DIAGN MICROBIOLINFECTDIS 1990;13:377-381
377
Cytokines and Septic Shock Thierry Calandra and Michel P. Glauser
INTRODUCTION Septic shock is a life-threatening complication of infection, which can be triggered by a wide spectrum of microorganisms. Bacteria, mainly Gram-negative bacilli and Gram-positive cocci, are the most frequent microbial agents isolated from patients with septic shock. However, other microorganisms such as spirochetes, rickettsia, viruses, fungi, or parasites may also occasionally cause septic shock. Most of our present knowledge on the pathogenic mechanisms involved in septic shock is derived from experimental and clinical data obtained either in animals or in humans infected with Gram-negative bacteria. With the exception of some Gram-positive infections, in most infections the pathogenic mechanisms responsible for the development of shock are either unknown or not completely understood (Sanford; 1985). In the present article we will therefore focus on Gram-negative infections and on the central role played by cytokines, especially tumor necrosis factor/cachectin (TNF), in the pathogenesis of Gramnegative septic shock. Over the past decades, the incidence of Gramnegative bacteremias has increased markedly in most medical centers (McCabe and Jackson, 1962; Kreger et al., 1980a). At theBoston City Hospital, the incidence of Gram-negative bacteremia increased from 0.9 cases per 1000 admissions in 1935 to 11.2 cases per 1000 admissions in 1972 (McGowan et al., 1975). In 1988, the incidence of Gram-negative bacteremia was 8 per 1000 admissions in our hospital. If comparable, the data from Boston in 1972 and from our institution in 1988 suggest that the incidence of Gramnegative bacteremia might have reached a plateau. One of the most striking and disappointing observations is that the mortality of these Gram-negative From the Division of InfectiousDiseases, Departmentof Internal Medicine, Centre Hospitalier UniversitaireVaudois, Lausanne, Switzerland. © 1990ElsevierSciencePublishing Co., Inc. 655 Avenue of the Americas,New York, NY 10010 0732-8893/90/$3.50
infections has been unchanged in spite of major breakthrough in both supportive care and antimicrobial therapy. Gram-negative bacteremias are still associated with a mortality of 20-35% (Felty and Keefer, 1924; Wolff and Bennett, 1974; McGowan et al., 1975; Kreger et al., 1980b; Bryan et a1.,1983). In patients developing Gram-negative septic shock, fatality ratios may be as high as 50% to 80% (Ziegler et al., 1982; Sprung et al., 1984; Bone et al., 1987; Veteran's Administration Systemic Sepsis Cooperative Study Group, 1987; Calandra et al., 1988). With the possible exception of passive immunotherapy with antibodies directed against the endotoxin core of Gram-negative bacteria (Ziegler et al., 1982), new therapeutic approaches such as treatment with corticosteriods or opiate antagonists have failed to reduce the mortality of Gram-negative bacteremia or septic shock (Sprung et al., 1984; De Maria et al., 1985; Bone et al., 1987; Veterans Administration Systemic Sepsis Cooperative Study Group, 1987).
TUMOR NECROSIS FACTOR/CACHECTIN (TNF) Most of the toxic manifestations induced by Gramnegative bacteria are triggered by endotoxin, also called lipopolysaccharide (LPS), which is a component of the outer membrane of these bacteria. Endotoxin can be released from the surface of the bacteria, either during bacterial growth or during antibiotic therapy (Shenep and Morgan, 1984). The lipid A, which is the innermost structure of endotoxin, is the moiety responsible for the toxic effects of endotoxin. In human volunteers it has been recently shown that the injection of a single intravenous bolus of Escherichi coli endotoxin reproduced the hemodynamic effects observed in patients with septic shock (Suffredini et al., 1989). A major breakthrough in our understanding of the pathophysiology of Gram-negative infections has been the recent recognition of the central role played by TNF in the pathogenesis of endotoxic shock
378
(Beutler et al., 1987 and 1988; Sherry and Cerami, 1988). TNF is a cytokine produced by activated macrophages upon exposure to endotoxin (Beutler et al., 1985c). Five lines of arguments strongly suggest that TNF is a primary mediator of the deleterious effects of endotoxin. First, in rabbits TNF was shown to be released in plasma after an intravenous challenge with E. coli LPS (Beutler et al., 1985a; Mathison et al., 1988). In a canine model of septic shock, a thrombin fibrin clot containing purified E. coli endotoxin implanted into the peritonuem or the intravenous infusion of recombinant TNF was shown to trigger cardiovascular changes similar to those observed in human septic shock (Natanson et al., 1989). Second, in rats and in dogs, the infusion of recombinant human TNF, in quantities similar to those produced endogeneously in response to