Portal de Periódicos da CAPES

Acquired antithrombin deficiency in sepsis

2001; Wiley; Volume: 112; Issue: 1 Linguagem: Inglês

10.1046/j.1365-2141.2001.02396.x

1365-2141

Barry White, David J. Perry,

Sepsis Diagnosis and Treatment

Antithrombin (AT) is a natural anticoagulant that plays a pivotal role in coagulation homeostasis by inhibiting thrombin and factor Xa (FXa) and, to a lesser extent, factor IXa and factor XIa (Abildgaard, 1969; Damus et al, 1973; Rosenberg & Damus, 1973; Kurachi et al, 1976). In addition to its anticoagulant function, there is increasing evidence that AT also has potent anti-inflammatory properties and is protective in animal models of sepsis (Taylor et al, 1988; Dickneite & Paques, 1993; Kessler et al, 1997; Dickneite & Leithauser, 1999; Minnema et al, 2000). Acquired AT deficiency is commonly found in sepsis and the level is predictive of outcome (Fourrier et al, 1992; Lorente et al, 1993; Martinez et al, 1999). Furthermore, recent data from randomized and non-randomized clinical studies suggests that AT replacement therapy in patients with septic shock may shorten the duration of disseminated intravascular coagulopathy (DIC), reduce organ dysfunction and improve survival (Fourrier et al, 1993; Schuster & Matthias, 1995; Baudo et al, 1998; Eisele et al, 1998). In this article, we review the anticoagulant and anti-inflammatory properties of AT, the causes of acquired AT deficiency in sepsis and the clinical evidence supporting the use of AT replacement therapy in these patients. Human antithrombin is a single-chain polypeptide produced in the liver and is a member of a family of related inhibitory proteins known collectively as the serpins (serine protease inhibitors) (Hunt & Dayhoff, 1980). Their inhibitory activity is dependent on the formation of a stable 1:1 inactive complex between the serine at the active site of the protease and a reactive peptide bond of the serpin, designated P1– (Hunt & Dayhoff, 1980). The anticoagulant effect of AT is accelerated at least 1000-fold by binding to heparin or other related sulphated glycosaminoglycans (Damus et al, 1973). While free heparin is not present in the circulation under normal physiological conditions, it is probable that the binding of AT to heparan sulphate, located on vascular endothelium, or dermatan sulphate, in the deeper layers of the vessel wall, provides a mechanism for accelerating its activity in vivo (Nydahl et al, 1989; Wight, 1989; Frebelius et al, 1990). In addition to its anticoagulant function, recent data suggest that AT also has potent anti-inflammatory properties. Preloading baboons with recombinant or plasma-derived AT in doses sufficient to achieve supraphysiological levels significantly reduced multiorgan failure, DIC and mortality, associated with the infusion of lethal doses of Escherichia coli (Taylor et al, 1988; Minnema et al, 2000). This decrease in morbidity and mortality is accompanied by a reduction in the pro-inflammatory cytokines, interleukin 6 (IL-6) and interleukin 8 (IL-8), although tumour necrosis factor (TNF)-α remains elevated. In porcine, guinea pig and rat models of sepsis, AT was associated with similar protective effects (Taylor et al, 1988; Dickneite & Paques, 1993; Kessler et al, 1997; Dickneite & Leithauser, 1999; Minnema et al, 2000). Furthermore, in the rat model of sepsis, AT inhibits an endotoxin-induced increase in vascular permeability and pulmonary accumulation of leucocytes, a critical component in the development of respiratory failure associated with sepsis (Uchiba et al, 1995a,b, 1996). In all preclinical studies, the protective effects of AT were only evident at supraphysiological plasma levels and the beneficial effects were significantly reduced when treatment was initiated after, rather than before, the exposure to live bacteria or endotoxins (Taylor et al, 1988; Dickneite & Paques, 1993; Kessler et al, 1997; Dickneite & Leithauser, 1999; Minnema et al, 2000). There is increasing data to suggest that the coagulation cascade is responsible not only for the formation of the fibrin clot and the activation of the natural anticoagulant and fibrinolytic pathways, but also for the generation of serine proteases with potent pro-inflammatory properties (Esmon, 2000). Thrombin is the most frequently recognized enzyme to be implicated in augmentation of the inflammatory response. Cell activation by thrombin is mediated by binding to protease-activated receptors (PARs) 1, 3 and 4, and this results in the generation of secondary messengers (Coughlin, 1999; Kahn et al, 1999). Thrombin has been shown to induce IL-6 and IL-8 release from cultured endothelial cells and monocytes and to augment endotoxin-mediated release of IL-1β from macrophages (Jones & Geczy, 1990; Johnson