Disseminated intravascular coagulation (DIC) is an acquired syndrome characterized by widespread intravascular activation of coagulation that can be caused by infectious insults (such as sepsis) and non-infectious insults (such as trauma). The main pathophysiological mechanisms of DIC are inflammatory cytokine-initiated activation of tissue factor-dependent coagulation, insufficient control of anticoagulant pathways and plasminogen activator inhibitor 1-mediated suppression of fibrinolysis. Together, these changes give rise to endothelial dysfunction and microvascular thrombosis, which can cause organ dysfunction and seriously affect patient prognosis. Recent observations have pointed to an important role for extracellular DNA and DNA-binding proteins, such as histones, in the pathogenesis of DIC. The International Society on Thrombosis and Haemostasis (ISTH) established a DIC diagnostic scoring system consisting of global haemostatic test parameters. This scoring system has now been well validated in diverse clinical settings. The theoretical cornerstone of DIC management is the specific and vigorous treatment of the underlying conditions, and DIC should be simultaneously managed to improve patient outcomes. The ISTH guidance for the treatment of DIC recommends treatment strategies that are based on current evidence. In this Primer, we provide an updated overview of the pathophysiology, diagnosis and management of DIC and discuss the future directions of basic and clinical research in this field.
Since it was described a half-century ago, the concept of disseminated intravascular coagulation (DIC) and its underlying pathogenesis have taken shape in accordance with the understanding of the mechanisms of blood coagulation and the advancement of laboratory tests1. At the beginning of the 1980s, Spero and colleagues2 correctly proclaimed that DIC is a sign that “death is coming”. Since then, DIC has been recognized as a serious, well-defined and life-threatening condition, which is elicited by diverse infectious and non-infectious insults.
The Scientific and Standardization Committee (SSC) on DIC of the International Society on Thrombosis and Haemostasis (ISTH) defined DIC as an acquired syndrome characterized by the intravascular activation of coagulation with a loss of localization arising from different causes. It can both originate from and cause damage to the microvasculature, which, if sufficiently severe, can produce organ dysfunction3. Importantly, this definition highlights that the core features of DIC are indicative of systemic thrombin generation that is not restricted to the site of insult and endothelial cell injury, which gives rise to organ dysfunction. Combined with these changes, inhibition of fibrinolysis synergistically results in microvascular thrombosis that — in concert with haemodynamic and metabolic derangements — contributes to organ dysfunction. DIC is therefore an independent predictor of mortality in critical illness3. On the basis of these insights into the pathogenesis of DIC, the ISTH has further contributed to the establishment of diagnostic criteria and better strategies for the clinical management of DIC3,4.
DIC is an old concept that still attracts a substantial amount of attention among physicians worldwide. In this Primer, we provide an updated overview of the pathophysiology, diagnosis and management of DIC and discuss the future directions of basic and clinical research in this field.
DIC is a frequent complication of a systemic inflammatory response syndrome (SIRS)5,6. SIRS can be caused by infectious insults (for example, sepsis, which is life-threatening organ dysfunction caused by a dysregulated host response to infection) and non-infectious insults (for example, trauma); indeed, sepsis and trauma are two predominant clinical conditions associated with DIC2,3. For example, two validation studies of the ISTH and the Japanese Association for Acute Medicine (JAAM) DIC diagnostic criteria revealed the prevalence of sepsis or infection (30–51% of patients) and trauma or major surgery (45% of patients) as underlying conditions of DIC7,8. Other important underlying DIC-associated disorders include organ destruction (such as severe pancreatitis), malignancy (solid tumours and haematological cancers), obstetrical calamities (such as amniotic fluid embolism, placental abruption, serious pre-eclampsia and postpartum haemorrhage), fulminant hepatic failure and severe toxic or immunological reactions2,3,8.
The incidence and mortality of DIC vary according to the period, country, place of treatment (ward or intensive care unit), diagnostic criteria and underlying disorders (Table 1). The differences in mortality between patients who are diagnosed according to the ISTH (a mortality rate of 46%) and the JAAM scoring system (a mortality rate of 22%) depend on the differences in the nature of the two diagnostic criteria7,8. The ISTH criteria diagnose full-blown DIC, whereas the JAAM criteria diagnose DIC that has not yet reached the stage of decompensation8. Rough estimations show an improvement in mortality due to DIC over the past two decades. For instance, a nationwide epidemiological survey by the Ministry of Health, Labour and Welfare of Japan showed that the mortality of patients with DIC was as high as 65% in 1992 but declined to 56% in 1998 (Refs 9,10). In addition, a study based on the Japanese national administrative database demonstrated that mortality due to DIC further declined to 46% between 2010 and 2012 (Ref. 11). This study also showed a significant decrease in mortality due to DIC in patients with infectious diseases in the same period. The same trend was observed in population-based studies in the United States, where mortality due to DIC decreased from 76% to 51% between 2006 and 2004–2010 (Refs 12,13). However, it is not clear whether these better outcomes are owing to better understanding of DIC or to improvements in the general management of patients who are critically ill.
The clinical progression from SIRS to sepsis to severe sepsis and septic shock has been recognized to accompany an increase in the incidence of DIC. Progression to DIC, in turn, leads to organ dysfunction, which is associated with increased mortality14. In fact, DIC is an independent predictor of morbidity, 28-day mortality and hospital mortality in patients with severe sepsis15. Subgroup analyses of large studies in patients with severe sepsis, such as the PROWESS and the KyberSept trials, showed that when activated protein C (APC) or antithrombin treatments were not allocated, the incidence of DIC was 28.9% and 40.7%, and the mortality was 43% and 40%, respectively16,17. A multicentre, prospective validation study of the JAAM DIC scoring system in patients with severe sepsis demonstrated that the incidence of DIC was 46.8% and the mortality in patients with both severe sepsis and DIC was 38.4%, a rate that was almost twofold higher than patients without DIC18. Moreover, other studies have shown that, in severely injured trauma patients, the incidence of DIC was 45–54% in patients diagnosed with the JAAM criteria19 and 9–17% in patients diagnosed using the ISTH criteria20. These studies also showed that the mortality of trauma patients with DIC ranged from 25–34%; these rates doubled in patients who met the ISTH criteria for DIC19,20. Finally, a Canadian study using the ISTH criteria found that the incidence of DIC in obstetrical calamities was as low as 0.03% from 1980 to 2009, whereas the DIC-related maternal mortality rate was 3%21. New data on the incidence of DIC in other underlying conditions have been insufficient.
