Interleukin 6 and Haemostasis

Abstract

Interleukin 6 (IL-6) is a 26 kDa protein produced by many cell types including activated monocytes and macrophages, endothelial cells, adipose cells and the Th-2 subset of T-helper cells (Aarden et al, 1987; Jirik et al, 1989; Mohamed-Ali et al, 1997; Laharrague et al, 2000). After interaction with a specific saturable receptor, present on a great variety of responsive cells, it promotes a range of activities (Fig 1 and Table I) including antiviral effects (Sehgal & Sagar, 1980), enhancement of proliferation of haemopoietic progenitors (Wong et al, 1988) and induction of acute-phase reactant release from hepatocytes (Gauldie et al, 1990) – a feature shared with other cytokines, collectively known as the IL-6 cytokine family. It is also this last action that leads to the production of C-reactive protein (CRP), a useful surrogate marker of IL-6 (Heinrich et al, 1990). IL-6 may also be involved in the pathogenesis of disease as has been demonstrated in myeloma (Kawano et al, 1988). Summary of some of the diverse stimuli, sites of production and effects of IL-6. The subject of this review, however, is the role of IL-6 in haemostasis, which is mediated through a series of differing effects on endothelial cells, leucocytes and hepatocytes with promotion of synthesis of coagulation factors such as fibrinogen, tissue factor and factor VIII (Amrani, 1990; Cermak et al, 1993; Stirling et al, 1998), and also through stimulation of platelet production (Burstein, 1997). IL-6 was discovered almost simultaneously in 1986 by a number of groups who had been thought to be investigating different proteins – a factor related to its diverse activity (Hirano et al, 1986; Zilberstein et al, 1986). It consists of 212 amino acids with a hydrophilic signal sequence of 28 residues (Hirano et al, 1986). The gene for IL-6, mapped to chromosome 7p21 (Bowcock et al, 1988), consists of four introns and five exons, and has three transcriptional initiation sites (Zilberstein et al, 1986; Yasukawa et al, 1987). IL-6 expression in different cell types is regulated in response to a number of stimuli including endotoxins, IL-1, tumour necrosis factor-α (TNF-α), IL-4 and interferon-γ (IFN-γ) (Van Damme et al, 1987; Shalaby et al, 1989; Howells et al, 1991). Different cell types may respond differently to these stimuli. Indeed, IL-6 expression may be promoted by some cytokines in certain cell types, whereas the same cytokine may inhibit IL-6 expression in other cell types (Howells et al, 1991). The interleukin-6 receptor (IL6-R) is a specific receptor for IL-6 (Fig 2). It is a 468 amino acid protein that has three main regions (Yamasaki et al, 1988; Kishimoto et al, 1995). The first is an outer immunoglobulin-(Ig) like domain that is not required for IL-6 binding (Yawata et al, 1993). The remaining extracellular portion binds IL-6 and is composed of two fibrinonectin type III-like modules (Bazan, 1990). An intracytoplasmic domain, which is also not necessary for IL-6 binding, completes the protein (Yamasaki et al, 1988). Mechanisms of interaction of IL-6 with its receptor. (1) Cells with the IL-6R may bind IL-6 directly inducing homodimerization of gp130 leading to signal transduction. (2) Alternatively, the outer domain of the IL-6R may be shed after proteolytic cleavage by protein kinase C to form soluble IL-6R which may also bind IL-6. (3) The IL-6–soluble IL-6R complex can then cause homodimerization of gp130 and subsequent signal transduction in cells that lack the IL-6R (Kishimoto et al, 1995; Peters et al, 1998). As the IL-6R lacks a tyrosine kinase domain, an alternative method of signal transduction after IL-6 binding must occur. This led to the discovery of an accessory signal transducer, gp130, a 130 kb glycoprotein (Taga et al, 1989) that also acts as a signal transducer for other cytokines such as IL-11. Binding of the IL-6/IL−6R complex to gp130 induces its homodimerization and this is followed by the tyrosine-specific phosphorylation and subsequent activation of a variety of transcription factors (Kishimoto et al, 1995). Several examples of such transcription factors that are involved in fibrinogen gene expression are described below. Soluble IL-6R has subsequently been discovered, and it is formed as a result of cleavage by protein kinase C from the membrane-bound receptor protein (Mullberg et al, 1993). The IL-6R is only expressed on certain cell types, e.g. hepatocytes and B cells (Yamasaki et al, 1988), whereas not on others, e.g. endothelial and haemopoietic cells (Peters et al, 1997;Romano et al, 1997). In