B. Eisele M. Lamy L. G. Thijs H.-O.Keinecke H.-P. Schuster F. R. Matthias E Fourrier H. Heinrichs U.Delvos
Antithrombin III in patients with severe sepsis
A randomized, placebo-controlled, double-blind multicenter trial plus a meta-analysis on all randomized, placebo-controlled, double-blind trials with antithrombin III in severe sepsis
Received: 4 June 1997 Accepted: 24 April 1998
B. Eisele ( 9 ) . H. Heinrichs. U. Delvos Clinical Research Department, Centeon Pharma GmbH, Marburg, Germany
M. Lamy Centre Hospitalier Universitaire de Li6ge, Li6ge, Belgium
L. G. Thijs Medical Intensive Care Unit, Free University of Amsterdam, Amsterdam, The Netherlands
Abstract Objectives: To evaluate the safety and potential efficacy of antithrombin III (AT Ill) in reduc- ing mortality in patients with severe sepsis. Design: Prospective, randomized, placebo-controlled, double-blind, phase II, multicenter, multinational clinical trial. Setting: Seven academic medical center intensive care units (ICU) in Belgium, Denmark, the Nether- lands, Norway and Sweden. Patients: 42 patients with severe sepsis who received standard sup- portive care and antimicrobial ther- apy, in addition to the administra- tion of AT ]II or placebo. Interventions: Patients received ei- ther an intravenous loading dose of 3000 IU AT III followed by a main- tenance dose of 1500 IU every 12 h for 5 days or equivalent amounts of placebo. Measurements and results: All pati- ents were evaluated for safety and for 30-day all-cause mortality. Conclusions: The administration of AT III was safe and well-tolerated. It was followed by a 39 % reduction in 30-day all-cause mortality (NS). The reduction in mortality was ac- companied by a considerably short- er stay in the ICU. Patients treated with AT III exhibited a better per- formance in overall severity of ill-
ness and organ failure scores (Acute Physiology and Chronic Health Evaluation II, multiple organ fail- ure, organ system failure), which was noticeable soon after initiation of treatment. Patients treated with AT III demonstrated a better reso- lution of pre-existing organ failures and a lower incidence of new organ failures during the observation peri- od. A meta-analysis comprising this and two other double-blind, place- bo-controlled trials with AT III with a total of 122 patients suffering from severe sepsis confirms the positive trend. The results of the meta-anal- ysis demonstrate a 22.9 % reduction in 30-day all-cause mortality in pati- ents treated with AT III. Although still too small to be confirmative, the meta-analysis clearly points to the fact that a sufficiently powered phase III trial is warranted to prove whether AT III has a beneficial role in the treatment of severe sepsis.
Key words Anti thrombin III • Sepsis • Multiorgan failure • Clinical trial • Meta-analysis
664
Introduction
Sepsis describes a constellation of maladaptive respons- es to infection often including hypotension and organ failure [1]. Despite major advances in antimicrobial therapy, surgical techniques, and intensive care, there has been little improvement in morbidity and mortality due to sepsis during the past 30 years. The average mor- tality from severe sepsis remains at approximately 35-45 %, and the incidence of sepsis continues to in- crease [2, 3].
For quite a long time, our understanding of the pathophysiology of sepsis was clearly focused on endo- toxin and proinflammatory cytokines. Large random- ized placebo-controlled trials in patients with sepsis, however, failed to demonstrate the clinical benefit of a specific blockade of endotoxin, tumor necrosis factor-al- pha (TNF a) or interleukin-1 (IL-1) ~ alone [3-9].
It is now appreciated that many signs and symptoms of sepsis are not the direct effects of cytokines but are transmitted through other mediator systems [10, 11]. When the septic process progresses to malperfusion and organ failure, the unbalanced - and cytokine-medi- ated - activation of the coagulation system comes into play [12-16].
Antithrombin III (AT III) is an important physiolog- ical regulator of blood coagulation as it inhibits the clot- ting process at various levels. Its inhibitory profile in- cludes the intrinsic (F XIa, F IXa), extrinsic (tissue fac- tor bound F VIIa), and common pathway (F Xa, throm- bin) of coagulation [17]. AT III is a single-chain glyco- protein which is produced by the liver. It belongs to the SERPIN (serine protease inhibitor) superfamily. The AT III molecule contains two main functional domains: a reactive site domain binding to the target proteases and a heparin-binding domain. The inactivation of tar- get proteases (e. g., thrombin) occurs through a stoichio- metric complex formation between AT III and the en- zyme. This reaction is then accelerated several hundred times by complex formation with heparin.
In healthy individuals, its half-life ranges from 18 to 27 h but may be markedly reduced in patients with se- vere sepsis [18], even in the absence of clinically mani- fest defibrinating disseminated intravascular coagula- tion [19]. The acute consumption of the molecule during the process of coagulation, its increased leakage from the intravascular compartment, and the proteolytic in- activation by elastase are the pathophysiologic mecha- nisms responsible for the acquired AT III deficiency in these patients [20, 21].