Thrombi (blood clots) are formed as the result of coagulation and comprise fibrin and activated platelets. The coagulation cascade involves a series of proteolytic reactions in which inactive serine proteases become activated and, in turn, activate subsequent proteases in the pathway. According to recent insights, the traditional division of the coagulation system into an intrinsic and extrinsic pathway seems outdated. An abbreviated scheme demonstrating the contemporary model of activation of coagulation in vivo is shown in Fig. 1. Initiation of activation of blood coagulation occurs through the tissue factor–factor VII pathway (formerly known as the ‘extrinsic system’) and ultimately results in the generation of thrombin. Thrombin is the central protease in the activation of coagulation. Generation of thrombin is not only crucial for the conversion of fibrinogen into fibrin but thrombin also augments its own generation by activating various other coagulant enzymes and cofactors (such as factor VIII, factor IX and factor XI). In addition, thrombin is a potent agonist of platelet aggregation. The creation of crosslinked fibrin is the eventual step in the activation of coagulation. Thrombin-induced cleavage of small fragments from fibrinogen leads to the formation of fibrin monomers and, successively, polymers. To further strengthen the clot, crosslinking of fibrin is mediated by thrombin-activated factor XIII (factor XIIIa). Activation of coagulation is regulated by three main anticoagulant pathways: antithrombin, the protein C system and tissue factor pathway inhibitor (TFPI).
Pathogenetic pathways in DIC
Innate immunity and coagulation are closely related and regulate each other when activated by diverse insults22. These processes, which consist of inflammation, haemostasis and immunothrombosis, maintain body homeostasis and promote recovery from the insults22,23. However, severe insults perturb these control mechanisms, leading to the systemic activation of coagulation and the inflammatory cascade, followed by DIC. In turn, DIC gives rise to multiple organ dysfunction and affects patient outcomes. The mechanisms involved in the pathological derangement of coagulation in patients with DIC have become increasingly clear in recent decades. Specifically, it seems that various mechanisms at different sites in the haemostatic balance act simultaneously towards a procoagulant state. Although DIC is a complication of various underlying disorders (Box 1), once initiated, the mechanisms that lead to this coagulopathy follow similar lines.
Triggers of coagulation activation in DIC. In sepsis and trauma, the pathogenesis of DIC is triggered by the systemic inflammatory response, in which inflammatory cytokines are the most important mediators24. Increasing evidence supports that extensive crosstalk between inflammation and coagulation occurs, such that inflammation leads to the activation of coagulation and coagulation also considerably affects inflammatory activity25 (Fig. 2). Interestingly, some organ dysfunctions in DIC are specific to severe sepsis owing to the systemic activation of coagulation and inflammation that occurs in sepsis26. In other specific underlying disorders that cause DIC, the activation of coagulation can initially be triggered by other routes, such as the expression of procoagulant factors (including tissue factor or factor X-activating cysteine protease) in patients with cancer27 or the release of coagulation-initiating molecules in obstetrical calamities, such as placental abruption or amniotic fluid embolism28.
The principal initiator of thrombin generation in DIC is tissue factor. For example, it has been shown that a moderate systemic inflammatory challenge such as low-dose endotoxaemia in humans results in an 125-fold increase in tissue factor mRNA levels in blood monocytes and consequent activation of coagulation29. In addition, the expression of tissue factor on human monocytes can be induced in response to experimental systemic exposure to microorganisms30. In line with this finding, in animals challenged with microorganisms or lipopolysaccharides (which are found on the surface of Gram-negative bacteria), inhibition of the tissue factor–factor VIIa pathway by specific antibodies or agents that block the activity of tissue factor or factor VIIa attenuated both thrombin formation and coagulopathy, thereby decreasing mortality31,32. Similarly, in patients with severe trauma or cancer, studies have shown that DIC is triggered by the tissue factor–factor VIIa pathway33,34. In addition to monocytes, perturbed epithelial cells might be a source of tissue factor35,36. Furthermore, tissue factor may be present on the surface of other leukocytes, particularly neutrophils37, although it is doubtful whether these cells are capable of producing tissue factor38. It is more likely that other leukocytes acquire surface-bound tissue factor exogenously, such as from microparticles that are shed from activated mononuclear cells and possibly endothelial cells39.
Platelets have a pivotal role in the pathogenesis of coagulation abnormalities in DIC36. Platelets can be activated directly, for example, by pro-inflammatory mediators such as platelet-activating factor40. In addition, the expression of tissue factor results in the generation of thrombin, which may further activate platelets. The activated platelet membrane then forms a perfect scaffold on which further coagulation activation can occur41. Another pathway by which activated platelets may stimulate thrombin generation involves P-selectin. Platelets express P-selectin on their surface, which regulates the adhesion of platelets to leukocytes and the vascular endothelium and also boosts the expression of tissue factor on mononuclear cells42. This increased expression is caused by the binding of platelets to mononuclear cells and the subsequent activation of nuclear factor-κB (NF-κB). P-selectin is released from the platelet surface, and soluble P-selectin is a precise marker of systemic inflammation42. In addition, disruption of the endothelium both enhances platelet–vessel wall interactions and involves the substantial release of ultra-large von Willebrand factor (vWF) multimers from the endothelium. vWF is an important mediator of platelet adhesion and coagulation, and its degradation is usually catalysed by a disintegrin and metalloproteinase with thrombospondin motifs 13 (ADAMTS13). Relative insufficient cleavage of vWF multimers due to consumption of ADAMTS13 might contribute to DIC43. Indeed, ultra-large vWF multimers have been detected in patients with DIC and ADAMTS13 deficiency, and the association between low levels of ADAMTS13 and the severity of DIC in sepsis has been confirmed44,
Propagation of coagulation activation. During sepsis-induced activation of coagulation, the function of all three physiological anticoagulant pathways can be impaired. First, antithrombin, which forms complexes with and inhibits thrombin and factor Xa (Fig. 1), is one of the most important inhibitors of coagulation, and reduced levels of antithrombin are a characteristic feature of DIC. Reductions in the levels of antithrombin are caused by a combination of processes, including reduced protein synthesis, increased clearance through the formation of protease–antithrombin complexes, extravascular loss due to increased vascular permeability and degradation by neutrophil elastase36. In addition, heparin sulfate increases the activity of antithrombin, and, in DIC, cytokines might impair proteoglycan synthesis in the vessel wall and thereby reduce the availability of heparin sulfate47.
