Severity of illness scoring

Table 5. Injury severity scoring

ISS Injury Severity Score

Table 6. Scoring systems for health status evaluation

SIP - Sickness Impact Profile (general health status)

Table 7. Injured extremity scores

MESI Mangled Extremity Syndrome Index

Table 8. Futility score

Any fatal illness with a life expectancy of six months or less

Methodologic principles of the MOD score

Table 1. The multiple organ dysfunction (MOD) score

Organ dysfunction scales: unresolved questions

each of the variables are recorded without consideration of therapeutic interven-

When should variables be measured?

How are variables selected and calibrated?

How is cardiovascular dysfunction measured?