The decrease in AT III is related to outcome, and lev- els are significantly lower in nonsurvivors than in survi- vors [18]. A decrease in AT Ill levels below 50 % was demonstrated to be a good prognostic predictor of sub- sequent death with a sensitivity of 96 % and a specificity of 76 % [151.
The first administration of AT III in patients with sepsis was reported in 1978 [22]. Subsequently, many other clinical studies were published. Yet, despite the fact that the therapeutic administration of AT III was generally found to be beneficial, the open, and in some cases nonrandomized, design of the trials limited the level of evidence for demonstration of treatment effica- cy.
In the present study, the safety and potential efficacy of AT III in reducing mortality in patients with severe sepsis was assessed in a randomized, double-blind, pla- cebo-controlled, multinational setting. The study com- prised 42 patients suffering from severe sepsis. The pri- mary endpoint was comparison of 30-day all-cause mor- tality for AT III relative to placebo. Seven academic medical centers in Northern and Western Europe par- ticipated in this trial. The study protocol was approved by the ethical review committee at each participating center, and informed consent for participation was ob- tained for all patients (either from the patients them- selves or from relatives, if a patient was not able to give informed consent due to the severity of the sepsis) prior to the performance of any study-related procedures.
In order to base the evaluation on a larger number of patients, a meta-analysis comprising 122 patients was also made. The analysis combines the results of the pre- sent study with those of two other double-blind, place- bo-controlled clinical trials with AT III in patients with severe sepsis. The results of both these trials were pre- sented recently [23, 24].
Patients and methods
Study design and treatment
The study was designed as a phase II pilot trial not aiming at statis- tical significance for the main criterion. It consisted of three phases: screening, treatment, and follow-up. During the screening phase, patient eligibility was determined, informed consent was obtained, and patients were randomly assigned to receive either AT III (Ky- bernin R Centeon Pharma GmbH, Marburg, Germany) or placebo. During the treatment phase, all patients received an intravenous loading dose of either AT III 3000 IU, or an equivalent volume of placebo, administered over approximately 1 h. Subsequently, AT III (1500 IU) or placebo was infused every 12 h for 5 days. Thus, the cumulative dose of AT Ill was 18000 IU. Decisions regarding the use of antimicrobial agents, intravenous fluids, cardiovascular and respiratory support, and surgical intervention were made by each patient's attending physician and were not dictated by the study protocol. After completion of the 5-day treatment phase, pa- tients were followed-up for the remainder of 30 days.
Patient selection
Patients with a diagnosis of sepsis were enrolled if they fulfilled the following criteria within a 6-h time window prior to initiation of treatment with the study medication: (a) clinical evidence of sepsis
665
with a suspected source of infection, (b) temperature > 38.5 ° C (rectally) or <35.5°C (rectally), (c)leukocyte count > 10000/ mm 3 or < 3500/mm 3. Furthermore, three of the following six signs for sepsis had to be fulfilled: tachycardia > 100 beats/min, tachyp- nea > 24 breaths/rain or mechanical ventilation because of septic indication, hypotension with systolic blood pressure < 90 mmHg when no vasoactive agents or fluid replacement was administered, thrombocytopenia with a platelet count < 100000/mm 3, elevated lactate level > 1.6 mmol/1, oliguria with urine output < 20 ml/h.
Patients were not eligible if they met any of the following crite- ria: age under 18 or over 80 years; need for anticoagulant therapy other than subcutaneous low-dose heparin; incurable malignancies with documented metastases or hematologic neoplasia except when in complete remission; chronic treatment with high-dose im- munosuppressive drugs or high dose nonsteroid anti-inflammatory drugs within the previous 2 days; known bleeding disorders; ongo- ing massive surgical bleeding; multiple organ failure with at least four organs involved existing for more than 24 h when the patient was checked for eligibility; acute myocardial infarction; chronic compensated organ dysfunction, such as chronic liver disease, dial- ysis-dependent renal failure, moderate to severe chronic heart fail- ure (New York Heart Association 3-4) or pre-existing complicated diabetes mellitus or severe obstructive pulmonary disease; burn patients; transplant patients; AIDS; severe cranial trauma and/or pathologic changes on computed tomography and Glasgow Coma Scale < 6; pregnancy; treatment with plasma expanders (Rheomac- rodex or HAES); prior enrollment in this clinical trial or enroll- ment in another clinical trial within the previous 30 days.
Patient evaluation
Patients were followed up throughout the 30-day study period af- ter receipt of study medication or until death, whichever occurred first. During screening, the patient's history was recorded and a physical evaluation was performed. Samples of blood and other body fluids and specimens from suspected sites of infection were obtained for culture. Vital signs were obtained and hematologic and biochemical tests carried out at screening and on days 1, 2, 3, 7, 10, 20, and 30. These laboratory values and organ-specific pa- rameters were also used to calculate the Acute Physiology and Chronic Health Evaluation (APACHE) II [25], Multiple organ failure (MOF) [26], and organ system failure (OSF) [27] scores.
Special coagulation variables
Additional blood samples for measurement of special coagulation variables (e. g., AT III activity) were taken before the first adminis- tration of the study drug and every 12 h for the following 5 days (always before the new administration of the study drug) and once on days 6 and 7.