Second, APC and its cofactor protein S form an additional line of defence against the excessive activation of coagulation. Thrombin forms a complex with the endothelial cell membrane-associated molecule thrombomodulin, and this complex converts protein C to its active form, APC48. Furthermore, after binding to thrombin, thrombomodulin stimulates the activation of the thrombin activatable fibrinolysis inhibitor (TAFI), which impairs endogenous fibrinolysis and stimulates sustained fibrin deposition. APC proteolytically degrades factor Va and factor VIIIa, attenuating thrombin generation and fibrin formation (Fig. 1). Vascular endothelial cells express endothelial protein C receptor (EPCR), which binds to and enhances the activation of protein C at the cell surface49. In addition to its anticoagulant activity, APC exerts anti-inflammatory effects on leukocytes. Several studies have demonstrated anti-inflammatory effects of APC in vivo50. By contrast, impairment of the protein C system increases the severity of systemic inflammation and DIC. In clinical studies, decreased levels of protein C and protein S are associated with reduced survival51. Furthermore, abrogation of protein C activity by administration of C4-binding protein converted a sublethal sepsis model in baboons into a fatal model52. Inhibition of EPCR by specific antibodies also reduced survival in this septic baboon model53. Conversely, administration of protein C in this sepsis model prevented DIC and mortality. Hence, it seems that the protein C pathway is of crucial relevance in the host defence response that causes DIC.
A third coagulation-regulating system is based on TFPI. This inhibitor is present at the surface of the vascular endothelium or is bound to lipoproteins in the circulation and inhibits tissue factor that is in a complex with factor VIIa. Observations in patients with sepsis have not generated conclusive results regarding the relevance of this inhibitory system in DIC, as plasma concentrations of TFPI were not lower in most patients than in normal controls54. Despite this finding, two lines of evidence from animal models demonstrate the relevance of TFPI in DIC. First, deficiency of TFPI increased the sensitivity of rabbits to DIC induced by tissue factor55. Second, administration of TFPI diminished the detrimental effects of experimental bacteraemia in baboons. In this trial, TFPI not only prevented DIC but, in all animals that had received a lethal dose of Escherichia coli, TFPI treatment also resulted in a marked amelioration of vital functions and survival of the bacterial challenge56. In further support of a role for TFPI in DIC, a study in healthy humans showed the ability of TFPI to block endotoxin-induced coagulation57.
Experimental and clinical studies have demonstrated that, at the peak of thrombin generation in sepsis, endogenous fibrinolysis is almost completely turned off58. Plasminogen is the inactive form of plasmin, which is an enzyme that proteolytically degrades fibrin clots. The immediate fibrinolytic reaction to inflammatory mediators is a sharp increase in plasminogen activators (mainly tissue-type plasminogen activator (t-PA) and urokinase-type plasminogen activator (u-PA)) due to their release from the endothelium. However, this intensification of plasminogen activation and subsequent plasmin generation is followed by a persistent increase in the levels of plasminogen activator inhibitor 1 (PAI-1)59. This increase results in a sustained impairment of fibrinolysis in sepsis. Important experimental studies have confirmed the role of fibrinolysis in removing fibrin from various organs. In experimental sepsis models, fibrin deposition in the kidneys, lungs, liver and adrenal glands was greatly reliant on a reduction in the activity of plasminogen activator60.
Inflammation and coagulation in DIC. The derangement of coagulation and fibrinolysis in DIC is mediated by several cytokines. High cytokine concentrations have been detected in the circulation of patients with sepsis and coagulopathy and, in experimental models of sepsis, serum levels of these cytokines are increased36. In the course of sepsis, tumour necrosis factor (TNF) first reaches a peak, which is followed by a rise in serum levels of IL-6 and IL-1. Several studies have been performed to elucidate the roles of these cytokines in the pathogenesis of DIC.
As TNF is the first cytokine that peaks in experimental bacteraemia or endotoxaemia and has potent procoagulant properties in vitro, it was initially hypothesized that coagulation activation was caused by TNF. However, a trial that used various strategies to inhibit the activity of TNF showed that, although the effects of sepsis on coagulation inhibitors and fibrinolytic activity seemed to be mediated by TNF, complete blockage of sepsis-induced enhancement of TNF did not affect the activation of coagulation36. In addition, in a lethal bacteraemia model in baboons, an anti-TNF antibody had only a marginal effect on fibrinogen consumption61. Furthermore, studies in patients with sepsis who were administered an anti-TNF monoclonal antibody did not show any effect of this treatment on survival62. By contrast, the administration of a specific anti-IL-6 antibody resulted in the complete inhibition of lipopolysaccharide-induced activation of coagulation in primates63. Similarly, patients with cancer who received recombinant IL-6 showed a marked increase in the levels of thrombin64. Taken together, these results indicate that IL-6 rather than TNF is important in mediating the procoagulant response in DIC. IL-1 is also a potent stimulator of tissue factor expression in vitro; however, its role in vivo has not been elucidated. In support of a role for IL-1 in mediating DIC in sepsis, infusion of an IL-1 receptor antagonist partly inhibited the procoagulant response in an experimental sepsis model. In addition, administration of an IL-1 receptor inhibitor to patients with sepsis reduced thrombin generation65. However, as the endotoxin-induced effects on coagulation occur well before the levels of IL-1 become increased in the circulation, the question of whether IL-1 has a direct role in the coagulopathy associated with sepsis remains unresolved.