contrast, gp130 appears to be on all cells with relatively constant expression (Hibi et al, 1990). This allows cells lacking the IL-6R, which cannot be stimulated by IL-6 alone, to be stimulated by circulating IL-6–soluble IL-6R complex (Peters et al, 1998). Thus IL-6 alone will not stimulate endothelial cells directly, whereas in the presence of soluble IL-6R it will cause homodimerization of gp130 leading to cellular activation. Therefore, soluble IL-6R must be added for in vitro studies of endothelial cell response to IL-6 (Romano et al, 1997). IL-6 promotes coagulation without affecting fibrinolysis. It has been shown that activation of the coagulation cascade, as measured by thrombin–antithrombin (TAT) complex levels increases after infusion of IL-6 to patients with renal carcinoma (Stouthard et al, 1996). In contrast, levels of tissue plasminogen activator (t-PA), plasminogen activator inhibitor (PAI-1) and plasmin-α2–antiplasmin (PAP) complexes are unaffected suggesting that IL-6 does not affect fibrinolysis (Bergonzelli & Kruithof, 1991; Stouthard et al, 1996). IL-6 promotes haemostasis through a number of pathways (summarized in 3, 4) and the individual mechanisms are described below. Pathways by which IL-6 contributes towards haemostasis – an overview. IL-6 increases platelet production and enhances platelet activation. IL-6 promotes the coagulation cascade through a number of effects which are summarized in Fig 4. Vessel wall injury or inflammation stimulate IL-6 synthesis by endothelial cells. Endothelial cells also respond to IL-6 through the soluble IL-6 receptor. Thus, through these cumulative mechanisms IL-6 promotes haemostasis. Pathways by which IL-6 influences the coagulation cascade. Broken arrows indicate direct action of IL-6. Negative effects are accompanied by the (–) symbol; all other effects are positive. IL-6 increases tissue factor production by monocytes and factor VIII transcription in hepatocytes. Serum levels of VWF rose and protein S fell after administration of IL-6 in a canine model. IL-6 increases antithrombin production by hepatocytes and increases fibrinogen production by a variety of cells. Fibrinogen is a large dimeric protein, each half consisting of three polypeptides – Aα, Bβ and γ which are encoded by separate genes. The transcription of fibrinogen can be significantly upregulated by IL-6 (Amrani, 1990). Hexanucleotide CTGGGA clusters in the promoter region of the fibrinogen Aα− and γ-chain genes have been recognized as being IL-6-responsive elements (Liu & Fuller, 1995; Zhang et al, 1995). Studies of the fibrinogen γ-chain gene show that the signal transducer and activator of transcription, STAT 3, associates with the CTGGGA hexanucleotide promoter (Zhang et al, 1995). As its name suggests, STAT 3 carries out the dual functions of signal transduction and transcription activation (Zhong et al, 1994). It may become activated through phosphorylation in response to IL-6 or other stimuli such as epidermal growth factor (EGF). Interestingly, IL-1β which stimulates IL-6 synthesis (Van Damme et al, 1987) has also been shown to inhibit IL-6-mediated rat γ-fibrinogen gene expression by blocking STAT 3 binding (Zhang & Fuller, 2000). IL-4, IL-10 and IL-13 have also been shown to down-regulate the biosynthesis of fibrinogen by inhibiting the effect of IL-6 on hepatocytes (Vasse et al, 1996). This illustrates the complicated and not yet fully understood interacting cytokine effects on haemostasis. More recently, the transcription factor serum amyloid A activating factor (SAF)-1 has also been demonstrated to induce expression of the γ-fibrinogen promoter in response to IL-6 (Ray, 2000). In the case of Aα-fibrinogen, a novel DNA binding protein is associated with the CTGGGA site when it is not stimulated by IL-6. IL-6 stimulation causes this protein to leave the CTGGGA site and bind 12 basepairs downstream. This protein then re-associates with the CTGGGA site 1–2 h later. Through this effect on the kinetics of the protein, IL-6 controls Aα-fibrinogen transcription (Liu & Fuller, 1995). Most fibrinogen is synthesized in the liver but other sources such as epithelial cells have also been identified and these are also IL-6 responsive (Haidaris, 1997). Peak levels of fibrinogen tend to occur in the morning, which has been associated with an increase in ischaemic events, and this may relate to the circadian rhythm of IL-6 levels (Kanabrocki et al, 1999). The relationship between IL-6, fibrinogen and thrombotic disease is further discussed below. Tissue factor is a potent promoter of thrombin generation