et al, 1998). Thrombin also facilitates the binding of neutrophils to the vascular endothelium via the elaboration of adhesion molecules and stimulates the release of platelet-activating factor (PAF), which is a potent activator of neutrophils (Shankar, 1992, 1994; Sugama et al, 1992). The endothelium is normally protected from thrombin-mediated activation by an endothelial protein called thrombomodulin (Esmon, 1989). The binding of thrombomodulin to thrombin's anion-binding exosite-I prevents binding of procoagulant substrates and switches thrombin substrate specificity so that it binds to, and proteolytically cleaves, protein C, resulting in the generation of activated protein C (Grinnell & Berg, 1996; Fuentes-Prior et al, 2000). This protective effect may be lost in sepsis owing to the cytokine-mediated downregulation of thrombomodulin and its proteolytic cleavage by neutrophil elastase (Conway & Rosenberg, 1988; Takano et al, 1990). Consequently, the endothelium may become more dependent on endothelial-bound AT to protect it from the increased generation of thrombin as the inflammatory process proceeds. Factor Xa is an additional clotting factor that may possess potentially important pro-inflammatory properties. A receptor for FXa, termed effector cell protease receptor 1 (EPR-1), has been identified on endothelium, leucocytes and smooth muscle cells (Altieri & Edgington, 1990; Altieri, 1994; Nicholson et al, 1996; Herbert et al, 1998). Factor Xa activates cells either by EPR-1-dependent or -independent pathways. Factor Xa induces the increased synthesis and release of IL-6, IL-8 and monocyte chemotactic protein (MCP)-1 from endothelial cells in an EPR-1-independent reaction (Senden et al, 1998), and results in the augmentation of lymphocyte proliferation and oedema via an EPR-1-dependent pathway (Altieri & Stamnes, 1994; Cirino et al, 1997). It is tempting to speculate that the protective effects of AT in sepsis, the associated downregulation of IL-6 and IL-8, and the reduction in cell adhesion molecules are mediated by the binding of AT to thrombin and FXa and the subsequent inhibition of their pro-inflammatory effects. Systemic infusions of heparin or a competitive inhibitor of FXa, [5-(dimethylamino)1-naphthalenesulphonyl]-glutamylglycylarginyl chloromethyl ketone (DEGR-FXa), prevented the coagulation abnormalities in animal models of sepsis, but failed to protect the host from organ failure or improve the overall survival (Taylor et al, 1991; Kessler et al, 1997). The administration of heparin: AT complex in a porcine and guinea pig model of sepsis also prevented DIC, but again without decreasing the risk of multiorgan failure or improving survival (Spannagl et al, 1991; Kessler et al, 1997). The most probable explanation for these experiments is that the protective effects of AT are dependent on its binding to endothelial glycosaminoglycans and that this interaction is competitively inhibited by the co-administration of heparin. Therefore, endothelial-bound AT may provide a local mechanism whereby the pro-inflammatory effects of thrombin or FXa are inhibited at the interface between these serine proteases and endothelial cells or leucocytes. Endothelial-bound AT may also mediate its protective effect in sepsis by increasing the release of prostacylcin from endothelial cells (Yamauchi et al, 1989; Horie et al, 1990; Uchiba et al, 1995a). While the cellular mechanisms responsible for this effect have not been established, it is probably mediated by a cyclooxygenase, as it is inhibited by indomethacin (Uchiba et al, 1996). Prostacyclin promotes vasodilation and inhibits platelet aggregation, but also appears to have a range of anti-inflammatory properties including the inhibition of monocyte pro-inflammatory cytokine production and the inhibition of neutrophil activation and adhesion to endothelial cells (Kainoh et al, 1990; Eisenhut et al, 1993; Uchiba et al, 1998). Pharmacokinetic studies on septic patients with acquired AT deficiency, clearly demonstrated a reduced recovery and shortened half-life that probably reflects, at least in part, the increased consumption of AT demonstrated by the elevated levels of thrombin:antithrombin complexes (Harper et al, 1996). Antithrombin also appears to function as a negative acute-phase protein and in vitro studies using a Hep G2 liver cell line demonstrated that IL-6 and IL-1β have a synergistic effect in inhibiting AT production (Niessen et al, 1997). Therefore, the pathophysiological mechanism(s) responsible for acquired AT deficiency in sepsis is probably multifactorial, including decreased synthesis, increased consumption and possibly capillary leak owing to increased vascular permeability. Regardless of the mechanism, AT deficiency is invariably