High-Dose Antithrombin Therapy for Sepsis: Mechanisms of Action

INTRODUCTION Although extensive antithrombin (AT) supplementation has been shown to attenuate extent of organ failure and reduce the duration of disseminated intravascular coagulation (DIC) in septic patients, it has not yet been possible to show an improved survival rate (1–3). The concept of these sepsis trials is unique; that is, the natural coagulation inhibitor AT improves not only coagulation abnormalities or DIC, but also improves septic organ dysfunction (4). However, for that purpose, the physiological level of AT is insufficient, and 140–160% of the normal activity, which is obtained after administration of pharmacological dose of AT, is required to attain the full efficacy (5). Because this efficacy of AT to improve organ dysfunction and survival cannot be explained by its anticoagulatory effect itself, the so-called “anti-inflammatory effect” has recently been advocated (6). Although success in the animal models is no guarantee of success in the clinic, former reports on the efficacy of AT in clinical sepsis suggest that the results of the studies in animal models can be translated into clinical efficacy (7,8). Furthermore, the experimental investigations of AT in septic models have fulfilled all the recommended steps proposed by a recent expert panel on animal models in sepsis (9). Therefore, in this article, we summarize information from experimental data that is currently available regarding the “antiinflammatory effect” of AT, which includes effects on the coagulation system, PGI2 production, cytokine production, and leukocyte-endothelial interaction (Fig. 1).Fig. 1: Regulatory effect of AT. High-dose AT is expected to have so-called “anti-inflammatory effect.” These effects include the regulation of the coagulation system, stimulation in PGI2 production by endothelial cells, suppression of cytokine production from monocyte, and modulation of leukocyte-endothelial interactions.EFFECTS ON THE COAGULATION SYSTEM AT is a hepatically synthesized single-chain glycoprotein with a molecular weight of approximately 59 kDa. This natural serine protease inhibitor inhibits various steps of coagulation pathways because many of the clotting factors are serine proteases. AT forms a 1:1 complex with thrombin and inactivates its enzymatic activity. AT also inhibits contact factors XIIa and XIa, in addition to Factor IXa and Xa. This broad-spectrum enzymatic action of AT makes it a central inhibitor of the coagulation system. As is known in clinical use, AT contains a heparin-binding domain and its anticoagulation activity is enhanced several orders of magnitude after binding with heparin in the blood stream or glycosaminoglycans on the vascular endothelium (10). The anticoagulant activity of AT is reported to be maximized through the conformational changes after the binding with heparin or glycosaminoglycans on endothelium (11). It is generally accepted that hypercoagulability plays major roles in the development of septic organ failure. In this situation, AT activity is significantly reduced, and the decreased AT level relates to the severity of sepsis, DIC, and MODS (12,13). Furthermore, the correction of the coagulation abnormality can lead to a better outcome. Redens et al. (14) revealed that hematologic changes after lipopolysaccharide (LPS) challenge expressed by decreased fibrinogen levels and platelet counts, increases in prothrombin time and activated partial thromboplastin time were all attenuated by AT substitution. Our rat models, produced by continuous intravenous infusion of 5.0 mg/kg LPS for 24 h, showed AT activity of approximately 80%, which was approximately 40% lower than in normal rats. Among the treatment groups, one group received AT at a dose of 125 U/kg, which recovers the AT activity to the normal levels (low-dose group), and the other group received at a dose of 500 U/kg, which was sufficient to increase plasma AT activity to over 160% (high-dose group;Fig. 2, left). As a result of LPS challenge, thrombin-AT complex (TAT) levels were significantly elevated, and fibrinogen levels, an index of consumptive coagulopathy, were markedly decreased in the LPS control group. After AT supplementation, the elevated TAT levels were significantly reduced and the decrease of fibrinogen was less severe in both of the treatment groups. Because the fibrinogen levels of the low-dose and high-dose group were at a similar level, low-dose AT treatment appeared to be sufficient in terms of anti-coagulation (Fig. 2, right) (15).Fig. 2: Comparison of AT activity and fibrinogen in a rat LPS model. AT activity decreased to 80.1 ± 15.7% in the control group, whereas the activity reached 174.4 ± 12.0% with high-dose AT supplement, and 120.3 ± 13.2% with low-dose AT supplement (left). Decrease of fibrinogen level was significantly suppressed by supplementation of 500 U/kg AT(right). From Reference 15 with permission.EFFECTS ON SURVIVAL RATE AND ORGAN DYSFUNCTION The effect of AT substitution on the survival was first reported by Triantaphyllopoulos (16) in his endotoxemic rabbit. The similar effect was also confirmed in a