Definitions and criteria of organ dysfunction
Respiratory dysfunction was defined according to the OSF score as the presence of a respiratory rate _< 5/min or _> 49/min or an arte- rial carbon dioxide tension _> 50 mmHg or an alveolar-arterial oxygen difference _> 350 mm Hg or (from day 4 on) a need for me- chanical ventilation. Renal dysfunction was defined according to the OSF score as a urinary output < 480 ml/24 h or < 160 ml/8 h or serum blood urea nitrogen _> 100 rag/100 ml or serum creatinine ___ 3.5 mg/100 ml. Hepatic dysfunction was defined according to MOF score grade 2 as glutamic oxaloacetic transaminase _> 50
U/1 or bilirubin _> 6 mg/100 ml. Central nervous system dysfunc- tion was defined according to MOF score grade 2 as an acute dete- rioration in mental status with severely disturbed responsiveness. Dysfunction of the coagulation system was defined according the the Jordan score [28], as the presence of two or more of the follow- ing criteria: platelet count < 60000/mm 3, prothrombin time > 18 s or requirement for 4 U of fresh frozen plasma (FFP), fibrinogen < 150 mg/100 ml, fibrin(ogen) split products > 16 dilutions.
Statistical analysis
The primary analysis was based on the 30-day all-cause mortality for all randomized patients (intention-to-treat analysis). APACHE II, MOF, and OSF scores were calculated at screening on days 1, 2, 3, 7, 10, 20, and 30. Numeric data on scores, AT III levels, and amounts of FFP administered are represented as mean and stan- dard deviation (SD). Survival rates over time and rates of dis- charge from the intensive care unit (ICU) over time were estimat- ed by the Kaplan - Meier method. The following exploratory tests were performed: t-tests for continuous variables, Fisher's exact test for binary variables, and a generalized Wilcoxon test for the duration of stay on the ICU; tests were indicated as "significant" if p was < 0.05, two-sided. Due to the small number of patients in this pilot trial the statistical power was only adequate for large dif- ferences between treatment groups.
Meta-analysis
Mantel - Haenszel estimates of the risk ratio (RR) of mortality in the antithrombin III group relative to the placebo group and corre- sponding 95 % confidence intervals were calculated per study and stratified over all studies according to Kleinbaum et al. [29]. The RRs and the confidence intervals were transformed into the "change of risk": (1-RR) *100%. All confidence intervals were represented together in one figure following Galbraith [30].
For the meta-analysis, clinical trials were selected according to the following criteria: Indication: severe sepsis (criterion for sever- ity: sepsis-induced circulatory and/or organ failure); design: pro- spective, double-blind, randomized, placebo-controlled; treatment group: antithrombin III concentrate; control group: placebo; the literature search for sepsis trials with AT III was performed with the help of the CD-ROM "Sepsis in review" published by the Eu- ropean Society of Intensive Care Medicine.
Results
For ty - two pat ients with severe sepsis were r a n d o m i z e d to receive active t r e a t m e n t or placebo. A t screening, the fol lowing data were recorded: m e a n age 57.2 (+ 14.3)years , m e a n A P A C H E II score 14.6 (+ 5.5); 60 % of all pa t ien ts had p r o v e n bac t e r emia on admis- sion, and 79 % of all pa t ien ts were hypotens ive . This dis- t r ibu t ion , as well as the demograph ic character is t ics and unde r ly ing diseases (Table 1), the pa thogens responsi- ble for the septic episode (Table 2), and the d i s t r ibu t ion of the severi ty of sepsis at s tudy en t ry (Table 3) were we l l -ba lanced across the two t r e a t m e n t groups. The re were no signif icant differences (i.e., p > 0.05), Whereas base l ine A P A C H E II and M O F scores favoured the
666
Table 1 Demographic characteristics and underlying diseases by treatment group
AT III group, the baseline OSF score favoured the pla- cebo group (Table 4).
Effects of AT III in patients with severe sepsis
The effect of AT III t reatment on the mortality of pati- ents with severe sepsis was analyzed on an intention-to-
treat basis. There were 9 deaths among the 22 recipients of the placebo (41% mortality), compared with 5 deaths among 20 patients receiving AT III (25% mortality); the 95 % confidence intervals were 21 to 64 % (placebo) and 9 to 49 % (AT Ill). Compared with placebo, the mortality was reduced by 39 % in patients treated with AT III (NS) The Kaplan - Meier survival curves are shown in Fig. 1.
On admission, endogenous activities of AT III were 45.7 + 14.4% (AT III) and 49.0 + 19.1% (placebo). Thus, in both groups AT III activity was clearly below the range seen in healthy individuals (70-130 % ). In pa- tients treated with placebo, AT III activities remained low, whereas in AT III-treated patients, there was an in- crease and from day 4 on they were within the normal range, i. e., > 70 % (Fig. 2). On day 4 (i. e., 96 h after ran- domization) the difference was significant (p < 0.01). At 48, 60, 108, and 120 h after randomizat ionp values were found to be < 0.1. For all other measurements the differ- ences were not significant.