present in severe sepsis and the level is predictive of outcome (Fourrier et al, 1992; Lorente et al, 1993; Martinez et al, 1999). Several non-randomized clinical trials in patients with sepsis have demonstrated a reduction in predicted mortality and downregulation of pro-inflammatory cytokine production and adhesion molecule expression in favour of AT replacement therapy (Blauhut et al, 1982, 1985; Inthorn et al, 1997, 1998; Kreuz et al, 1999). The role of AT therapy in sepsis has also been assessed in four placebo-controlled randomized trials (Fourrier et al, 1993; Baudo et al, 1998; Eisele et al, 1998). In the study by Baudo et al (1998), 120 patients who were admitted to intensive care with plasma AT levels of less than 70 U/dl were randomly assigned to 4000 IU of AT as a bolus dose or placebo with a repeat dose of 2000 IU or placebo every 12 h for 5 d. While there was no difference in survival between the two groups, subgroup analysis of those patients with septic shock (n = 56) demonstrated a significant survival advantage in favour of AT treatment (predicted mortality of 13% vs. 30%). In the remaining three studies, which included a smaller number of patients with septic shock (n = 45, n = 35 and n = 42), there was a trend in favour of improved survival associated with AT replacement therapy (Fourrier et al, 1993; Eisele et al, 1998). The treatment doses of AT used in these studies were: 3000 IU loading dose followed by a bolus of 500 IU every 4 h for 7 d; 90–120 IU/kg over 3 h followed by 90–120 IU/kg/d as a continuous infusion for 4 d; and 3000 IU as a bolus dose over 1 h followed by 1500 IU every 12 h for 5 d (Fourrier et al, 1993; Eisele et al, 1998). In the meta-analysis of the three studies there was a 23% reduction in the 30 d all-cause mortality (45% vs. 35%), but this failed to achieve statistical significance (Eisele et al, 1998). While larger studies are clearly required, the results of these phase II clinical studies suggests that the anti-inflammatory properties of AT may provide an important mechanism whereby the morbidity and mortality associated with severe sepsis can be significantly improved. It is the authors' view that there is currently insufficient evidence to support the use of AT replacement therapy in sepsis outside randomized clinical trials. Some authors have argued that, based on preclinical and clinical studies, AT therapy should be considered in severe sepsis (Schinzel & Weilemann, 1998). Physicians who wish to proceed with this therapeutic strategy should consider the following clinically relevant issues. As with any novel adjunctive therapy, the administration of AT therapy is probably not of benefit if the standard clinical management is suboptimal. Failure to adhere to the basic principles of management of severe sepsis such as early antibiotic therapy, aggressive fluid resuscitation, correction of acid-base defects and appropriate ventilation support will compromise patient outcome and limit any potential benefit of AT therapy (Astic, 1998). Similarly, while AT has been associated with early resolution of DIC, it does not obviate the need for coagulation support. Fibrinogen values of less than 0·8 g/l and a platelet count of less than 20–50 × 109/l (depending upon the local practice and patient status) require correction with cryoprecipitate or fibrinogen concentrate and platelet transfusions respectively. The effective local inhibition of thrombin at the site of vessel injury may be associated with, at least, a theoretical risk of excessive bleeding, especially in those patients whose haemostatic response is already significantly compromised (e.g. severe hypofibrinogenaemia or thrombocytopenia). Despite these concerns, there is no evidence from clinical or preclinical studies that AT therapy is associated with an increased risk of haemorrhagic complications Currently available AT concentrates are plasma-derived blood products and, although they undergo viral inactivation, there is still a residual risk of infection especially from non-enveloped viruses, prion proteins or previously unidentified infectious agents. One should not assume that the AT that is present in commercial preparations is the same as native AT. Chang & Harper (1997) demonstrated that approximately 40% of a commercial AT preparation existed as the inactive latent forms (l-forms), presumably because of changes in the protein configuration during viral inactivation. The clinical significance of administering this altered material is uncertain, however, it will probably reduce the efficacy of AT replacement therapy in sepsis, especially in those patients in whom the response to treatment is not measured by functional AT assays. Furthermore, while the effect of the l-forms of AT on the immune