through direct activation of factor X by the tissue factor–factor VIIa complex (Osterud & Rapaport, 1977). IL-6, in a concentration of 100 ng/l, has been shown to increase tissue factor mRNA 4·5-fold and also increase the surface expression of tissue factor on cultured monocytes (Neumann et al, 1997). Effects of IL-6 on tissue factor may also be mediated indirectly through CRP, an acute-phase protein released by hepatocytes after IL-6 stimulation (Heinrich et al, 1990). CRP at concentrations found during inflammation has been shown to increase tissue factor procoagulant activity 75-fold in studies using monocyte cultures (Cermak et al, 1993). This is an example of how IL-6 can act synergistically to influence haemostasis through two pathways. Moreover, other cytokines have been shown to contribute to tissue factor upregulation including IL-1 and TNF-α in endothelial cells and monocytes, and IFN-γ and IL-8 in monocytes (Dosquet et al, 1995; Mulder et al, 1996; Osnes et al, 1996; Neumann et al, 1997; ten Cate et al, 1997). Other cytokines act to oppose such effects. IL-4, IL-10 and IL-13 have been shown to inhibit IL-1-induced tissue factor synthesis (Osnes et al, 1996) and IL-10 has also been shown to inhibit IFN-γ and TNF-α production during experimental endotoxaemia (Marchant et al, 1994). This illustrates a small part of the complexity of cytokine interactions and their effects on haemostatic balance. We have reported that transcription of the factor VIII gene is also promoted by IL-6 (Stirling et al, 1998). Levels of mRNA increased six- to ninefold basal values in liver cell lines after stimulation with IL-6. In contrast, other cytokines studied (IL-1 and IL-2) did not significantly affect factor VIII gene transcription. The transcription factors NFκB and C/EBP have been shown using mutational analysis to be necessary for increased factor VIII mRNA transcription in the acute-phase response as stimulated by lipopolysaccharide (LPS) (Begbie et al, 2000). Further study is required to elucidate whether these transcription factors are involved in the IL-6-mediated response and/or response to other cytokines associated with the acute-phase response such as IL-1 and TNF-α. Indeed, many protein binding sites have been identified within the factor VIII gene promoter region (Figueiredo & Brownlee, 1995; McGlynn et al, 1996). Thus, several proteins may act/interact with involvement of a variety of transcription factors to regulate factor VIII synthesis. In a canine model, IL-6 caused an increase in von Willebrand factor (VWF) and a decrease in protein S, the co-factor for protein C inactivation of factors V and VIII (Burstein et al, 1996). We have also shown in man that after infusion of desmopressin (1-desamino-8-arginine vasopressin) there is an increase in the plasma concentration of IL-6, and that this occurs after the increase in VWF. This rise in IL-6 is not secondary to the VWF rise, as in patients with severe VWD the IL-6 response is retained although there is no increase in VWF (Newby et al, 2000). The synthesis of the potent inhibitor of coagulation, antithrombin (AT) may also be influenced by IL-6. AT production has been shown to be inhibited by both IL-6 and IL-1 in cultured hepatocytes (Niessen et al, 1997) and acquired AT deficiency may occur in sepsis as part of the acute-phase response (White & Perry, 2001). IL-6 contributes further to haemostasis by enhancing platelet production as elegantly reviewed by Burstein (1997). IL-6 has been shown to promote megakaryocyte maturation in the absence of other growth factors. Also, it possibly stimulates megakaryocyte proliferation. Administration of IL-6 has been shown to result in a significant increase in the platelet count. IL-6 also alters platelet function and enhances thrombin-induced platelet activation. Thus by stimulating the mechanisms of both primary and secondary haemostasis, IL-6 appears to be an important mediator of the development of a stable fibrin clot. Endothelial cells are central to haemostasis through a number of vasoregulatory, coagulation and thrombolytic mechanisms. In the healthy resting state, endothelial cells inhibit haemostasis through production of the anticoagulants AT, tissue-factor pathway inhibitor (TFPI), membrane-bound thrombomodulin (which converts thrombin from a pro- to an anticoagulant role) and through vasodilators such as nitric oxide (NO) and prostaglandins. Endothelial cells also produce tissue plasminogen activator (t-PA). Endothelial procoagulants include VWF and thrombospondin, which promote platelet adhesion and aggregation, and