Resolution of pre-existing organ dysfunction (Table 5)
Acute Neurological dysfunction: 16 patients (AT III 6, placebo 10) already showed neurological dysfunction on admission. A reduction in the prevalence of central nervous system (CNS) dysfunction by at least 50 % was recorded on day 20 in the placebo group. In the AT III group, the corresponding reduction was 40 %.
Acute respiratory dysfunction: 14 patients (AT III 8, placebo 6) were in respiratory failure on admission. A reduction in the prevalence of respiratory dysfunction by at least 50 % was noted on day 2 in the AT I l l group and on day 3 in the placebo group.
Acute renal dysfunction: 16 patients (AT III 8, place- bo 8) had renal dysfunction on admission. A reduction in its prevalence by at least 50 % was seen on day i in the AT III group and on day 7 in the placebo group.
Acute hepatic dysfunction: 13 patients (AT III 5, pla- cebo 8) had hepatic failure on admission. In the AT III group the prevalence of hepatic dysfunction was re- duced by at least 50 % on day 3. In the placebo group, the corresponding reduction came to 38 %.
Coagulation failure: 5 patients (AT III 3, placebo 2) had coagulation failure on admission. A reduction in its prevalence by at least 50 % was recorded on day 2 in the AT III group and on day 1 in the placebo group.
Incidence of organ dysfunction (Table 6 and Fig. 3)
Acute neurological dysfunction: 8 of 24 patients without neurological dysfunction on admission experienced it during the observation period, 2 of 13 (15 %) in the AT III group and 6 of 11 (55 %) in the placebo group.
667
100
80
> ,
o 60 t -
4o
rr 2O
~.~2-~-~-- I c'--:~____~--_l " Antithrombin III (n = 20)
i
~ - . . . . . . . . . . . "O"O
Placebo (n = 22)
i , . . . . . . . . i . . . . . ~ , , r I ' ' ' ' ' ' ' ' ' I '
0 10 20 30 Surv iva l (days)
Fig.1 Survival time after admission to study; intention-to-treat analysis; Kaplan Meier estimation
Acute respiratory dysfunction: 21 of 27 patients (78 %) had an episode of acute respiratory dysfunction during the observation period, 7 of 11 (64 %) in the AT III group and 14 of 16 (88 %) in the placebo group.
Acute renal dysfunction: the incidence was 10 out of 25 patients (40%), with 3 of 11 (27%) in the AT III group and 7 of 14 (50 %) in the placebo group.
Acute hepatic dysfunction: the incidence was 12 of 26patients (46%), with 3 of 12 (25%) in the AT III group and 9 of 14 (64 %) in the placebo group.
Coagulation failure: the incidence was 7 of 35 pati- ents (20 %), with 4 of 15 (27 %) in the AT III group and 3 of 20 (15 %) in the placebo group.
Length of stay on ICU
Fourteen patients died during the observation period (AT III 5 patients, placebo 9 patients). Those patients who survived left the ICU earlier if treated with AT III (p = 0.09, generalized Wilcoxon test). Altogether, 50 % of the survivors in the AT [II group were discharged be- fore day 18, whereas only 16 % of the survivors in the placebo group were discharged before that day (and only 38 % before day 30).
%
100 -
9 0 -
8 0 -
70
6 0 -
5 0 -
4 0 -
3 0 -
0 ~
3,000 I.U, 1.500 I.U. each 2 = 18,000 I.U.
it tt • AT III group • Pacebo group * p < 0.01
On 24- h 48 h 7: ~ h 96 h 12(3 h 144- h admission
Fig.2 AT III activity (% of normal; mean ± SD) before, during, and after treatment with AT III concentrate or placebo. (AT III was measured after an interval of 12 h after the previous adminis- tration, i. e. just before the next dose was given)
Safety
Treatment with AT III was found to be well-tolerated and safe. There were four adverse events in the placebo group and none in the AT III group considered to be re- lated to the study drug. They were rash (1) and hemor- rhage (3). On admission, mean hemoglobin levels were 8.33 + 2.39 g/dl (AT III group) and 7.86 + 1.85 g/dl (pla- cebo group). There was no significant decrease during the course of the study. Eight patients (40 %) in the AT III group and 10 patients (45 %) in the placebo group received FFR In AT III-treated patients, the mean total dose per patient was 2821 + 4168 ml and in placebo- treated patients it was 1640 + 1475 ml. The difference was not statistically significant.
Discussion
Although too small to be confirmative, the results of this phase II trial with AT III in patients with severe sepsis
Table 4 APACHE II, MOF, and OSF scores over treatment/observation period
Score Treatment On Day l Day 2 Day 3 Day 7 Day 30 group admission
No. of A T III 0 1 1 2 3 5 pa t i en t s w h o P lacebo 0 0 0 1 2 9 d ied be fore
a Pa t i en t s w h o died are a s s u m e d no t to have r ecove red f r om o rgan failure. Pe r cen t ages are ca lcu la ted exc luding miss ing da ta b Miss ing da ta
Table 6 Inc idence of o rgan dys func t ion in the N & W E u r o p e a n s tudy
Inc idence of: A T III P lacebo
A c u t e neuro log ica l dys func t ion 2/13 6/11
A c u t e resp i ra tory dys func t ion 7/11 14/16
A c u t e rena l dys func t ion 3/11 7/14
A c u t e hepa t ic dys func t ion 3/12 9/14
Coagu la t i on fai lure 4/15 3/20
demonstrate a 39 % reduction in 30-day all-cause mor- tality. The survival benefit provided by treatment with AT III appeared to occur late in the septic process. It started within the second week after initiation of treat- ment and was most pronounced during the third week.
