Brain Regulation of Thrombosis and Hemostasis: From Theory to Practice

Introduction

Thrombosis and hemostasis impact stroke neurology and cerebrovascular disease. Decision-making for stroke prevention and stroke treatment almost invariably involves pharmacotherapies that involve coagulation, with drugs ranging from aspirin to anticoagulants, both new and old. Nevertheless, the focus on pharmacotherapy modulation of coagulation processes tends to overlook some basic pathophysiological realities: the organ affected by stroke has its own unique system for regulation of thrombosis and hemostasis.

The purpose of this paper is to review the brain’s intrinsic capacity for thrombosis and hemostasis regulation. At first glance, it may appear that this is esoteric subject matter. Indeed, until recently it appeared that this material was largely of theoretical interest.

It is now evident that thrombosis, hemostasis, and the brain have enormous practical significance for the field of stroke neurology. Intrinsic regulation of thrombosis and hemostasis can no longer be ignored if one is to intervene pharmacologically in ways that impact coagulation. Manipulation of systemic coagulation factors will necessarily have consequences implicating intrinsic brain regulatory mechanisms.

This paper will review thrombosis and hemostasis from an organ-specific perspective, and one organ in particular: the brain. The paper will address how brain regulation of thrombosis and hemostasis manifests itself in the context of organ-specific regulation. The phenomenon of cerebral microbleeds will then be discussed as a pivotal and perhaps paradigmatic example of the importance of this issue.

Thrombosis, Hemostasis, and Coagulation

The terms thrombosis, hemostasis, and coagulation are sometimes used interchangeably, but more precisely refer to different processes. Hemostasis defines the avoidance or arrest of bleeding by maintaining blood within a vessel. Thrombosis is the formation of clot within a blood vessel resulting in obstruction of flow, while coagulation refers to a liquid transformed into a coherent solid or semisolid mass (1).

Coagulation is primarily regulated by circulating soluble factors, circulating cells, and vessel wall constituents, combined with vascular integrity and blood flow. Circulating cells are considered the components of “primary hemostasis”, while circulating coagulation factors constitute “secondary hemostasis” (1). The four component model of coagulation (hepatic factors, bone marrow-derived hematopoietic cells, vascular tree, and endothelium) represents a useful conceptual approach to this complex system, in which primary hemostasis consists of the relevant bone marrow-derived cells (platelets and monocytes) while the liver provides the coagulation factors of secondary hemostasis (1).

Another model incorporates the classic Virchow’s Triad, consisting of alterations of blood flow, blood wall, and blood constituents. This model is typically used to explain predilection for thrombosis, but is also highly relevant for maintenance of hemostasis (1–3). While flow reduction, overproduction of coagulation factors, and exposure of subendothelial constituents all contribute to thrombosis, low levels of coagulation factors and disruption of vascular integrity will obviously impact hemostasis.

The process of coagulation (described in some detail in Figure 1) is initiated by activation of factor VII by tissue factor, and this extrinsic pathway activation is amplified by the intrinsic pathway (4). Coagulation activation leads to fibrin clot generation, and this process is largely regulated by four pathways, three of which are endothelial-based: the thrombomodulin (TM) protein C pathway, the tissue factor pathway inhibitor (TFPI) pathway, and the fibrinolytic pathway. Another important pathway involves circulating antithrombin III and cofactor heparan sulfate proteoglycans, which inhibit all coagulation proteases of the coagulation cascade (4, 5). Studies of knockout mice demonstrate that all three anticoagulant pathways are necessary for coagulation cascade regulation; in contrast, animals lacking components of fibrinolytic pathway generally survive until adulthood, suggesting overlap of function for tissue-type plasminogen activator (tPA) and urokinase-type plasminogen activator (uPA) (4).

Organ-specific thrombosis and hemostasis

Based on Virchow’s Triad, it would be expected that substantial alteration of circulating coagulation factors would result in diffuse thrombotic or hemorrhagic phenomena. And yet this is not the case, because focal or mutli-focal rather than diffuse events occur within the vasculature with these changes in circulating factors. These local events, occurring in the presence of systemic changes of the coagulation system, are indicative of organ-specific regulation of thrombosis and hemostasis (5,6).

Focal changes in the presence of a systemic prothrombotic state are evident in diseases of veins, arteries, and microvessels. There appears to be a dual basis for this focality, residing in the differential expression of anticoagulant and procoagulant factors within different elements of the vascular tree, along with specific effects of different organs (6). For example, the endothelial protein C receptor (EPCR) is expressed predominantly in large arteries and veins, while TFPI is principally in capillaries (6,7). Other endothelial-dependent anticoagulant (eg, nitric oxide) and procoagulant (von Willebrand Factor) molecules show predilection for arteries and veins, respectively (6,8).