endotoxin, was accompanied by hypotension, metabolic acidosis, hemoconcentration, and death within minutes to hours (Tracey et al., 1986 and 1987b). Third, mice passively immunized with polyclonal rabbit antiserum directed against murine TNF or with purified immunoglobulin were protected against the lethal effect of an intravenous injection of E. coli endotoxin (Beutler et al., 1985b). Similar results were obtained in rabbits passively immunized with a polyclonal anti-TNF antibody (Beutler et al., 1985a). Fourth, in a baboon model of septic shock, passive immunization with anti-TNF monoclonal antibodies administered 2 hr before a lethal challenge of live E. coli was shown to protect the animals from shock, organ dysfunction, persistent stress hormone release, and death (Tracey et al., 1987a). Finally, levels of immunoreactive TNF have been detected in the plasma of healthy human volunteers challenged with a single intravenous injection of 4 ng/kg of E. coli endotoxin (Hesse et al., 1988; Michie et al., 1988). In both studies the plasma concentration of TNF increased after 30-60 min and peaked at 90-120 min (358 --- 166 pg/ml and 240 _+70 pg/ml, respectively). Plasma TNF was not detectable 4-6 hr after the injection of endotoxin. The rise in TNF concentration was accompanied by chills, headache, myalgias, and nausea and by an increase in body temperature, heart rate, and stress hormones (i.e., ACTH and epinephrine) but was not associated with a drop in blood pressure. Moreover, in cancer patients the infusion of recombinant human TNF at a dose of -> 545 bLg/m2/24 hr produced peak plasma concentration of TNF, symptoms, and metabolic responses similar to those observed in volunteers challenged with endotoxin (Michie et al., 1988). Thus, in healthy humans the infusion of endotoxin was followed by a transient increase in plasma levels of TNF, which was associated with the development of the clinical signs and biochemical responses encountered in Gram-negative infections.
T. Calandra and M.P. Glauser
INTERLEUKIN-1, G A M M A INTERFERON, A N D INTERLEUKIN-6 It is obviously beyond the scope of this article to review the multiple biological properties of other mediators such as interleukin-1 (IL-1; Dinarello, 1988), gamma interferon (IFN-~/; Nathan and Yoshida, 1988), and interleukin-6 (IL-6; Wong and Clark, 1988) and to dissect out their respective roles and interplay in the pathogenesis of septic shock. Briefly, in human volunteers, concentrations of IL-1 remained at baseline values (Michie et al., 1988a and b) or increased only slightly (Hesse et al., 1988) after an intravenous injection of endotoxin or the infusion of TNF. Also, the injection of endotoxin or the infusion of TNF did not induce the release of IFN-~/in plasma. In rabbits, the infusion of a single dose of 5 ~g/kg of recombinant IL-113 was associated with the development of a transient hypotension, a decrease in systemic vascular resistance, and an increase in cardiac output and heart rate (Okusawa et al., 1988). These hemodynamic effects were prevented by ibuprofen pretreatment, indicating that it required cyclo-oxygenase products. In this model, IL-1 and TNF appeared to act synergistically, because the combination of the two cytokines was more potent than either agent alone. Thus, in rabbits, ILl was capable, like TNF, of inducing a shock state. Similar results were obtained by Waage and Espevik in a mouse model (1988).
CYTOKINES IN PATIENTS WITH SEPSIS A N D SEPTIC S H O C K In a study of the serum levels of immunoreactive TNF in 104 patients with various infectious diseases, Scuderi et al. (1986) detected elevated TNF (i.e., >39 pg/ml) in 18 of 27 (67%) patients with kala-azar, in 7 of 10 (70%) patients with malaria, and in only 2 of 20 (10%) patients with bacterial infections. None of the patients with other parasitic, viral, or fungal infections had elevated levels of serum TNF. Using a bioassay, Waage et al. (1986) detected a cytotoxic TNF activity in the serum of 3 of 23 (13%) patients with various infectious diseases but not in 20 cancer patients or in 25 controls. In two subsequent articles, the same group of investigators reported on the measurement of biologically active TNF, IL-6, and IL-1 in the serum of 79 patients with meningococcal disease (Waage et al., 1987 and 1989). TNF was detected in 10 of 11 patients who died and in only 8 of 68 survivors (p 440 U/ml survived. In this study, blood pressure, the presence of ecchymoses, and leukocyte and platelet counts were also shown to be associated with the patient's outcome (Waage et al., 1989). IL-6 was released in the serum
Cytokines and Septic Shock
of 69 patients (87%). Median serum levels of IL-6 were higher in patients with septic shock (189 ng/ml) than in patients with meningitis, bacteremia, or combined septic shock and meningitis (0.2 ng/ml in each of these groups of patients). All of the patients with IL-6 levels --