Activated coagulant factors and coagulation inhibitors not only interact with other coagulation proteins but also with specific cellular receptors that subsequently turn on signalling pathways. In particular, the binding of proteases to receptors that influence the activation of inflammation might be important in sepsis and DIC. The most relevant pathway by which coagulant proteases affect inflammatory activity is by binding to protease-activated receptors (PARs), which are transmembrane domain, G protein-coupled receptors66. An unusual property of PARs is that they act as their own ligand, which is in contrast to most other receptors. Binding and cleavage by an activated coagulant protease results in the exposure of a neo-amino terminus that in turn activates the same receptor (and possibly neighbouring receptors), causing transmembrane signalling. There are four different PARs: PAR1, PAR3 and PAR4 are thrombin receptors, whereas PAR2 is activated by factor Xa, the tissue factor–factor VIIa complex and trypsin. PAR1 can also act as a receptor for factor Xa and the tissue factor–factor VIIa complex.
Recent observations have pointed to an important role for extracellular DNA and DNA-binding proteins (such as histones and high mobility group protein B1 (HMGB1)) in the pathogenesis of DIC. This cell-free DNA and DNA-binding components are released from nucleosomes of degraded cells and might form a surface on which the assembly of activated coagulant factor complexes could be greatly facilitated67. In addition, histones activate platelets and stimulate thrombin generation68. The activation and binding of neutrophils by DNA components result in the formation of neutrophil extracellular traps (NETs), which have recently been identified as important contributors to vascular thrombosis and inflammation69. NETs mainly trap and kill pathogens70,71 by using their contents: histones, DNA and potent proteases. However, NETs have been found to promote excessive thrombosis by multiple mechanisms, including activation of factor XII72, inactivation of TFPI73 and provision of a mesh for platelet binding and aggregation74. NETs might also provide the availability of inflammatory cells that express tissue factor72. Activation of coagulation is further enhanced by the proteolytic cleavage of physiological anticoagulants by neutrophil elastase, which is abundant in NETs75. NETs might also induce endothelial cell death and detrimental inflammatory activity, effects that are probably mediated by NET-associated proteases or cationic proteins, such as histones69,76.
In addition to the role of inflammation in promoting coagulation, considerable crosstalk between regulatory anticoagulant systems and inflammatory mediators might occur. For instance, antithrombin can act as an anti-inflammatory mediator by directly binding to leukocytes and reducing their cytokine and chemokine receptor expression77. Indeed, in experimental animal models, the administration of antithrombin can reduce DIC intensity and mortality, and these effects are accompanied by a decrease in the levels of IL-6 and IL-8. In addition, convincing evidence supports that the protein C system plays a crucial part in modulating inflammation78. APC has been shown to block endotoxin-induced increases in the levels of TNF, IL-1β, IL-6 and IL-8 in vitro and in vivo79,80. In line with these results, the inhibition of APC in animals with experimental sepsis has been shown to aggravate the systemic inflammatory host response, as indicated by enhanced levels of pro-inflammatory cytokines, increased leukocyte infiltration and tissue destruction in various organs81. Furthermore, a study showed that targeted disruption of the gene encoding protein C in mice (causing a heterozygous protein C deficiency) resulted in a more severe coagulopathy after the administration of endotoxin than in wild-type mice and that this increase in coagulopathy severity was associated with a markedly increased inflammatory response, as shown by enhanced levels of several pro-inflammatory cytokines82.
Pathophysiology of organ dysfunction
As DIC is an intermediary condition that arises on the background of other disorders (such as sepsis, trauma, obstetrical calamities and cancer), it has frequently been classified as a ‘haemostatic complication’ of these underlying disorders83. Nonetheless, the observations that DIC increases the risk of mortality to levels that are higher than that of the initiating disorder and that DIC is an independent predictor of mortality regardless of the underlying condition84,85 suggest that the triggering of distinct processes that culminate in multiple organ dysfunction as a prelude to death represent a distinct pathological entity. In sepsis that is not complicated by DIC, for example, DIC increases the risk of death from 27% to 43%86. DIC is therefore a development that has net liability to the patient regardless of the inciting factors. This concept is further demonstrated and strengthened by the fact that removing or treating the primary disease does not necessarily result in a better outcome or indeed in the resolution of DIC87. Altogether, these points indicate that, although the DIC process can be triggered and shaped by an underlying disease-specific process, its pathogenetic effects and systemic dissemination that lead to multiple organ dysfunction and death are probably governed by somewhat independent pathophysiological pathways and mediators. Such pathways might include, for example, acute phase reactants from the liver that are often remotely linked to coagulation or to the inciting and triggering effectors. Thus, it is often a major challenge to identify the predominant mechanism that is driving DIC from the heterogeneous, overlapping effects of the condition that can vary with time. The pathology associated with DIC is further complicated by the presence of several intertwined feedback loops between coagulation, inflammation, complement and immune systems. This interplay between mechanisms tends to vary over the course of the clinical presentation and can be further exacerbated by the underlying disease-specific process. Emerging evidence also suggests a pivotal role for systemic cellular activation and death through the release of damage-associated molecular patterns (DAMPs), including nuclear breakdown products such as DNA and histones that can be found in NETs, among other locations, making the management of this devastating condition particularly challenging.
To understand the mechanisms of multiple organ dysfunctions in DIC, a clear appreciation of the main themes in the pathophysiology of DIC is required in the approach to the patient with DIC. These themes need to explain the mixed pathology, which includes microvascular thrombosis, haemorrhages and oedema caused by vascular leakage (summarized below).