Arterial thromboses are particularly dependent on loss of vascular integrity with consequent exposure of subendothelial surfaces to blood (6). Nevertheless, thrombotic occlusion of the coronary arteries does not substantially increase with deficiencies of the protein C, protein S, or antithrombin III pathway (5). On the other hand, deficiencies of these same factors clearly predispose to venous thrombosis (5,6). These venous thromboses tend to occur in the lower extremities, at sites of venous valve pockets where stagnation of flow and local hypoxia is common (6, 9). A vastly different distribution of thrombosis occurs in the presence of polycythemia vera, paroxysmal nocturnal hemoglobinuria, and essential thrombocythemia, in which intra-abdominal veins (where valves are sparse, if present at all) are common sites of thrombi (5,10).

Further demonstration of the complexity of thrombus distribution is observed in the smaller vessels. The syndrome of erythromylagia, due to thrombotic occlusion of arterioles, tends to localize to the toes and fingers, and is provoked by presence of myeloproliferative syndromes polycythemia vera and essential thrombocythemia (6,11). In contrast, warfarin-induced skin necrosis tends to occur in buttocks, thighs, and breasts, and is due to thrombotic occlusion of dermal venules in the presence of low circulating protein C (5,6,12).

The concept of organ-specific regulation has been more frequently applied to thrombosis, rather than hemostasis. Nevertheless, there are notable differences in the distribution of hemorrhage, with variations observed in both murine models and in humans (4). The principal differences appear to depend on whether there are deficiencies of the extrinsic or intrinsic pathways. For example, low expression of tissue factor and low levels of factor VII both lead to murine hemorrhage in heart, lung, testis, uterus, and placenta (4). Deficiencies of intrinsic pathway, specifically factors VIII or IX, lead to hemorrhage involving muscles and joints, where tissue factor expression is known to be low (4).

Additional evidence for organ-specific regulation of thrombosis derives from a variety of knock-out mice, involving tPA, thrombomodulin, and TFPI (13–18). Initial reports of mice with combined deficiency of both tPA and uPA, with or without endotoxin injection to provoke thrombosis, showed extensive fibrin deposition in liver, lung, intestine, and uterus; thrombosis of brain was not reported (13). A later study investigating mice deficient for either tPA or uPA demonstrated 10–20 fold increased fibrin deposition in vasculature of lung, spleen, heart, and liver. However, the brain and kidney were spared thrombosis (14).

Further studies examined effects of murine thrombomodulin deficiency. Mice with inactivation of both alleles of the thrombomodulin gene demonstrated extensive fibrin deposition in lung; brain was not specifically examined in this study (15). Studies of mice with heterozygous thrombomodulin deficiency and with modified thrombomodulin containing a single amino acid substitution (producing vastly reduced ability to activate protein C) showed extensive fibrin deposition in lung, heart, spleen, liver. Once again, brain and kidney did not demonstrate this fibrin deposition (14). The effect of the murine thrombomodulin gene mutation (single amino acid substitution) was further studied using endotoxin, which provoked thrombosis in kidney, heart, spleen, and lung; however, the brain was again spared (16).

These findings of organ-specific thrombosis (and organ-sparing of thrombosis) were further explored by examining the effects of combining deficiencies of thrombomodulin with tPA/uPA and TFPI deficiency. Studies of single- and double- knockouts for murine thrombomodulin, tPA, and/or uPA showed that tPA had greatest impact on fibrin deposition in the heart (17). Tissue deposition of fibrin was also studied in mice heterozygous for TFPI deficiency combined with thrombomodulin mutation (single amino acid substitution), with fibrin demonstrated in liver but not lung or heart. The impact of these combined defects included some evidence of fibrin deposition in pial vasculature (18).

These murine investigations are consistent with clinical studies indicating the focal and organ-specific nature of thrombosis and hemostasis. The murine investigations also emphasize a unique role for the brain in this organ-specific regulation. The concept of brain-specific regulation is supported by the paucity of brain thrombosis in these various knockout models. Taken together, these findings suggest that the brain, compared to other organs, has less reliance on antithrombotic and fibrinolytic pathways and imply that protection against hemorrhage is a higher priority for the brain.

Characteristics of Brain-specific Regulation

The central nervous system fits well within the context of organ-specific regulation of thrombosis and hemostasis. The brain microvasculature demonstrates a remarkably consistent pattern of structural and functional organization that offers an unusual degree of protection against hemorrhage. These findings are localized principally at the microvascular endothelial junctions, combined with constituent underexpression of a variety of antithrombotic molecules.