377
Cytokines and Septic Shock Thierry Calandra and Michel P. Glauser
INTRODUCTION Septic shock is a life-threatening complication of infection, which can be triggered by a wide spectrum of microorganisms. Bacteria, mainly Gram-negative bacilli and Gram-positive cocci, are the most frequent microbial agents isolated from patients with septic shock. However, other microorganisms such as spirochetes, rickettsia, viruses, fungi, or parasites may also occasionally cause septic shock. Most of our present knowledge on the pathogenic mechanisms involved in septic shock is derived from experimental and clinical data obtained either in animals or in humans infected with Gram-negative bacteria. With the exception of some Gram-positive infections, in most infections the pathogenic mechanisms responsible for the development of shock are either unknown or not completely understood (Sanford; 1985). In the present article we will therefore focus on Gram-negative infections and on the central role played by cytokines, especially tumor necrosis factor/cachectin (TNF), in the pathogenesis of Gramnegative septic shock. Over the past decades, the incidence of Gramnegative bacteremias has increased markedly in most medical centers (McCabe and Jackson, 1962; Kreger et al., 1980a). At theBoston City Hospital, the incidence of Gram-negative bacteremia increased from 0.9 cases per 1000 admissions in 1935 to 11.2 cases per 1000 admissions in 1972 (McGowan et al., 1975). In 1988, the incidence of Gram-negative bacteremia was 8 per 1000 admissions in our hospital. If comparable, the data from Boston in 1972 and from our institution in 1988 suggest that the incidence of Gramnegative bacteremia might have reached a plateau. One of the most striking and disappointing observations is that the mortality of these Gram-negative From the Division of InfectiousDiseases, Departmentof Internal Medicine, Centre Hospitalier UniversitaireVaudois, Lausanne, Switzerland. © 1990ElsevierSciencePublishing Co., Inc. 655 Avenue of the Americas,New York, NY 10010 0732-8893/90/$3.50
infections has been unchanged in spite of major breakthrough in both supportive care and antimicrobial therapy. Gram-negative bacteremias are still associated with a mortality of 20-35% (Felty and Keefer, 1924; Wolff and Bennett, 1974; McGowan et al., 1975; Kreger et al., 1980b; Bryan et a1.,1983). In patients developing Gram-negative septic shock, fatality ratios may be as high as 50% to 80% (Ziegler et al., 1982; Sprung et al., 1984; Bone et al., 1987; Veteran's Administration Systemic Sepsis Cooperative Study Group, 1987; Calandra et al., 1988). With the possible exception of passive immunotherapy with antibodies directed against the endotoxin core of Gram-negative bacteria (Ziegler et al., 1982), new therapeutic approaches such as treatment with corticosteriods or opiate antagonists have failed to reduce the mortality of Gram-negative bacteremia or septic shock (Sprung et al., 1984; De Maria et al., 1985; Bone et al., 1987; Veterans Administration Systemic Sepsis Cooperative Study Group, 1987).
TUMOR NECROSIS FACTOR/CACHECTIN (TNF) Most of the toxic manifestations induced by Gramnegative bacteria are triggered by endotoxin, also called lipopolysaccharide (LPS), which is a component of the outer membrane of these bacteria. Endotoxin can be released from the surface of the bacteria, either during bacterial growth or during antibiotic therapy (Shenep and Morgan, 1984). The lipid A, which is the innermost structure of endotoxin, is the moiety responsible for the toxic effects of endotoxin. In human volunteers it has been recently shown that the injection of a single intravenous bolus of Escherichi coli endotoxin reproduced the hemodynamic effects observed in patients with septic shock (Suffredini et al., 1989). A major breakthrough in our understanding of the pathophysiology of Gram-negative infections has been the recent recognition of the central role played by TNF in the pathogenesis of endotoxic shock
378
(Beutler et al., 1987 and 1988; Sherry and Cerami, 1988). TNF is a cytokine produced by activated macrophages upon exposure to endotoxin (Beutler et al., 1985c). Five lines of arguments strongly suggest that TNF is a primary mediator of the deleterious effects of endotoxin. First, in rabbits TNF was shown to be released in plasma after an intravenous challenge with E. coli LPS (Beutler et al., 1985a; Mathison et al., 1988). In a canine model of septic shock, a thrombin fibrin clot containing purified E. coli endotoxin implanted into the peritonuem or the intravenous infusion of recombinant TNF was shown to trigger cardiovascular changes similar to those observed in human septic shock (Natanson et al., 1989). Second, in rats and in dogs, the infusion of recombinant human TNF, in quantities similar to those produced endogeneously in response to endotoxin, was accompanied