endothelin, which causes vasoconstriction. In addition, production of plasminogen activator inhibitor-1 (PAI-1) by endothelial cells inhibits fibrinolysis (Pearson, 1999). Recent evidence suggests that factor VIII may also be synthesized by endothelial cells (Do et al, 1999). Although this remains controversial, it is supported by the observation that endothelial cells synthesize and release factor V (Cerveny et al, 1984), which has a major degree of strucutural homology with factor VIII, and which also has a very similar co-function in enzyme reactions on the platelet surface. After activation of endothelial cells, and in some disease states, there is increased production of pro-coagulants and decreased production of anticoagulants. As a result, the balance of endothelial cell activity moves to favour haemostasis (Fig 5) (Gross & Aird, 2000). Cytokines play a major role in the control of endothelial cell effects on haemostasis. After stimulation with endotoxins, TNF-α and IL-1, endothelial cells respond by producing a number of cytokines and growth factors including IL-1, IL-5, IL-6, IL-8, IL-11, IL-15, granulocyte colony-stimulating factor (G-CSF), macrophage colony-stimulating factor (M-CSF) and GM-CSF. These in turn not only change the balance in haemostatic factors as above, but also activate platelets and leucocytes. IL-1 and IL-6 appear to be particularly responsive to the local microenvironment (Bierhaus et al, 2000). IL-6 is not only produced by endothelial cells, they may also respond to this cytokine, for example, by upregulating protein S synthesis (Hooper et al, 1997). It has been shown recently that elevated levels of IL-6 are independently linked to markers of endothelial cell activation such as tPA, thrombomodulin and VWF (Yudkin et al, 1999). Heterogeneity in endothelial cell function in different tissues cannot be discerned in these in vivo studies, and it is only possible to speculate about their relationship with local thrombotic disease such as myocardial infarction. However, in vitro studies of endothelial cells from different tissues are limited by the need for soluble IL-6 receptor and also the removal of these cells from their normal microenvironment through which many interactions may occur. The effects of endothelial cell activation. Following activation the balance of endothelial cell activity moves to favour haemostasis. IL-6 is both secreted by endothelial cells and activates endothelial cells. AT-antithrombin; TFPI-tissue factor pathway inhibitor; TM – thrombomodulin; NO – nitric oxide; PG – prostaglandin; TF – tissue factor; VWF – von Willebrand factor; ET – endothelin; TS – thrombospondin A number of other related cytokines which have affinity for the gp130 receptor and which promote the synthesis of acute-phase proteins have been described and together with IL-6 are known as the IL-6 cytokine family (Taga & Kishimoto, 1997). These include leukaemia inhibitory factor (LIF), oncostatin M (OSM), ciliary neurotrophic factor (CNTF), cardiotrophin-1 (CT-1) and IL-11. These cytokines may also act to promote haemostasis. For example, OSM has been shown to induce hepatocyte fibrinogen biosynthesis and human umbilical vein endothelial cell (HUVEC) secretion of PAI-1 and VWF (Pourtau et al, 1998). In mice, administration of intravenous IL-11 has been shown to significantly increase VWF and factor VIII, which appears to be mediated through an effect on secretion rather than transcription (Denis et al, 2001). However, other actions of cytokines of the IL-6 family may actually be antagonistic to IL-6. For example, IL-11 and LIF inhibit adipocytes, whereas these cells are an abundant source of IL-6 (Ohsumi et al, 1991). Serum IL-6 levels increase with age as reviewed by Ershler & Keller (2000). In contrast, no apparent effect of ageing is seen for the cytokines TNFα and IL-1β (Roubenoff et al, 1998). An increase in IL-6 is particularly found in women after the menopause (McKane et al, 1994; Kania et al, 1995). In pregnancy, IL-6 levels rise significantly in the second trimester, and there is a further rise with the onset of labour (Opsjln et al, 1993). Ageing and pregnancy are both well established risk factors for thrombotic disease. The rise in serum IL-6 seen may be an additional contributing factor to the multiple established factors that are implicated in the pathogenesis of thrombotic disease in these groups. The possible role of IL-6 in thrombotic disease, particularly myocardial infarction, has generated a great deal of interest recently and there are several excellent reviews on this subject (Tracy, 1999; Yudkin