Cerebral microbleeds and brain-specific hemostasis

There has been considerable attention to the phenomenon of cerebral microbleeds during the past decade. These focal areas of hemosiderin-iron were initially studied by MRI using gradient echo sequence, and later investigations showed that susceptibility weighted imaging was even more effective in demonstrating microbleeds (59). The consensus view is that cerebral microbleeds represent small foci of hemorrhage that are largely age-dependent (60). In addition to age, principal risk factors for microbleeds are cerebral amyloid angiopathy (for cortical microbleeds) and hypertensive vasculopathy (for subcortical lesions) (60, 61). Microbleeds limited to lobar location are more prevalent in ApoE €4 carriers and with the ApoE €2/€2 genotype (60). There is a strong correlation between cerebral microbleeds and cerebral white matter disease of aging (leukoaraiosis) (62–68), and microbleeds are also associated with ischemic and hemorrhagic stroke as well as Alzheimer’s Disease (69–71).

Most studies of cerebral microbleeds have focused on their imaging and their correlates with aging and a variety of disease entities as noted above. Increasingly, attention has been directed to what the presence of microbleeds tell us about the brain itself. The latter is an issue that involves neurological consequences of microbleeds, the impact of pharmacotherapies on microbleeds, and perhaps most importantly the underlying pathophysiological mechanisms for cerebral microbleeds.

Clinical and radiologic correlates of cerebral microbleeds provide clues to their underlying pathophysiology. 1) Cerebral microbleeds show strong age-dependence. In a population-based cohort study, microbleeds were demonstrable in only 6.5% of subjects age 45–50 years, but increasing to 35.7% in individuals age 80 year and older (60). 2) Cerebral amyloid angiopathy is a well-defined risk factor for cerebral microbleeds, particularly cortical or “peripheral” microbleeds (60, 61). 3) Hypertension is another well-defined risk factor for microbleeds, particularly subcortical or “deep” microbleeds (60,61). 4) A strong and consistent association has been observed between cerebral microbleeds and white matter disease of aging (62–68).

What do these correlates tell us about the process(es) which produces cerebral microbleeds? The current consensus view is that cerebral microbleeds are produced by focal tears in small arteries or arterioles leading to local bleeding (72, 73), and this seems highly likely at least for some cases given the strong association between cerebral microbleeds and clinical intracerebral hemorrhage (74); indeed, there is evidence suggesting heterogeneity among cerebral microbleeds (75). Nevertheless, the radiographic appearance of cerebral microbleeds can sometimes suggest a different etiology. Figure 4 shows examples of cerebral microbleeds from two different patients in which a diffuse, disseminated process, rather than a focal or multi-focal process, would appear to be more likely. These cases, while perhaps clinically extreme, are shown to emphasize a point: a disseminated process appears capable of producing the MRI appearance of cerebral microbleeds.

The likely source of such a diffuse process underlying cerebral microbleeds would involve the microvasculature. There are multiple lines of evidence supporting this contention: 1) White matter disease of aging, strongly correlated with cerebral microbleeds, appears to have a microvascular origin, probably involving the blood-brain barrier (76), and is likely to involve inflammatory and oxidative injury (77). 2) The blood-brain barrier is well-known to exhibit age-dependent changes, with increased permeability demonstrable with aging (78,79). 3) Both hypertension and cerebral amyloid angiopathy have been shown to independently contribute to blood-brain barrier dysfunction, by oxidative injury, inflammation, and tight junction alterations (80,81); 4) MRI of high-altitude cerebral edema patients, a clinical syndrome known to be microvascular in origin and almost certainly involving blood-brain barrier disruption, demonstrates cerebral microbleeds (82).

Given these elements of circumstantial evidence relating cerebral microbleeds to microvascular dysfunction, it is noteworthy that careful neuropathological evidence has demonstrated age-dependent capillary hemorrhage in human brain (27, 83–86). This was first described as high prevalence of cortical microscopic hemorrhage in aging brain (83), and high prevalence of capillary-derived cerebral microscopic hemorrhage has subsequently been confirmed by multiple studies (27, 84–86). Indeed, presence of microscopic hemorrhage in putamen is almost invariably present in human brain from subjects over the age of 70, and existing independent of either hypertension or local deposition of amyloid (27).