Multifaceted consequences of thrombin generation in vivo. Regardless of the underlying disease-specific process, the excessive generation of thrombin and its systemic dissemination is usually the hallmark of DIC development88. This mainly relates to the central physiological role of thrombin and the multiple well-orchestrated coagulant89,90, anticoagulant89,91, profibrinolytic92 and antifibrinolytic93,94 functions of thrombin in the coagulation cascade (Fig. 3). The main theme in DIC pathophysiology is the loss of these opposing effects owing to the excessive generation of thrombin, which not only disrupts this balance but also results in the disproportionate consumption of these different components at rates that are difficult to predict, tend to vary over the course of the illness and may be shaped by the underlying inciting pathology95. One example of disease-specific tailoring of DIC is the loss of the protein C receptors (thrombomodulin and EPCR) from endothelial surfaces at vulnerable vascular sites, as has been shown in the case of DIC in cerebral malaria, which involves the cytoadherence of Plasmodium falciparum-infected erythrocytes in microvessels within the brain96. Cerebral malaria can result in localized cerebral microvascular thrombosis, especially as infected erythrocytes bind to EPCR, thereby reducing the generation of APC97,98. The loss of EPCR can also contribute to compromised vascular endothelial barrier function, vascular leakage, oedema and microhaemorrhages due to the loss of cytoprotective signalling through the APC–EPCR–PAR1 pathway99.
The excessive generation of thrombin in DIC can manifest as varying phenotypes, which are not necessarily restricted to excessive thrombosis (Figs 3,4). This is exemplified by an early hyperfibrinolyic (haemorrhagic) phase, which is a response to the surge of thrombin in the early stages following severe trauma100,101. The haemorrhagic phase is followed by a procoagulant (thrombotic) phase after 24–48 hours102, which is mainly attributable to the excessive expression of PAI-1 on the surface of platelets and activated endothelial cells, as well as to the suppression of the protein C anticoagulant pathway101.
Mechanisms that disseminate thrombin generation. In DIC, organ dysfunction typically occurs distant to the site of injury, for instance, acute lung injury in severe trauma and necrotizing pancreatitis. As such, factors that disseminate and increase the generation of thrombin are particularly important. These factors include abnormal and excessive expression of tissue factor on the surfaces of activated cells and derived microparticles103,104, platelet polyphosphate-dependent activation of factor XI68,105, increased consumption (and reduced production) of anticoagulant factors (protein C and antithrombin), increased exposure of negatively charged surfaces that dramatically enhance thrombin generation106,107 and, finally, the newly recognized role of DAMPs in promoting and fuelling thrombin generation68,108. DAMPs are discussed in more detail below.
The generation and dissemination of microvascular thrombi in DIC leads to organ ischaemia and ischaemia–reperfusion injury, which in turn results in nonspecific body responses with inflammation and further coagulation activation, fuelling a vicious circle that substantially contributes to multiple organ dysfunction109. Thus, the development of multiple organ dysfunction in the course of DIC is directly linked to the systemic dissemination of thrombin generation that overwhelms the fine balance between the opposing physiological effects controlled by thrombin and the phenotype (thrombotic or haemorrhagic) that follows thrombin generation during the DIC process. This phenotype relies on the predominant pathway that overwhelms the haemostatic balance and might be partly, but not completely, shaped by the underlying pathology.
Immunothrombosis and DAMPs in multiple organ dysfunction. Although clot formation was classically described as a function of activated clotting factors, subendothelial collagen and platelets, recent evidence suggests that the picture is much wider and involves activated neutrophils and monocytes — what has come to be known as ‘immunothrombosis’ (Refs 22,23). As described above, monocyte-derived tissue factor expression is believed to be an important mediator of immunothrombosis103,104. Furthermore, the interactions between neutrophils, monocytes and platelets to generate and propagate thrombosis in vivo is increasingly compelling72,110.
Emerging evidence also highlights the pivotal part played by DAMPs, such as the individual components of NETs (histones and DNA) in mediating not only immunothrombosis74,108 but also direct cellular toxicity that contributes to organ injury and multiple organ dysfunctions111,
Diagnosis, screening and prevention
Clinically, the disordered coagulation associated with DIC can manifest at any point in the spectrum between bleeding and thrombosis (Fig. 3). Although bleeding — ranging from oozing at venipuncture sites to major organ haemorrhage — is one manifestation of DIC, organ dysfunction from microvascular thrombosis is usually evident. The archetypal expression of this organ dysfunction is the skin manifestation of purpura fulminans, which appears bruised owing to bleeding under the surface, but also ischaemic owing to the reduced blood supply (Fig. 4). Equally, the underlying and predisposing clinical condition is likely to influence the balance between thrombosis and bleeding and affect the resulting phenotype (Box 1).
As such, the diagnosis of DIC is always made in the context of the underlying clinical condition121 and this is typically in the setting of acute clinical deterioration. Thus, diagnostic tests should be simple, robust and rapid to keep pace with the DIC process. As increased in vivo thrombin generation is central to the pathogenesis of DIC, assays can detect its generation (such as the thrombin generation assay and the detection of increased levels of thrombin–antithrombin complexes), the activation of the protein C pathway (such as measuring increased levels of APC and detecting APC–inhibitor complexes) and its activity on fibrinogen (such as measuring the levels of fibrinopeptide A or soluble fibrin monomer). However, none of these sensitive indicators of thrombin generation meet the practical challenges present in the acute diagnostic laboratory of turning out such results rapidly. Moreover, serial testing of these molecular markers to monitor evolving DIC is unlikely to be cost and clinically effective.