primate model when Taylor et al. (17) reported survival benefit in Escherichia coli-infused baboons. Other than this, a protective effect of AT on mortality was reported in various models of sepsis (18,19). In our rat model, when treated with a continuous infusion of 10.0 mg/kg LPS over 24 h, all rats in the LPS control group died, 50% of the rats in the low-dose AT group survived, and all of the rats in the high-dose AT group survived. A similar pattern was seen when 5.0 mg/kg LPS was administered. Thus, improvement in the survival rate, in a dose-dependent manner, was ascertained (Table 1). In the next step, the effect of AT administration on organ dysfunction was investigated. Alanine aminotransferase (ALT), an indicator of liver dysfunction, markedly increased in the LPS control group, but this increase was attenuated in both treatment groups (Fig. 3, left). Blood urea nitrogen (BUN), an indicator of renal dysfunction, also increased in the LPS control group, and again the increase was reduced in the high-dose group (Fig. 3, right). In LPS controls, prominent pathological changes in the liver and kidneys were characterized by focal necrosis and glomerular thrombosis, respectively, and these changes were markedly less severe in the high-dose AT group (20). Other than the liver and kidney, Giebler et al. (21) reported decreases in intravascular fibrin deposition of the lung in endotoxin-challenged rabbits. Uchiba et al. (22,23) and Okajima (24) also reported the protective effects on pulmonary vasculature. Interestingly, although the anticoagulatory effect was saturated even in the low-dose AT treatment, the organ-protective effect was more prominent in the high-dose group. A similar phenomenon was reported from other laboratories (25,26). From these data, it is strongly suggested that the pharmacological level of AT obtained after high-dose infusion induces not only an anticoagulant but also an anti-inflammatory effect. This anti-inflammatory effect, which can be the key for sepsis treatment, will be discussed in the following part.Table 1: Comparison of survival rates in a rat LPS modelFig. 3: Comparison of ALT and BUN. Significant elevation in ALT level was observed in the control (1392.7 ± 1335.4 IU/L). This elevation was suppressed in both low-dose AT (256.7 ± 297.7 IU/L) and high-dose AT (174.3 ± 145.4 IU/L) groups (left). The elevation in BUN was significantly suppressed in the high-dose AT group (right). From Reference 20 with permission.EFFECTS ON LEUKOCYTE-ENDOTHELIAL CELL INTERACTION The mechanisms underlying better survival and organ function were investigated in observations of organ microvasculature. Again, rat models of sepsis were produced by the continuous infusion of LPS; in addition to LPS, only physiological saline was given to the LPS control group. The physiological AT activity (approximately 120%) was maintained in the low-dose AT group, and AT activity reached over 160% in the high-dose AT group. After the treatment, microvascular observations were made by intravital microscopy. After LPS treatment, the microvasculature in the liver showed reduced hepatic sinusoidal width, increased numbers of adhesive leukocytes, decreased sinusoidal blood flow, and intravascular coagulation. The treatment resulted in hepatic sinusoidal width, which was significantly greater in both low-dose and high-dose groups; by approximately 20% in comparison with the LPS control group (Fig. 4, left). Adhesive leukocytes were found three times as often in the LPS controls as in the high-dose AT treatment group (Fig. 4, middle). Sinusoidal blood flow was similar in the LPS control group and the low-dose AT group, but it was more than 1.5-fold greater in the high-dose AT group (Fig. 4, right) (27). These results suggested that high-dose AT treatment is useful in maintaining vascular endothelial function and in preventing leukocyte adhesion, in addition to preventing abnormal blood coagulation. It is known that these phenomena are not specific in sinusoid. Hoffmann et al. (28) reported the regulatory effect of AT on leukocyte adhesion in a skinfold of Syrian hamsters. The other laboratory confirmed our findings of similar effects in rat mesenteric microcirculation (29) (Fig. 5).Fig. 4: Comparison of sinusoidal width, adhesive neutrophil count, and blood flow. Both the low-dose and high-dose groups showed a significant increase in sinusoidal width compared with the control group (left). Adhesive neutrophil counts in a visual field in a 90-s period at the sinusoid were significantly suppressed in the high-dose group (middle). The high-dose group showed significant elevation in sinusoidal blood flow compared with either the control or the low-dose group (right). From Reference 27 with permission.Fig. 5: Changes in microcirculation 3 h after LPS injection. A marked adhesion of leukocytes and an aggregation of platelets are observed. These changes are followed by the swelling of endothelial cells with indistinct organellas, and the generation of microthrombi.One of the hypothesis underlying these phenomena is as follows: thrombin has been implicated in the inflammatory cascade (30), specifically