The fact there is age-dependent accumulation of small foci of hemorrhage within the brain is striking. As discussed above, the brain is characterized by a unique system of brain –specific hemostasis regulation, focused on the microvasculature, largely encompassed by the blood-brain barrier, and, by all appearances, designed to protect the brain against occurrences such as microbleeds. It is logical to consider the likelihood that development of cerebral microbleeds represents progressive failure of this system of brain specific hemostasis, a scenario that is outlined in Figure 5. In this scenario, age-dependent changes of brain-specific hemostasis are amplified by effects of hypertension and amyloid angiopathy, with superimposed transient injury contributing to formation of microhemorrhages/microbleeds. The transient injury may be inflammatory, known to produce enhanced transcellular permeability (87), and amplification of cerebral microhemorrhages has been demonstrated in this setting (88).

Therapeutic challenges: Mixed cerebrovascular disease

Stroke prevention efforts typically are either/or affairs, focusing on either ischemic stroke or hemorrhagic stroke. It has become increasingly clear that that approach is overly simplistic, due to frequent coexistence of ischemic and hemorrhagic cerebrovascular disease. It thus becomes critical to clarify: For what kind of cerebrovascular disease is preventative treatment being offered?

This difficult clinical context has led to the development of a new diagnostic formulation for stroke: mixed cerebrovascular disease (89, 90). This diagnostic categorization incorporates both ischemic and hemorrhagic stroke, clinical and subclinical. Ischemic syndromes include ischemic stroke (clinical) and silent infarction (subclinical), the later occurring as much as four times more frequently than clinical infarction (91). White matter disease of aging is included on the ischemic side, with the acknowledged difficulties that it may be sometimes difficult to distinguish normal white matter changes of aging (present in more than 95% of the population over the age of 65) and pathologic white matter disease (leukoariaosis) (92). Hemorrhagic stroke syndromes include intracerebral hemorrhage (clinical) and cerebral microbleeds (subclinical).

The advantages of a diagnostic categorization of “mixed cerebrovascular disease” is its emphasis on coexistence of ischemic and hemorrhagic processes, and the resultant implication that a more specific prevention strategy is indicated. The necessity for this specific prevention approach is based on the substantial body of data indicating that risk of hemorrhagic stroke is predicted by presence of cerebral microbleeds (74,75, 93). Adding to the complexity of this situation are the observations indicating that microbleeds themselves likely contribute to neurological dysfunction, thereby emphasizing the importance of treatment strategies that limit progression of cerebral microbleeds.

Platelet medications used for stroke prevention are known to increase risk for intracerebral hemorrhage, with aspirin increasing risk for hemorrhagic stroke by 84% (94) and combined treatment with aspirin-clopidogrel increasing hemorrhagic stroke risk beyond what is encountered with clopidogrel alone (95). Multiple reports indicate an important linkage between cerebral microbleeds and risk for intracerebral hemorrhage, for patients with both hypertension and cerebral amyloid angiopathy (74, 75). Cerebral microbleeds therefore represent a plausible mechanistic link between use of platelet medications and risk of intracerebral hemorrhage; this has in fact been demonstrated, as Wong et al and others reported that presence of cerebral microbleeds increased risk for intracerebral hemorrhage in patients using aspirin (93, 96).

Intracerebral hemorrhage is the most feared complication of anticoagulant therapy, with warfarin therapy carrying a risk of intracerebral hemorrhage generally ranging from 0.3% to 1% annually (97). This risk is substantially lessened, by as much as 33–60%, with the use of the new generation anticoagulants rivaroxaban and dabigatran (98, 99). Cerebral microbleeds once again represent a plausible mechanistic link, and substantial increased risk of intracerebral hemorrhage (as high as more than 80-fold increase) has been reported in patients with cerebral microbleeds using anticoagulant therapy (100–102). Moreover, cerebral white matter disease of aging is linked to increased risk (nearly 13-fold increase) of intracerebral hemorrhage in patients using warfarin (103), further emphasizing the likely common pathophysiologic origin for both white matter disease of aging and cerebral microbleeds.

While cerebral microbleeds represent an apparent mechanistic link between antithrombotic therapy and risk of intracerebral hemorrhage, the microbleeds themselves are increasingly associated with neurological dysfunction independent of hemorrhage. For example, Qiu et al reported that presence of multiple cerebral microbleeds more than doubled the risk for vascular cognitive impairment (104). These findings, relating cerebral microbleeds and cognitive impairment, were largely confirmed by later work (105–107). A relationship between microbleeds and neurological dysfunction is not surprising, given the demonstrable effects of heme on microglial activation via toll-like receptor-4, with resultant enhanced cytokine expression producing inflammatory injury (108).