et al, 2000). IL-6 has other effects, in addition to modulating haemostasis and endothelial cell function, which may contribute to the development of coronary heart disease. For example, IL-6 reduces circulating lipoprotein lipase concentration and has been shown to cause an increase in fatty lesion size in the arteries of a murine model (Huber et al, 1999). In addition, IL-6 increases CRP levels, which then binds to lipoprotein leading to complement activation, and this may contribute to atherogenesis (Bhakdi et al, 1999). Elevated IL-6 is also associated with insulin resistance (Yudkin et al, 1999). Some previously well known risk factors for ischaemic heart disease may be linked to IL-6. Adipose tissue has been shown to contribute as much as one third of the total circulating IL-6, and it has been hypothesized that obesity may resemble a low-grade inflammatory disorder. Stress may also affect IL-6 levels as catecholamines may stimulate IL-6 release from adipose tissue (Yudkin et al, 1999). A guanine to cytosine polymorphism at position 174 in the IL-6 gene promoter has been discovered in patients with juvenile chronic arthritis. Those individuals with the 174 cytosine allele show reduced IL-6 expression after stimulation with endotoxin or IL-1 and lower basal levels of IL-6 (Fishman et al, 1998). These individuals may be protected from thrombotic disease. Several studies have demonstrated an association between IL-6 and/or CRP and ischaemic heart disease (IHD). These are summarized below and in Table II. In the elderly, IL-6 concentrations in the top quartile (≥ 3·19 pg/ml) are associated with a relative risk (RR) of mortality from both IHD and all causes of 1·9 times that for concentrations in the lowest quartile (Harris et al, 1999). This study also demonstrated a RR of 1·9 for CRP in the top versus the bottom quartile of the normal range. If both IL-6 and CRP are in the top quartile, the RR of mortality rises to 2·6. IL-6 has also been shown to correlate significantly with risk factors for cardiovascular disease: age, fibrinogen, white cell count, plasma and blood viscosity, and smoking (Woodward et al, 1999). CRP in the top quartile of the normal range has been shown to be associated strongly with coronary heart disease and coronary heart disease risk factors – age, smoking, increased body mass index (BMI), increased cholesterol and Helicobacter Pylori and Chlamydia infection in middle aged men (Mendall et al, 1996; Koenig et al, 1999). CRP in the top quartile was also associated with increased incidence of myocardial infarction (RR = 2·9) and ischaemic stroke (RR = 1·9) but not venous thrombosis in a follow-up study of healthy men (Ridker et al, 1997). This study also showed aspirin to be most beneficial in those with high CRP. We have shown a significant elevation of IL-6, CRP and soluble intercellular adhesion molcule-1 (sICAM-1) in men presenting with chest pain (O'Malley et al, 2001). ICAM-1 is expressed on activated endothelium to facilitate localization of leucocytes to the site of injury and its upregulation may be under the influence of IL-6. Although, in this small study, a relationship to short-term cardiovascular outcome was not identified for any of these markers, interestingly, at 3 month review, IL-6 and CRP levels had fallen but ICAM-1 levels remained unchanged. ICAM-1 may be an alternative, superior marker to IL-6 for prediction of risk in IHD. The Multiple Risk Factor Intervention Trial (MRFIT) has shown CRP to be associated with coronary heart disease mortality in high-risk patients independent of other risk factors (RR = 4·3 for fourth versus first quartile). An elevated CRP was a particularly poor prognostic marker in smokers (Kuller et al, 1996). Patients with angina who have a CRP in the upper quintile of normal have been found to have a twofold increase in risk of having a coronary event compared with those in the first four quintiles (Haverkate et al, 1997). An increase in risk of peripheral vascular disease has also been found to be associated with an elevated CRP (RR = 4·1 for fourth versus first quartile) (Ridker et al, 1998). As discussed earlier, IL-6 increases transcription of fibrinogen and factor VIII, both of which are associated with an increased risk of arterial thrombosis. A fibrinogen level in the upper quintile is associated with an increased risk of coronary heart disease (RR = 2·1), cerebrovascular disease (RR = 1·3) and mortality at 2·5 years follow-up (RR = 5·8), and a factor VIII in the upper quintile is associated with an increased risk of coronary heart disease (RR = 1·5) and