by hypotension, metabolic acidosis, hemoconcentration, and death within minutes to hours (Tracey et al., 1986 and 1987b). Third, mice passively immunized with polyclonal rabbit antiserum directed against murine TNF or with purified immunoglobulin were protected against the lethal effect of an intravenous injection of E. coli endotoxin (Beutler et al., 1985b). Similar results were obtained in rabbits passively immunized with a polyclonal anti-TNF antibody (Beutler et al., 1985a). Fourth, in a baboon model of septic shock, passive immunization with anti-TNF monoclonal antibodies administered 2 hr before a lethal challenge of live E. coli was shown to protect the animals from shock, organ dysfunction, persistent stress hormone release, and death (Tracey et al., 1987a). Finally, levels of immunoreactive TNF have been detected in the plasma of healthy human volunteers challenged with a single intravenous injection of 4 ng/kg of E. coli endotoxin (Hesse et al., 1988; Michie et al., 1988). In both studies the plasma concentration of TNF increased after 30-60 min and peaked at 90-120 min (358 --- 166 pg/ml and 240 _+70 pg/ml, respectively). Plasma TNF was not detectable 4-6 hr after the injection of endotoxin. The rise in TNF concentration was accompanied by chills, headache, myalgias, and nausea and by an increase in body temperature, heart rate, and stress hormones (i.e., ACTH and epinephrine) but was not associated with a drop in blood pressure. Moreover, in cancer patients the infusion of recombinant human TNF at a dose of -> 545 bLg/m2/24 hr produced peak plasma concentration of TNF, symptoms, and metabolic responses similar to those observed in volunteers challenged with endotoxin (Michie et al., 1988). Thus, in healthy humans the infusion of endotoxin was followed by a transient increase in plasma levels of TNF, which was associated with the development of the clinical signs and biochemical responses encountered in Gram-negative infections.
T. Calandra and M.P. Glauser
INTERLEUKIN-1, G A M M A INTERFERON, A N D INTERLEUKIN-6 It is obviously beyond the scope of this article to review the multiple biological properties of other mediators such as interleukin-1 (IL-1; Dinarello, 1988), gamma interferon (IFN-~/; Nathan and Yoshida, 1988), and interleukin-6 (IL-6; Wong and Clark, 1988) and to dissect out their respective roles and interplay in the pathogenesis of septic shock. Briefly, in human volunteers, concentrations of IL-1 remained at baseline values (Michie et al., 1988a and b) or increased only slightly (Hesse et al., 1988) after an intravenous injection of endotoxin or the infusion of TNF. Also, the injection of endotoxin or the infusion of TNF did not induce the release of IFN-~/in plasma. In rabbits, the infusion of a single dose of 5 ~g/kg of recombinant IL-113 was associated with the development of a transient hypotension, a decrease in systemic vascular resistance, and an increase in cardiac output and heart rate (Okusawa et al., 1988). These hemodynamic effects were prevented by ibuprofen pretreatment, indicating that it required cyclo-oxygenase products. In this model, IL-1 and TNF appeared to act synergistically, because the combination of the two cytokines was more potent than either agent alone. Thus, in rabbits, ILl was capable, like TNF, of inducing a shock state. Similar results were obtained by Waage and Espevik in a mouse model (1988).
CYTOKINES IN PATIENTS WITH SEPSIS A N D SEPTIC S H O C K In a study of the serum levels of immunoreactive TNF in 104 patients with various infectious diseases, Scuderi et al. (1986) detected elevated TNF (i.e., >39 pg/ml) in 18 of 27 (67%) patients with kala-azar, in 7 of 10 (70%) patients with malaria, and in only 2 of 20 (10%) patients with bacterial infections. None of the patients with other parasitic, viral, or fungal infections had elevated levels of serum TNF. Using a bioassay, Waage et al. (1986) detected a cytotoxic TNF activity in the serum of 3 of 23 (13%) patients with various infectious diseases but not in 20 cancer patients or in 25 controls. In two subsequent articles, the same group of investigators reported on the measurement of biologically active TNF, IL-6, and IL-1 in the serum of 79 patients with meningococcal disease (Waage et al., 1987 and 1989). TNF was detected in 10 of 11 patients who died and in only 8 of 68 survivors (p 440 U/ml survived. In this study, blood pressure, the presence of ecchymoses, and leukocyte and platelet counts were also shown to be associated with the patient's outcome (Waage et al., 1989). IL-6 was released in the serum
Cytokines and Septic Shock
of 69 patients (87%). Median serum levels of IL-6 were higher in patients with septic shock (189 ng/ml) than in patients with meningitis, bacteremia, or combined septic shock and meningitis (0.2 ng/ml in each of these groups of patients). All of the patients with IL-6 levels --