to increase leukocyte recruitment (31). Thrombin increases the expression of endothelial P- and E-selectin (32,33) and consequently is likely to increase leukocyte rolling. In addition, thrombin induces the rapid expression of intercellular adhesion molecule-1 (ICAM-1) (33), thereby promoting leukocyte adhesion. Thrombin is known to mediate inflammatory reactions via specific receptors on platelets, endothelial cells, and white blood cells known as protease-activated receptors (PARs) (34). PARs are coupling-type receptors for protein G and they induce emergence of tissue factors, release of PAI-1, and down-regulation of thrombomodulin expression (35). The process of signal transduction after activation of PAR was shown to be the activation of NF-κB through the activation of C kinase and tyrosine kinase (36,37). After all, if thrombin were responsible for leukocyte-endothelial interaction and subsequent tissue injury, AT might have a potential therapeutic benefit in attenuating organ damage. Other than the anti-inflammatory effect through thrombin-PAR pathways, it is reported that AT induces anti-inflammatory effects via signal transduction cascade-coupled membrane receptors such as syndecan-4 (38). EFFECTS ON VASOACTIVE SUBSTANCES PRODUCED BY ENDOTHELIAL CELLS The reduced hepatic sinusoidal width observed in our rat models of sepsis was less marked with high-dose AT treatment. This suggests that AT produces a regulatory effect on vasoactive substance production. Endothelin 1 (ET-1) is known to be a major vasoconstrictor during endotoxemia, and recently, AT has received attention in relation to an apparent effect on the regulation of prostaglandin I2 (PGI2) production first in vitro (39,40). PGI2 is known to have beneficial effects both on hypercoagulability and inflammation (41). We, therefore, focused on the production of these endothelium-derived vasoactive substances. Rat models were treated with LPS, and again with AT at low- and high-dose levels by continuous infusion for 24 h; circulating levels of 6-keto-PG F1α, a stable metabolite of PG12, and ET-1 were measured thereafter. In response to LPS administration, 6-keto-PG F1α was increased approximately 2 fold and was further increased with AT treatment to a level approximately 1.5 times that in the LPS-treated control group (Fig. 6, left). In contrast, ET-1 was increased by LPS up to approximately seven times the normal level, however, this increase was not affected by AT supplementation (Fig. 6, right) (42). Uchiba et al. (23) also reported that PGI2 production increased in normal rats treated with AT in a dose-dependent manner. So far, theirs and ours are the only reports that elucidate AT's stimulatory effect of PGI2 production. However, in Uchiba's study, PGI2 production in rats treated with LPS and AT (250 U/kg) was no more than that of the treatment with LPS alone (43). Interestingly, in their other experiment, neither AT, whose heparin-binding ability was inhibited, nor AT substitution with heparin showed the stimulatory effect on PGI2 production (44). These observations suggested that the increased PGI2 production following AT administration may be induced through a receptor mechanism related to heparin-like glycosaminoglycan on the endothelial surface. The efficacy of heparin is still controversial. Coalson et al. (45) revealed that the use of heparin was revealed to decrease the endotoxin-initiated clotting, but was ineffective in preventing organ failure and death in the baboon model. On the other hand, Hermida et al. (46) showed that the simultaneous infusion of low molecular weight heparin and AT improved the coagulation abnormality and pathologic organ damage in their endotoxin-induced rabbit DIC model. In our model, because concomitant use of low dose heparin with the high-dose AT could further suppress the organ damage, we think heparin is effective only if the amount of AT is sufficient enough. Because it is generally impossible to use this sufficient amount of AT in clinic, the concomitant use of heparin is not recommended (7,47).Fig. 6: Comparison of 6-keto-PG F1α and endothelin 1. 6-keto-PG F1α level was significantly elevated in the control group (236.5 ± 41.9 pg/mL), and this level was further increased in the high-dose group (327.8 ± 84.8 pg/mL, left). In contrast, ET-1 level increased to approximately seven times higher compared with the normal level, and the level was not affected by AT supplementation (right). From Reference 51 with permission.Regarding the results of our study, the increased production of PGI2 by LPS was considered to be due to thrombin generation stimulated by LPS. After low-dose AT treatment, PGI2 levels were found to be similar in the LPS control group, and the levels were significantly higher after high-dose AT treatment. Increased PGI2 may explain the changes of sinusoidal width and the prevention of leukocyte adhesion. PGI2 is considered to have an anti-inflammatory effect because of the effects on vasodilation, inhibition of platelet aggregation, potent monocyte suppression, and inhibition of neutrophil adhesion to vascular endothelium (48). The fact that the inhibition of adhesion