mortality (RR = 1·8) (Kannel et al, 1987; Tracy et al, 1999). There has also been much interest in a link between infections such as Chlamydia pneumoniae, IL-6 and coronary heart disease (Mlot, 1996; Capron, 1996). Human vascular smooth muscle and endothelial cells infected with C. pneumoniae have been shown to have increased IL-6 expression, which appears to be mediated through upregulation of the transcription factor NF-κB (Dechend et al, 1999; Rodel et al, 2000). Tissue factor and PAI-1 were also found to be increased in infected cells. Furthermore, treatment of C. pneumoniae infection results in improvement of the inflammatory markers IL-1, IL-6, CRP and TNF-α (Anderson et al, 1999). Further studies are required to elucidate whether this results in a reduction in coronary heart disease. Other Chlamydia strains and Rickettsia rickettsii have also been shown to induce expression of pro-coagulants (Fryer et al, 1997; Shi et al, 1998). Although Ridker et al (1997) did not find an association between IL-6 and venous thrombosis), an association has been suggested using a baboon model. After experimentally induced thrombosis, elevated IL-6 levels were associated with distal fibrin formation measured using radiolabelled fibrinogen scanning and thrombus extension assessed using venography (Wakefield et al, 1993). Although it cannot be determined clearly from this study whether elevation of IL-6 was the cause or effect of extension of thrombus, no significant association was found with thrombus formation for the other cytokines studied, IL-8 and TNF-α. A further study has shown IL-6 levels to be up to 100-fold higher in drainage fluid than peripheral blood after abdominal surgery, and that elevated levels were associated with increased post-operative complications including deep venous thrombosis (Sakamoto et al, 1994). Similar findings were not found in this study for other acute-phase proteins, IL-1 and TNF-α. As described earlier factor VIII transcription is promoted by IL-6. Factor VIII levels have been defined recently as a major risk factor for primary and recurrent venous thrombo-embolism (Kraaijenhagen et al, 2000; Kyrle et al, 2000). These studies have aimed to exclude any effect of the acute-phase response during which IL-6 levels (and therefore thrombotic risk) may be even higher. As IL-6 has been shown to be central to haemostasis, interventions to modulate its effects may have therapeutic potential. Options include increasing IL-6 to improve baseline coagulation factors in mild haemophilia and VWD. This may reduce the use of plasma-derived clotting factor concentrates as has been demonstrated with desmopressin (Mannucci, 2000). Alternatively, reducing synthesis or blocking the function of IL-6 may be useful in thrombotic disease. However, neither of these approaches is without significant potential problems. Because of the diversity of function of IL-6, any alteration in its levels will also affect haemopoietic, immune and acute-phase responses. Thus, any therapy would need to be more specific – for example, blocking the effect of IL-6 on fibrinogen and factor VIII transcription. We have shown desmopressin to cause a rise in plasma IL-6 levels, which correlates with a rise in VWF and factor VIII (Newby et al, 2000). This rise occurs too quickly for IL-6-mediated transcription, and other mediators may be involved. IL-6 synthesis may be reduced by essential fatty acids, ethanol and pentoxifyllin (McCarty, 1999). This may explain the cardioprotective role associated with these substances. Aspirin may have an anti-inflammatory as well as an antiplatelet role in the prevention of myocardial infarction. Healthy men with CRP in the highest quartile were shown to have a 55·7% reduction in risk of myocardial infarction in a double-blind placebo control trial of aspirin versus placebo compared with a reduction of only 13·9% in those with CRP in the lowest quartile (Ridker et al, 1997). IL-6 induces a prothrombotic state by increasing expression of tissue factor, fibrinogen, factor VIII and VWF, activation of endothelial cells and increasing platelet production and by reducing the levels of inhibitors of haemostasis such as antithrombin and protein S. Although currently limited to use for research purposes, measurement of IL-6 may become a useful clinical tool in evaluating thrombotic risk. Mechanisms to reduce IL-6, or block its action, offer the potential for novel therapies to reduce the risk of venous thromboembolism and inhibit the progression of atherosclerosis. We are very grateful to Dr Chris Prowse for his helpful comments.