As falling levels of the endogenous anticoagulants (protein C and antithrombin) have been linked to clinical outcome in DIC51,122, measuring the levels of anticoagulants might offer a more practical approach to establishing coagulation activation at the molecular level. Results can now be generated in real time, but the sensitivity of this approach for non-overt DIC is not quite clear. Moreover, overt DIC may already be evident through routinely available tests of global coagulation (prothrombin time and activated partial thromboplastin time) by the time that levels of protein C and antithrombin fall to <50% of normal2. Prognostically, studies have shown that the more-readily available measures of prothrombin time (a global measure of the extrinsic pathway of coagulation that can be increased in DIC) and D-dimer changes (a measure of lysis of crosslinked fibrin and an increase in DIC), rather than levels of protein C and antithrombin, relate significantly to mortality in DIC when analysed as continuous variables123.
As such, diagnosing DIC is still very much reliant on a composite of simple, rapid and practical tests of global coagulation, such as the prothrombin time, activated partial thromboplastin time (aPTT; which is a global measure of the intrinsic pathway of coagulation) and the number of platelets in circulation (which can fall as part of DIC consumption) alongside the levels of fibrinogen and markers of fibrin formation and its lysis, such as D-dimer. None of these markers are sufficiently sensitive or specific in isolation, and a combination of results at different time points is particularly helpful in determining the presence of DIC. However, normal prothrombin time or aPTT do not exclude coagulation activation124 and it is more useful to look for time-dependent longitudinal changes to capture the typically evolving nature of DIC. Nonetheless, it is well validated that the degree of abnormality in global coagulation tests has pathogenetic relevance in indicating the degree of multiple organ failure and the likelihood of death, which has led the ISTH SSC on DIC to develop and harmonize a composite scoring framework of global haemostatic tests, in which a score of ≥5 is indicative of overt DIC3,125 (Box 2).
Once overt DIC can be identified, the condition might already be at a stage of irreversible decompensation and, therefore, late from a therapeutic perspective. With this in mind, a standardized method of identifying non-overt DIC at a point when haemostatic dysfunction is subtle but starting to decompensate would be important clinically and for inclusion criteria in clinical trials. Screening for this has been proposed within the ISTH DIC guidelines through scoring for abnormal trends and abnormal results in global coagulation tests3. Although this approach needs prospective validation, similar approaches for capturing worsening coagulopathy have been shown to correlate with clinical deterioration, increasing number and severity of organ dysfunctions and adverse outcome in general. As such, a high index of suspicion of abnormal changes in the global coagulation tests is the mainstay for alerting the clinician to identify the trigger so that its removal can prevent the overt transformation of DIC.
Point-of-care tests that can assess the combined effect of the different haemostatic components, including platelets and the fibrinolytic system, would be advantageous if truly indicative of the predominant effect of multifaceted thrombin generation in vivo. In this regard, automated thromboelastographic techniques have the potential for real-time capture of the evolving DIC process126; no prospective studies have been conducted on their sensitivity or specificity in DIC. The only evidence for the cost and clinical effectiveness of these techniques relates to predicting blood loss in cardiovascular surgery127,128. In addition to coagulation-based tests, there is a potential for improving DIC diagnosis by incorporating biomarkers of the crosstalk between coagulation, inflammation and innate immune activation, such as DAMPs, that have a pathogenetic link to increasing thrombin generation in vivo111,115. However, none of these assays are sufficiently simple and rapid, and their robustness in DIC needs to be comprehensively evaluated.
As presented in Fig. 5, the theoretical cornerstone of DIC management is the specific and vigorous treatment of the underlying conditions (the insults)4. DIC should be simultaneously well managed to improve a patient's outcome. Discrimination between controlled and uncontrolled DIC is important. In controlled DIC, the endothelial regulatory network for coagulation control is temporarily overridden and DIC will reverse quickly when the predisposing condition is removed or stopped (such as in cases of transfusion reactions or placental abruption). By contrast, uncontrolled DIC is characterized by the overriding of the regulatory factors and the degradation of the endothelial network (which occurs, for instance, in sepsis and trauma)3. Thus, in cases of uncontrolled DIC, in addition to the management of underlying disorders, it is essential to install supportive treatment aimed at DIC itself. The timing of the start of treatment is also important. Early treatment, before DIC is diagnosed, might deteriorate physiological haemostasis and immunothrombosis against injuries and pathogens and exacerbate the underlying conditions. Indeed, Fourrier129 demonstrated that anticoagulant factor concentrates only improve the outcome of severe sepsis when administered to patients with DIC. The same results were obtained from the subgroup analyses of two large-scale trials in which anticoagulant factor concentrates were used for the treatment of severe sepsis16,17. These results suggest that the target of these concentrates is not severe sepsis, but severe sepsis with established DIC and that DIC should be treated after its definitive diagnosis (Fig. 6). Finally, during treatment, repeated evaluation of the DIC score is required to monitor the DIC status, which will improve the diagnostic accuracy and the ability of the DIC scoring system to predict clinical outcomes3,18,100.
Although the evidence-based benefits of the transfusion of platelets, fresh frozen plasma (FFP) and coagulant factor concentrates have not yet been established in randomized controlled trials, these therapies seem to be supported in patients who are at risk of bleeding or those with bleeding due to consumption coagulopathy. The recommended treatment thresholds for transfusion of platelets, FFP, fibrinogen concentrates or cryoprecipitates have been presented in the ISTH guidance for treatment of DIC4.
The ISTH guidance recommends the use of prothrombin complex concentrate (PCC) in actively bleeding patients to promote clotting; however, the following should be considered: PCC is a concentrated product composed of three or four vitamin K-dependent coagulant factors and contains no (or a very small amounts) of anticoagulant proteins, such as protein C, protein S and antithrombin130. This means that, at least theoretically, the extremely high ratio of procoagulants to anticoagulants in PCC might induce both thromboembolism and DIC130,
Anticoagulant treatment is an appropriate approach based on the ISTH definition of DIC, in which DIC is characterized by extensive activation of coagulation due to systemic thrombin generation. However, the use of anticoagulants in patients with bleeding due to consumption coagulopathy or increased fibrinolysis (or fibrinogenolysis) is controversial. Anticoagulants are contraindicated in DIC caused by severe trauma and traumatic shock accompanied by critical bleeding due to both consumption coagulopathy and hyperfibrinolysis (or hyperfibrinogenolysis)101.