by AT is completely abolished by the pretreatment with the cyclooxygenase inhibitor indomethacin (47) supports the view that PGI2, plays major roles in the anti-inflammatory effect of AT. The finding that PGI2 production—already increased by thrombin—is further increased appears contradictory. However, it can be explained as follows: both thrombin and AT stimulate vascular endothelial cells through independent pathways. Thrombin, through the thrombin receptor on the cell membrane, stimulates PGI2 production in the endothelium (49), and AT, through heparin-like glycosaminoglycan (another receptor on the cell membrane), acts on vascular endothelial cells (50), thereby exerting synergistic effects on PGI2 production (Fig. 7).Fig. 7: Changes in PGI2 production during sepsis. The production of PGI2 from EC is stimulated by the binding of thrombin and the thrombin receptor on endothelial cells during severe sepsis. This stimulation is further upregulated by AT-GAGs binding. However, AT was largely diminished by extravasation, consumption by thrombin (TAT formation), and degradation by proteases during sepsis. Th, thrombin; GAGs, glycosaminoglycans.ET-1, which has a potent vasoconstrictive effect, was reported to be increased in severe sepsis (51), and might be involved in the worsening of organ dysfunction. We observed that ET-1 levels increased to seven times the normal level after LPS administration. It is possible that ET-1 production is regulated through thrombin in the same way as PGI2 (52). However, because ET-1 levels were not affected by the administration of AT, ET-1 production is likely to be regulated by some other factors, including cytokines, etc. (53). ET-1 is thought to act on Ito cells, resulting in the reduction of sinusoidal width, and thereby regulating blood flow (54). In our study, however, we did not observe the AT regulation of ET-1. Thus, AT appears to exert a vasodilatating effect through its regulation of a vasoactive substance other than ET-1. EFFECTS ON CYTOKINE PRODUCTION BY ENDOTHELIAL CELLS Jochum et al. (5), in their clinical investigation, found decreased TNF and IL-8 levels 2–3 days after the beginning of AT supplement therapy. Inthorn et al. (55) also observed decreased IL-6, soluble ICAM-1, and soluble E-selectin levels in patients treated with AT. However, whether these regulatory effects are direct or indirect is still unclear. Therefore, we investigated the effects of AT on cytokine production. The methods and results are as follows: rats were assigned to three groups and received either low-dose AT, high-dose AT, or saline alone (control group). Endotoxin (5.0 mg/kg) was infused for 3 h in all three groups. Blood samples were collected at 0, 3, and 6 h after endotoxin administration, and the serum levels of rat TNF, IL-4, and cytokine-induced neutrophil chemoattractant (CINC) were measured. TNF levels determined at 0 h in the high-dose AT group were less than one-half of those in the control group, although the difference was not significant (Fig. 8, left). No difference was seen in either IL-4 or CINC at any time point in three groups (Fig. 8, middle and right). Therefore, the inhibitory effect of AT directly on cytokine production was not apparent in this experiment (56). However, because TNF particularly tended to increase in the control group but with great variability, this remains to be investigated. Recently, Dickneite et al. (57) reported that no difference in TNF levels was seen in experiments in porcine models. On the contrary, Souter et al. (58) revealed that AT downregulates the production of proinflammatory cytokine from monocyte in a dose-dependent manner. Therefore, this issue remains to be discussed.Fig. 8: Changes of TNF, IL-4, and CINC levels after AT supplement. Although TNFα levels in the high-dose group (closed circle) at 0 h was approximately one-half that in the low-dose group (gray circle) and less than one-half that in the control group (open circle), the differences were not statistically significant (left). IL-4 level reached maximum at 3 h after endotoxin infusion. No significant difference was seen in plasma levels of IL-4 at any time point (middle). CINC level reached maximum at 6 h after endotoxin infusion. No significant difference was seen in plasma levels of CINC at any time point (right). From Reference 56 with permission.CONCLUSION The physiological level of AT (approximately 100%) strongly regulates coagulation abnormalities. In addition to this effect, the pharmacological levels of AT (140%–160%) can induce regulation of vasoactive substance production in the vascular endothelium and the regulation of leukocyte-endothelial cell interaction. The key substance of the latter effect is thought to be PGI2. PGI2 has favorable effects on sepsis, including vasodilation, anticoagulation, and anti-leukocyte adhesion, and these effects were recognized only at pharmacological levels, but not physiological levels of AT. High-dose AT therapy has the capability of improving clinical outcomes in cases of severe sepsis by favorably influencing coagulopathy, endothelial PGI2 production, and leukocyte-endothelial cell interaction.