Aspirin remains one of the most commonly utilized platelet medications for stroke prevention, but its use has been implicated in development of cerebral microbleeds. Vernooij et al reported approximately 70% increased risk of cerebral microbleeds with use of platelet medications (109), and chronic use of aspirin (>five years) was found to be associated with greater than five-fold increased risk of cerebral microbleeds in a Chinese population treated for cerebrovascular disease (110). The apparent aspirin-cerebral microbleed linkage, independent of the risk of intracerebral hemorrhage, emphasizes the importance of developing a more refined stroke prevention strategy that is less likely to contribute to development of cerebral microbleeds.

For the patient with mixed cerebrovascular disease, with coexistent ischemic and hemorrhagic processes, what will be the optimal approach for preventative treatment? Given the observed associations linking platelet medications and anticoagulants with microbleeds and intracerebral hemorrhage, it is obvious that the standard “one size fits all” approach for stroke prevention is insufficient. For the present, perhaps the most attractive strategy involves combining platelet effects with vascular wall protection.

A simple approach to achieve platelet inhibition and vessel wall protection is via modulation of intracellular cyclic nucleotides, for both platelets and vascular endothelial cells. This is achievable via modification of signal transduction pathways using phosphodiesterase inhibitors to regulate intracellular levels of cAMP and cGMP. Phosphodiesterase (PDE) inhibitors, of which there are eleven families with more than 60 isoforms, modulate hydrolysis of these cyclic nucleotides (111). Elevation of platelet cAMP and cGMP interferes with all known platelet activation pathways (111), while cAMP pathways have a critical role in development of the blood-brain barrier (112). Both dipyridamole and cilostazol are PDE inhibitors that have already been studied in stroke prevention trials and shown to be beneficial (113–116).

Dipyridamole, with stroke prevention effects comparable to aspirin (113), acts via relatively nonspecific PDE inhibition impacting both PDE3 and PDE5 (111). Dipyridamole produces platelet effects via elevation of plasma adenosine (by reduced red cell uptake) and increasing effects of prostacyclin and nitric oxide, while vessel wall protection is achieved by anti-oxidative effects and reduction of interactions between platelets and monocytes (117). Reduction of infarct size in experimental stroke has been demonstrated with dipyridamole (118). Moreover, in a mouse model of cerebral microbleeds and at clinically relevant plasma levels, dipyridamole did not worsen cerebral microscopic hemorrhage in aged transgenic animals subjected to immunotherapy-induced hemorrhagic worsening (88).

Cilostazol is a specific PDE3 inhibitor, resulting in relatively selective inhibition of cAMP hydrolysis (111). Cilostazol inhibits multiple pathways of platelet activation and aggregation, while vessel wall protection has been demonstrated in vitro by enhancement of endothelial cell barrier properties and reduction of histamine-induced transient barrier disruption (111, 119). Cilostazol has also been shown to reduce hemorrhagic conversion in several murine models of experimental stroke (120, 121). In clinical stroke prevention trials, cilostazol has been demonstrated effective compared to both placebo and aspirin. Hemorrhagic events, including intracerebral hemorrhage, were reduced by more than one-half for cilostazol, compared to aspirin treatment (116).

The patient with mixed cerebrovascular disease presents a unique therapeutic challenge, in which both ischemic risk and hemorrhagic tendencies must be addressed simultaneously. Therapy concurrently directed to both platelets and the vessel wall appears to be an attractive way to address this dilemma. This therapeutic challenge thus appears to represent a consequence of changes in the specific system of hemostasis regulation that resides in the brain.

Conclusions

The unique hemostasis regulatory system present in the brain resides in the microvasculature, and appears primed to protect the brain against hemorrhagic injury. This regulatory system has important structural and functional components, and appears largely as a component of the blood-brain barrier or neurovascular unit. It is proposed that changes in brain-specific hemostasis regulation are a critical underlying factor for age-dependent hemorrhagic changes of the brain, which manifest pathologically as microscopic hemorrhage and radiographically as cerebral microbleeds.

Much work needs to be done to fully flesh out this conceptual framework. Development of new animal models of cerebral microbleeds will be an important step (122), as well as definition of molecular elements of age-dependent changes of the blood-brain barrier. Exquisitely careful correlations will be necessary, between MRI cerebral microbleeds and neuropathologically-demonstrable cerebral microscopic hemorrhage. The relationship between cerebral microinfarcts (123) and cerebral microscopic hemorrhage (86) represents an important area for investigation, particularly with potential contribution of hemorrhagic microinfarction to development of cerebral microbleeds (86).

The future evolution of stroke prevention efforts will need to more carefully address the underlying pathophysiology of the ongoing cerebrovascular processes of concern. Just as the revision of our concepts of transient ischemic attacks (124), newer definitions such as mixed cerebrovascular disease may help lead to more effective efforts to further reduce the prevalence of stroke, which will only come about with optimal efforts addressing stroke prevention.