The ISTH guidance recommends the use of unfractionated heparin or low-molecular-weight heparin (LMWH) in DIC with the thrombotic phenotype4. However, there have been no randomized controlled trials demonstrating a clinically relevant outcome for patients with DIC who are treated with heparin. Unfractionated heparin was found to target sepsis, but not sepsis with DIC, in a failed randomized trial135. No significant differences in DIC scores or mortality of patients with DIC who received unfractionated heparin or LMWH were noted in a small randomized trial136. Importantly, the prophylaxis of venous thromboembolism with unfractionated heparin or LMWH is advocated in critically ill, non-bleeding patients with DIC4,137.
Anticoagulant factor concentrates
The escape of thrombin from the site of insult into the circulation is inhibited by anticoagulant mechanisms, including TFPI, antithrombin and the thrombomodulin–protein C systems of the endothelium. When anticoagulant mechanisms are overwhelmed by severe insults, systemic thrombin generation ensues138. In DIC, endothelial injury as well as the consumption and dysfunction of these anticoagulant proteins enhance dissemination of thrombin generation; thus, the use of agents that are capable of restoring impaired anticoagulant pathways is recommended4,139. Three large trials of antithrombin140, TFPI141 and APC142 for severe sepsis have failed. However, it is important to note that, in these trials, although the treated patients had severe sepsis, the majority did not have both severe sepsis and DIC. In addition, the subgroup analyses of patients with or without DIC16,17 and the analyses across the subgroup that was diagnosed with DIC at entry129 showed significant efficacy of antithrombin and APC in decreasing mortality in patients with severe sepsis associated with DIC. Accordingly, studies that focus on the treatment of anticoagulant factor concentrates for severe sepsis with DIC, but not for severe sepsis without significant coagulopathy, are required.
Recombinant APC has been removed from the market based on the results of the PROWESS-SHOCK study142. Plasma-derived APC is still available and a double-blind, randomized trial of the use of APC and unfractionated heparin in the treatment of DIC has been conducted143. APC significantly improved 28-day mortality (20.4% for APC versus 40% for placebo; P < 0.05) without increasing bleeding. However, this drug is locally approved and cannot be used worldwide.
A Cochrane review144 concluded that antithrombin did not significantly reduce overall mortality compared with a control group of patients who were critically ill with various underlying conditions. Several subgroup analyses did not investigate the effects of antithrombin in patients with DIC; no data on the effects of antithrombin in DIC were available for this review. By contrast, a meta-analysis on mortality in patients with DIC and sepsis, and a systematic review and meta-analysis on mortality due to DIC in severe sepsis showed significant reduction in mortality with antithrombin treatment145,146. The systematic review and meta-analysis146 included a subgroup analysis of the KyberSept trial, which showed the efficacy of antithrombin in the treatment of DIC17. In addition, a randomized controlled study of the effects of antithrombin on DIC in sepsis in patients who had initial antithrombin levels of 50–80% of normal showed that antithrombin significantly improved DIC scores and doubled the rate of recovery from DIC without any risk of bleeding147. The small sample sizes and the low mortality in the control (13.3%) and antithrombin (10%) groups explain why there was no improvement in the 28-day mortality. Furthermore, a multicentre survey showed that patients with sepsis and DIC who had an initial antithrombin level of <40% of normal and had received a high dose of antithrombin (3,000 IU per day) had a higher DIC recovery rate and 28-day survival rate than patients who received a lower dose of antithrombin (1,500 IU per day), without an increase in the bleeding risk148. Moreover, two large nationwide database studies have been published149,150, the first with 2,194 patients and the second with 518 patients, which used propensity scores and instrumental variable analyses. In these studies, the administration of antithrombin was associated with significant reductions in 28-day mortality in patients with severe pneumonia and sepsis-induced DIC and in patients with sepsis-induced DIC after emergency laparotomy for intestinal perforation. Together, these results suggest the need for a randomized controlled trial of antithrombin without adjuvant heparin and good bias protection in patients who meet prespecified criteria for DIC144.
Although somewhat varied, the data on the efficacy of using recombinant human soluble thrombomodulin (rhTM) to treat DIC are generally positive. Following the study that demonstrated the efficacy and safety of rhTM in DIC151, many studies confirmed these results; however, the results of these subsequent studies have been heterogeneous. For example, one study showed little or no association between the use of rhTM and mortality in patients with severe pneumonia and sepsis-associated DIC152. By contrast, a historical control study and a multicentre propensity score-matched study on sepsis-induced DIC showed that rhTM led to significant improvements in organ dysfunction, DIC and hospital mortality153,154. In addition, rhTM has been shown to improve mechanical ventilation times and the length of stay in the intensive care unit. A systematic review and meta-analysis found beneficial effects of rhTM on 28–30-day mortality and the resolution of DIC in patients with sepsis-induced DIC155. Importantly, no studies have found an increased incidence of major bleeding or transfusion associated with rhTM use151,
In DIC, fibrinolysis is primarily blocked by increases in the levels of PAI-1. Thus, DIC should not be treated with antifibrinolytic agents that may in fact cause deterioration of microvascular thrombosis4. In some cases, DIC and pathological systemic hyperfibrinolysis (or hyperfibrinogenolysis) may coexist — known as DIC with the hyperfibrinolytic phenotype158,159. Examples of this clinical condition include acute promyelocytic leukaemia (APL) and prostatic carcinoma. In these cases, antifibrinolytic treatment may be applicable4.