There was a trend to clinical benefit in the AT III group with respect to the respiratory, renal, and liver systems. In AT III-treated patients, there was a better
resolution of these pre-existing organ failures, as well as a lower incidence of new ones. There was no advan- tage in the AT III group with respect to the coagulation system (see below). A positive trend in the AT III group could also be seen in the patients' cerebral function. It remains open, however, whether the possible improve- ment in CNS function is a directly mediated beneficial effect of the drug or rather an indirect consequence of the better performance of the patients' respiratory, kid- ney, and liver systems.
The trend for a better performance of individual or- gan systems in patients treated with AT III was also re- flected in overall organ failure scores. The improvement in APACHE II, MOF, and OSF scores started soon af- ter initiation of treatment (Table 4).
The beneficial and potentially life-saving effect of AT III was not obtained at the expense of a long and costly need for intensive care treatment. In patients treated with AT III, the stay in the ICU was consider- ably shorter than for patients treated with placebo.
On admission, patients had markedly reduced plasma levels of AT III. During the treatment period with AT III concentrate or placebo, AT III levels remained low in
669
Acute neurological dysfunction
Acute hepatic dysfunction
Acute renal dysfunction
Acute respirator}
~ 1 5 % 55%
;rl I t 25°/° 64 %
! . . . . . :]27% 5O %
.............. :,!.i, i~,~ii,.i i ......... i ] 64 % dysfunction 88 %
C o a g u l a t i o n ~ 27 % failure ! 15,% ,
0 10 20 30 40 5'0 6'0 7'0 8'0 9'0 lC)0 %
[ ] Antithrombin III • Placebo
Fig.3 Incidence of organ dysfunction after start of treatment with study medication
placebo patients and increased to normal levels in pa- tients treated with AT III. Nevertheless, the increase of AT III activity in AT III-treated patients was more pro- tracted and lower than expected. Due to the fact that AT III levels were measured after an interval of 12 h after the previous administration (i. e., just before the next dose was given), the relatively low increase in AT III lev- els despite high-dose replacement therapy could repre- sent the net result of the following processes: degrada- tion by target proteases (e. g., thrombin) or by nontarget proteases (e.g., elastase, metalloproteases), and loss from circulation due to capillary leak syndrome [21, 31].
It appears to be difficult to definitely unravel the mechanism by which the beneficial effect of AT III was brought about. For quite a long time the proposed mode of action of AT III in patients with sepsis was re- stricted to the drug's anticoagulant activity. The results of this trial, however, do not support an explanation pri- marily based on the anticoagulant effect of AT III with- in the fluid phase. In fact, there was a slight trend fa- vouring the placebo group but both the prevalence and the incidence of coagulation failure or clinically overt disseminated intravascular coagulation were too low to allow any definite conclusion. Likewise, whether or not there is a beneficial effect of AT III in the prevention of fibrin formation in the microvasculature also remains open. As this process is thought to be involved in the pathogenesis of multiple organ failure [15, 31-33], the better performance of organ systems in AT III-treated patients would be in accordance with this assumption.
Recently, evidence has been accumulating to show that there is an additional anti-inflammatory potential of AT III besides the drug's anticoagulant properties. Studies in nonhuman primates given a lethal dose of liv- ing Escherichia coli have demonstrated that infusion of AT III in a high-dose regimen reduced the intensity of the coagulopathic and cell injury responses and prevent- ed death [34]. Yet, in the same animal model, infusion of
DEGR-F Xa (i. e., F Xa blocked in the active center competing with naturally occurring F Xa for binding to F Va) did not prevent death, although the coagulant re- sponse to the E. coli infusion was also completely pre- vented [35]. This indicates that the life-saving effect of AT III in this animal model might have been brought about by some mechanism other than one blocking fi- brin formation.
It has been demonstrated that AT III is able to bind to the glycosaminoglycans (GAGs) of the endothelial cell via its heparin-binding region. This binding to GAGs is followed by the release of prostacyclin (PGI2). The release of PGI 2 in the microcirculation has marked anti-inflammatory effects, as it reduces the re- lease of oxygen radicals from activated neutrophils and of TNF from activated monocytes. Furthermore, it re- duces platelet aggregability [36-38]. The latter effect es- pecially could be expected to further improve oxygen delivery in the organs' microcirculation.