response has not been established, there is evidence that proteolytically altered serpins exacerbate the inflammatory process in the baboon sepsis model. The Pittsburgh variant of α1-anti-trypsin (358 Met-Arg) is a novel protease inhibitor and, in common with AT, has activity against both thrombin and the contact proteases. The Pittsburgh mutant was expected to improve the outcome of septic shock in baboons, however, it unexpectedly exacerbated the features of shock and the associated coagulopathy (Harper et al, 1998). In vitro experiments showed that more than 50% of the Pittsburgh variant underwent proteolytic cleavage rather than complex formation in the presence of thrombin, and immunoblot of plasma from treated animals revealed the presence of significant amounts of the cleaved protein. Previous studies have demonstrated that the proteolytically cleaved serpins have pro-inflammatory properties and therefore it is probable that the cleaved Pittsburgh variant enhanced the inflammatory response and contributed to the rapid deterioration in treated animals (Banda et al, 1988; Corbin et al, 1990; Kurdowska & Travis, 1990). While latent AT is different to cleaved AT, this data clearly illustrates the potential hazards associated with the administration of conformationally altered serpins to patients with sepsis. Further studies are required to determine the concentration of l-forms of AT in different concentrates and to assess their effect, if any, on the host inflammatory response. The decision to treat with AT therapy should be based on the clinical status of the patient and the predicted mortality. The subgroup of patients with multiple organ failure and high-predicted mortality will probably benefit most from AT supplementation. While an argument may be made in favour of AT therapy in these patients, there is no justification at the present time for the use of this expensive, unproved and unlicensed blood product in the management of patients with low predicted mortality. It is difficult to identify a threshold level of plasma AT below which replacement therapy should be considered. Plasma levels of AT will invariably be reduced in patients with severe sepsis in proportion to the severity of the disease, however, the level will also reflect the time-point in the evolution of the inflammatory process at which the sample was collected (Fourrier et al, 1992; Lorente et al, 1993; Martinez et al, 1999). Furthermore, the circulating plasma level may not accurately reflect the concentration of endothelial AT, which appears to be critical for the protective effects of AT in sepsis. Therefore, until further data is available, the absolute level of plasma AT cannot be used to determine the appropriateness of AT therapy, although the plasma level should, at least, be reduced before treatment is considered. Prior to the administration of AT, consent should be obtained from the next of kin, pretreatment blood samples should be stored for baseline virology studies and hypofibrinogenaemia and thrombocytopenia should be corrected, as previously discussed. Pre-clinical studies suggest that AT therapy is only of benefit when given early in the course of the inflammatory process and at doses required to achieve supraphysiological plasma levels. Therefore, once the decision has been made to use AT, patients should be treated as early as possible. A suggested treatment schedule is a loading dose of 100 IU/kg, followed by daily treatment by continuous infusion to achieve AT plasma levels of approximately 100–150 U/dl. Prolonging treatment beyond 5 d of intensive therapy unit (ITU) admission will probably not be associated with additional benefit and may prove to be prohibitively expensive. The administration of heparin is of unproven benefit and may even negate the potential benefits of AT therapy, as illustrated in the porcine and guinea pig models of sepsis (Spannagl et al, 1991; Kessler et al, 1997). Antithrombin is a natural anticoagulant that plays a pivotal role in coagulation homeostasis. Recent data suggest that this glycoprotein also has potent anti-inflammatory properties and is protective in animal models of sepsis. The molecular mechanism responsible for the anti-inflammatory effect has not been clearly established, but probably relates to the local inhibition of the pro-inflammatory properties of thrombin and FXa by endothelial AT, and the increased release of prostacyclin from endothelial cells. While the results of randomized clinical trials are encouraging, there is currently insufficient evidence to support the use of AT therapy in severe sepsis, and larger studies are required to define the role of this treatment strategy. We would like to express our appreciation to Robin Carrell for his helpful comments.