Acute promyelocytic leukaemia. Massive plasmin is formed on the assembly of plasminogen and t-PA on cell surface-associated annexin II. Its formation leads to the consumption of α2-plasmin inhibitor by the plasmin–α2-plasmin inhibitor complex in the plasma, which results in excess plasmin. In turn, excess plasmin induces haemorrhagic disorder in patients with APL160. A small double-blind study showed that tranexamic acid (an antifibrinolytic agent that inhibits plasmin-mediated degradation of fibrin) was effective for the control of haemorrhage without thromboembolic complications in APL161. However, after the introduction of all-trans retinoic acid (ATLA) as a front-line therapy for APL, a recent historical control study showed no potential benefit of the prophylactic use of tranexamic acid in patients with APL who were treated with ATLA162. Because of the complication of fatal thrombosis when ATLA and tranexamic acid treatments are combined163, antifibrinolytic agents should only be considered in cases of life-threatening bleeding4,164.
Trauma. Weibel–Palade bodies are storage granules of endothelial cells. Upon traumatic shock-induced hypoperfusion, these bodies release t-PA into the circulation, which results in systemic fibrinolysis (or fibrinogenolysis) in addition to DIC-induced secondary fibrinolysis33,158,165. In contrast to the immediate t-PA release from the endothelium, the induction and expression of PAI1 (also known as SERPINE1) mRNA usually takes several hours166. Indeed, immediately after trauma, one study showed that patients with DIC had significantly higher levels of t-PA and plasmin than those without DIC, whereas the levels of PAI-1 were almost identical between patients with and those without DIC33. These studies suggest that an extreme imbalance between the levels of t-PA and PAI-1 are the main cause of hyperfibrinolysis in patients with DIC during the first few hours after trauma. Tranexamic acid can reduce the risk of death in patients with bleeding trauma167 and should be given as early as possible because any delay reduces its efficacy and might be harmful168. Although some debates exist169, these studies provide the theoretical basis for antifibrinolytic therapy in DIC with the hyperfibrinolytic phenotype in the early phase of trauma.
In summary, the management of multiple organ dysfunctions in general and in DIC specifically has had many failed therapeutic dawns140,
Quality of life
The quality of life of patients who develop DIC is highly dependent on the conditions and prognoses of the disorders underlying DIC. Similarly, the occurrence of DIC markedly influences the prognoses of the various disorders that underlie DIC and the development of multiple organ dysfunctions8.
A logistic regression analysis demonstrated that, on the day of diagnosis of severe sepsis, the complication of DIC was an independent predictor of 28-day and hospital mortality and that the DIC score could also predict a patient's prognosis15,18,100. Importantly, the Kaplan–Meier method showed that the 1-year survival rate of patients with severe sepsis and DIC was significantly lower than that in severe sepsis without DIC18. Another study has shown that, on the day of the injury, the complication of DIC was found to be an independent predictor of death of patients with trauma19,171.
In addition, DIC could predict the need for massive transfusions, and DIC scores on admission to the emergency department were found to be significantly correlated with the transfusion volume of platelet concentrates, packed red blood cells and FFP20,171. These findings clearly indicate that the quality of life of patients with DIC can be aided by the improvement of both the underlying disease and DIC itself.
Although insight into the mechanisms underlying the development of DIC in various settings has improved considerably in recent decades, important questions remain. For instance, as it is likely that activation of coagulation in DIC occurs at the surface of the endothelium that interacts with inflammatory cells and mediators, we must gain more knowledge on the exact interactions between these components at this surface in vivo. As most of our current insights are based on in vitro observations (for example, by using cultured cells and isolated molecules), which may sometimes lead to spurious results, and studies in experimental animals and humans are mostly based on ex vivo observations, the real challenge may be to be able to more directly and incisively analyse inflammatory-driven activation of coagulation at the vascular wall surface in vivo. This approach could potentially yield new targets for improved treatment strategies for DIC.
Current therapeutic interventions are mostly supportive and only partly effective. Although these interventions can lead to improvement of the coagulopathy or more-rapid resolution of DIC, they do not affect clinically relevant outcomes, such as organ dysfunction or mortality. Further refinement of (supportive) treatment might come from the notion that the effect of the coagulopathy in DIC might vary from organ to organ172,173. Hypothetically, it could be argued that therapy should be tailored to the organs that are most affected. For example, if acute lung injury is the most prominent feature of DIC, therapy should be aimed at restoration of physiological anticoagulant pathways, such as antithrombin or thrombomodulin. In DIC presenting with purpura fulminans, there are ample indications that restoration of the APC pathway might be most effective. By contrast, in acute renal failure, interventions aimed at the deranged platelet–vessel wall interaction (for example, restoring the levels of ADAMTS13) can be most helpful.
Management of DIC might also benefit from improvement in early patient identification and risk stratification. Although the diagnosis of DIC has been greatly improved and facilitated after the introduction of diagnostic scoring algorithms, these systems are particularly effective in establishing overt DIC and less sensitive and specific for DIC in its early stage. In addition, tests that can be used to assess endothelial cell perturbation in combination with early-stage systemic coagulopathy would be helpful to identify patients at high risk of developing uncontrolled DIC and would facilitate early (and thereby potentially more effective) treatment.
In addition, genetic variation between patients might be important in the vulnerability for the development of DIC and the severity of the coagulopathy174. For instance, genetic mutations and polymorphisms have been shown to affect coagulation and fibrinolysis in DIC. Mice with targeted disruption of one allele of the gene that encodes protein C, which causes heterozygous protein C deficiency, exhibited a more-severe DIC and associated inflammatory response82. In addition, factor V Leiden heterozygosity has been associated with the incidence and outcome of DIC in sepsis175. Furthermore, a functional mutation in PAI1 — the 4G/5G polymorphism — not only influences the plasma levels of PAI-1 but has also been linked to clinical outcome of meningococcal sepsis and DIC176. More insight into the genetic variation that influences the host response to conditions underlying DIC might be helpful for identifying patients who are more susceptible to DIC and for individually tailoring therapy to the most vulnerable patients.
C.-H.T. has received funding from the US National Institute of Health Research. The authors thank Y. Alhamdi (University of Liverpool, UK) for assistance in preparing the manuscript.
About this article
Intensive Care Medicine (2017)