Meta analysis
There have been numerous reports on uncontrolled as well as controlled studies with AT III in severe sepsis, most of them demonstrating a possible beneficial effect of the drug in patients with sepsis and septic shock [21, 39]. Only two, however, were carried out under a random- ized, double-blind, placebo-controlled protocol (Schus- ter & Matthias, 1995: "German study" [24]; Fourrier et al. 1993: "French study" [23]. Both recruited a study pop- ulation similar to the present one - that is, patients suffer- ing from severe sepsis and organ failure with obligatory [23] or facultative [24] signs of disseminated intravascular coagulation and shock. In the German study, 45 patients received 3000 IU AT III or placebo as loading dose, fol- lowed by 500 IU every 4 h over 7 days. Both treatment groups additionally received low-dose i.v. heparin at 6 IU/kg bodyweight and hourly during the 7-day treat- ment period. The study resulted in a 14 % lower 30-day all-cause mortality in the AT III group (NS) [24].
In the French study, 35 patients received 90-120 IU/ kg b.w. AT III or placebo as loading infusion, followed by a continuous daily infusion of 90-120 IU/kg b.w. AT III or placebo over 4 days. In the primary analysis, there was a 42 % reduction in the 28-day all-cause mortality in the AT III group (NS) [23]. This analysis did not include 3 patients in the AT III group who suffered from a rap- idly fatal disease or from intractable rapidly fatal shock and who were therefore excluded (in a blinded fashion) from the analysis. In previous studies of immunotherapy for sepsis, an improved outcome was always found, if the analyses were based on some subsets of patients who were analyzed post hoc. For that reason, we decid- ed to include these patients in the meta-analysis, thus basing the analysis on the intention-to-treat population.
670
N & W Europe ,L (n = 42) T
France (n = 35}
Germany (n = 45)
+
1 T
Metaanalysis (n = 122)
-'80 ; ' . . . . < 0 -40 - 0 0 +25 +67 +150 >+400
Risk ratio change (%)
Fig.4 Meta-analysis criteria: patients with severe sepsis; random- ized, double-blind, placebo-controlled trials: asymptomatic confi- dence interval for Mantel-Haenszel estimates
100 -
8 0 -
=~ 60
12r
# 4o
n- 20
Antithrombin III (n = 60)
u o-o-o
Placebo (n = 62)
i , , r , , , , , , i , , , , i , , , , [ , , , , , E , , , i ,
0 10 20 30 Surv ival (days)
Fig.5 Meta-analysis: survival time after admission to study; Inten- tion-to-treat analysis; Kaplan Meier estimation
Results of the meta-analysis
The meta-analysis comprised 122 patients (AT III 60, placebo 62). Mean age was 54+16yea r s . Mean A P A C H E II score on admission was 13 points (AT III 11.5, placebo 14) in the German and Northern and West- ern European study. In the French study, the Simplified Acute Physiology Score was calculated and mean values came to 19 points (AT III 20, placebo 19). The analysis was well-balanced for background characteristics.
In all three trials administration of AT III was found to be safe and well-tolerated.
In the aggregate, the 45 % (28/62) 30-day mortality in placebo patients was reduced to 35 % (21/60) in patients t reated with AT III. This corresponds to a relative re- duction in the 30-day all-cause mortality by 22.9 % (con- fidence interval: - 50 to 20 %, see Fig. 4). Kaplan Meier survival analysis is given in Fig. 5.
In the German trial, the impact of AT II! on the sur- vival rate appears to be less pronounced than in the oth- er two trials. The reason could be that the interaction with G A G s occurs via the heparin-binding site of AT
III. Recently it was shown that coadministration of hep- arin - as administered in the German trial - blocks the heparin-binding site of AT III and thus abolishes the drug's anti-inflammatory effect [38]. Whether or not this mechanism was responsible for the difference in mortality remains open. However, a case can be made that in future trials i.v. coadministration of high doses of heparin should be avoided.
A comparison of the impact of AT III on organ dys- function between the studies was not feasible. The defi- nitions of organ dysfunction were too heterogeneous, Fur thermore, in the French trial, data on organ dysfunc- tion were recorded only over the 4-day t reatment peri- od. This period appears too short to allow a conclusion to be drawn on any difference in preventing new organ dysfunction.
In conclusion, the meta-analysis with AT III in cases of severe sepsis resulted in a relative 22.9 % reduction in 30-day all-cause mortality in patients t reated with AT III. To prove the beneficial role of AT III in these patients, the performance of a sufficiently powered pla- cebo-controlled phase I l l trial is warranted.
Aeknowledgemenls We thank Ms. Sylvia Herget for her commit- ted monitoring assistance, Ms. Diana Happel-Schfifer for carefully typing the manuscript, and Ms. Eva Hamers for careful data man- agement and programming•
Appendix A
Members of the study groups
N & W European study Steering Committee: M.Lamy, M.D., University of Liege, Bel- gium; L. G. Thijs, M. D., Free University of Amsterdam, The Neth- erlands. Study group: J.S. R Gimbr~re, M.D., S. J. Van Leeuwen, M. D., Hospital of Nijmegen, The Netherlands; R Lauwers, M.D., R Lormans, M. D., O. A. Kick, M. D., D. Vertommen, M. D., Hospi- tal of Leuven, Belgium; C. Christiansen, M.D., C.Laerkholm, M.D., T.Pedersen, M.D., Hospital of Frederiksberg, Denmark; A. Aasen, M. D., T. E. Ruud, M. D., N. Smith-Erichsen, M. D., Hos- pital of Oslo, Norway; H.A. Bruining, M.D., D.Poldermans, M.D., University of Rotterdam, The Netherlands; RDamas, M.D., University of Liege, Belgium; K. Benneg~rd, M.D., Hospi- tal of Gothenburg, Sweden.
German study Steering committee: U. Hartenauer, M.D., University of Miinster; E R. Matthias, M.D., University of Giessen; W. Oettinger, M.D., University of Ulm; H.-R Schuster, M.D., Municipal Hospital of Hildesheim. Study group: A.S.Kasper, M.D., ELechler, M.D., J. Miiller, M. D., University of K61n; S. Kljucar, M. D., Central Hos- pital St.-Jt~rgen-Str., Bremen; N. Speichermann, M.D., D. Spilker, M.D., Municipal Hospital, Ludwigsburg; U.Hartenauer, M.D., N. Mertes, M.D., J. Zander, M.D., University of Mt~nster; D. Nit- sche, M.D., University of Kiel; K.Falke, M.D., M.Nienhaus, M.D., R. Rossaint, M.D., University Hospital "Rudolf Virchow", Berlin; G. Boeden, M.D., E. Martin, M.D., Hospital of Niirnberg; E Jakschick, M.D., K. Rommelsheim, M. D., University of Bonn.
671
References
1. Bone RC, Balk RA, Cerra FB (1992) Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. Chest 101: 1644-1655
2. Bone RC, Fisher CJ Jr, Clemmer TE Slotman GJ, Metz CA, Balk RA (1989) Sepsis syndrome: a valid clinical entity. Crit Care Med 17:389-393
3. Fisher C J, Dhainaut JFA, Opal SM, Pribble JE Balk RA, Slotman GJ, Iber- tie TJ, Rackow EC, Shapiro M J, Green- man RL, Reines HD, Shelly MR Th- ompson BW, LaBrecque JE Catalano MA, Knaus WA, Sadoff JC (1994) Re- combinant human interleukin-1 recep- tor antagonist in the treatment of pati- ents with sepsis syndrome. JAMA 271: 1836-1843
4. Greenman RL, Schein RMH, Martin MA, Wenzel RE Maclntytre NR, Em- manuel G, Chmel H, Kohler RB, Mc- Carthy M, Plouffe J, Russell RN (1991) A controlled clinical trial of E5 murine monoclonal IgM antibody to endotoxin in the treatment of gram-negative sep- sis. JAMA 266:1097-1102
5. Ziegler EJ, Fisher CJ, Sprung CL, Stra- ube RC, Sadoff JC, Foulke GE, Wortel CH, Fink ME Dellinger E Teng NNH, Allen IE, Berger HJ, Knatterud GL, LoBuglio AE Smith CR (1991) Treat- ment of gram-negative bacteremia and septic shock with HA-1A human mono- clonal antibody against endotoxin. A randomized, double-blind, placebo con- trolled trial. N Engl J Med 324: 429-436, 486-487
6. Dinarello CA, Gelfand JA, Wolff SM (1993) Anticytokine strategies in the treatment of the systemic inflammatory response syndrome. JAMA 269: 1829-1835
7. Dhainaut F-J A, Tenaillon A, Le Tulzo Y, Schlemmer B, Solet J-R Wolff M, Holzapfel L, Zeni E Dreyfuss D, Mira J-R De Vathaire F, Guinot P (1994) Platelet-activating factor receptor an- tagonist BN52021 in the treatment of severe sepsis: a randomized, double- blind, placebo-controlled, multicenter clinical trial. Crit Care Med 22: 1720-1728
8. McCloskey RV, Straube RC, Sanders C, Smith SM, Smith CR (1994) Treatment of septic shock with human monoclonal antibody HA-1A. Ann Intern Med 121: 1-5
9. Abraham E, Wunderink R, Silverman H, Perl TM, Nasraway S, Levy H, Bone R, Wenzel RR Balk R, Allred R, Pennington JE, Wherry JC (1995) Effi- cacy and safety of monoclonal antibody to human tumor necrosis factor ~ in pa- tients with sepsis syndrome, JAMA 273: 934-941
10. Piguet PF, Grau GE, Vassalli P (1990) Subcutaneous perfusion of tumor ne- crosis factor induces local proliferation of fibroblasts, capillaries, and epidermal cells, or massive tissue necrosis. Am J Pathol 136:103-110
11. Remick DG, Kunkel RG, Larric JW, Kunkel SL (1987) Acute in vivo effects of human recombinant tumor necrosis factor. Lab Invest 56:583-590
12. van Deventer SJH, Buller HR, ten Cate JW, Aarden LA, Hack CE, Sturk A (1990) Experimental endotoxemia in humans: analysis of cytokine release and coagulation, fibrinolytic and com- plement pathways. Blood 76:2520-2526
13. Bauer KA, ten Cate H, Barzegar S, Spriggs DR~ Sherman ML, Rosenberg RD (1989) Tumor necrosis factor infu- sions have a procoagulant effect on the hemostatic mechanism of humans. Blood 74:165-172
14. van der Poll T, Buller HR, ten Cate H (1990) Activation of coagulation after administration of tumor necrosis factor to normal subjects. N Engl J Med 322: 1622-1627
15. Fourrier F, Chopin C, Goudemand J, Hendrycx S, Caron C, Rime A, Marey A, Lestavei P (1992) Septic shock, mul- tiple organ failure, and disseminated in- travascular coagulation. Chest 101: 816-823
16. Dickneite G, P~ques EP (1993) Reduc- tion of mortality with antithrombin III in septicemic rats: a study of Klebsiella pneumoniae induced sepsis. Thromb Haemost 2:98-102
17. Rao LVM, Nordfang O, Hoang AD, Pendurthi UR (1995) Mechanism of an- tithrombin III inhibition of factor VIIa/ tissue factor activity on cell surfaces. Comparison with tissue factor pathway inhibitor/factor Xa-induced inhibition of factor VIIa/tissue factor activity. Blood 85:121-129 Smith-Erichsen N, Aasen AO, Galli- more MJ, Amundsen E (1982) Studies of components of the coagulation sys- tems in normal individuals and septic shock patients. Circ Shock 9:491-497
18.
19. Voss R, Matthias FR, Borkowski G, Re- itz D (1990) Activation and inhibition of fibrinolysis in septic patients in an in- ternal intensive care unit. Br J Haema- tol 75:99-105
20. Havemann K, Gramse M (1984) Physi- ology and pathophysiology of neutral proteinases of human granulocytes. In: H6rl WH, Heidland A (eds) Proteases: potential role in health and disease. Ple- num, New York, pp 1-20
21. Seitz R, Wolf M, Egbring R, Havemann K (1989) The disturbance of haemosta- sis in septic shock: role of neutrophil elastase and thrombin, effects of anti- thrombin III and plasma. Eur J Haema- tol 43:22-28
22. Schipper HG, Jenkins CSR Kahl LH, Ten Cate JW (1978) Antithrombin III transfusion in disseminated intravascu- lar coagulation. Lancet I: 854-856
23. Fourrier F, Chopin C, Huart J, Runge I, Caron C, Goudemand J (1993) Double- blind, placebo-controlled trial of anti- thrombin III concentrates in septic shock with disseminated intravascular coagulation. Chest 104:882-888
24. Schuster HR Matthias FR (1995) Anti- thrombin III in severe sepsis. 15th Inter- national Symposium of Intensive Care and Emergency Medicine, Brussels, March, 21-24, 1995
25. Knaus WA, Draper EA, Wagner DE Zimmermann JE (1985) APACHE II: a severity of disease classification sys- tem. Crit Care Med 13:818-829
27. Knaus WA, Draper EA, Wagner DR Zimmermann JE (1985) Prognosis in acute organ system failure (OSF). Ann Surg 202:685-693
28. Jordan DA, Miller CE Kubos KL, Rog- ers MC (1987) Evaluation of sepsis in a critically III surgical population. Crit Care Med 15:897-904
29, Kleinbaum DG, Kupper LL, Morgen- stern H (1982) Epidemiologic research: principles and quantitative methods. Wadsworth, Belmont, CA
30. Galbraith RJ (1988) A note on graphi- cal presentations of estimated odds ra- tios from several clinical trials. Stat Med 7:889-894
31. Williams J, Maier RV (1992) The in- flammatory response. J Intensive Care Med 7:53-66
672
32. Regoeczi E, Brain MC (1969) Organ distribution of fibrin in disseminated in- travascular coagulation. Br J Haematol 17:73-79
33. Robboy S J, Colman RW, Minna JD (1972) Pathology of disseminated intra- vascular coagulation (DIC): analysis of twenty-six cases. Hum Pathol 3:327
34. Taylor FB, Emerson TE, Jordan R, Chang AK, Blick KE (1988) Anti- thrombin III prevents the lethal effects of Escherichia coil infusion in baboons. Circ Shock 26:227-235
35. Taylor FB, Chang ACK, Peer GT, Mather T, Blick K, Catlett R, Lockhart MS, Esmon CT (199l) DEGR-factor Xa blocks disseminated intravascular coagulation initiated by Escherichia coli without preventing shock or organ damage. Blood 78:364-368
36. Yamauchi T, Umeda F, Inoguchi T, Nawata H (1989) Antithrombin III stimulates prostacyclin production by cultured aortic endothelial cells. Bio- chem Biophys Res Commun 163: 1404-1411
37. Horie S, Ishii H, Kazama M (1990) Heparin-like glycosaminoglycan is a re- ceptor for antithrombin I I I - dependent but not for thrombin-dependent prosta- cyclin production in human endothelial cells. Thromb Res 59:895-904
38. Okajima K, Uchiba M, Murakami K (1995) Antithrombin replacement in DIC and MOF. In: Vincent JL (ed) Yearbook of intensive care and emer- gency medicine. Springer, Berlin Hei- delberg New York, pp 457-464
39. Blauhut B, Kramar H~ Vinazzer H, Bergmann H (1985) Substitution of an- tithrombin III in shock and DIC: a ran- domized study. Thromb Res 39:81-89
