Risk factors for venous and arterial thrombosis

Introduction

Thrombophilia is considered as a condition predisposing to the development of thrombosis. Arterial thrombosis usually occurs after the erosion or rupture of an atherosclerotic plaque and, through platelet-mediated thrombi, can cause ischaemic injuries especially in tissues with a terminal vascular bed. Indeed, cardiac ischaemia and stroke are the most severe clinical manifestations of atherothrombosis. Ischaemia can arise slowly from the progression of atherosclerotic disease (stable angina, claudication) or acutely in the case of vascular (atherosclerotic plaque rupture) or intracardiac (atrial fibrillation, mechanical valve prostheses) thromboembolisation.

Venous thromboembolism (VTE) is the most common vascular disease after acute myocardial infarction and stroke. It is represented by two main clinical events: deep venous thrombosis (DVT) and pulmonary embolism (PE), which often constitute an unique clinical picture in which PE follows DVT. Although VTE is a common disease, the underlying pathogenic mechanisms are only partially known, particularly in comparison to those of atherothrombosis. During the past decades, progress has been made in the identification and characterisation of the cellular and molecular mechanisms that interdependently influence Virchow’s triad. It is now accepted that the combination of stasis and hypercoagulability, much more than endothelial damage, is crucial for the occurrence of VTE; venous thrombi are mainly constituted by fibrin and red blood cells, and less by platelets. In contrast, platelets are essential for primary haemostasis, repair of damaged endothelium and play a pivotal role in the development of atherosclerosis. Inflammation, lipids and the immune system, through a complex interplay, are also important determinants of arterial and, albeit to a lesser extent, of venous thrombosis. Pathophysiological and epidemiological findings have enabled the definition of the main risk factors for atherothrombosis and VTE, listed in Tables I and II. This review summarises the recent epidemiological data on the main risk factors for venous and arterial thrombosis, and considers the mechanisms by which they mediate the disease.

Table I.

Classical risk factors for cardiovascular disease³⁵.

Risk factor	OR (99%CI)
Hyperlipidaemia	3.25 (2.81–3.76)
Smoking	2.87 (2.58–3.19)
Diabetes	2.37 (2.07–2.71)
Hypertension	1.91 (1.74–2.10)
Abdominal obesity	1.62 (1.45–1.80)

OR: odds ratio; CI: confidence intervals.

Table II.

Classical risk factors for venous thromboembolism.

Strong risk factors (odds ratio >10)

trauma or fractures

major orthopaedic surgery

oncological surgery

Moderate risk factors (odds ratio 2–9)

non-oncological surgery

oral contraceptives and hormone replacement therapy

pregnancy and puerperium

hypercoagulability

previous venous thromboembolism

Weak risk factors (odds ratio <2)

age

bed rest (> 3 days)

prolonged travel

metabolic syndrome

air pollution

Age

There is an exponential increase in the risk of both arterial and venous thrombotic events with age^1,2, and the increase in life expectancy in the second half of the 20^th century is a major cause of the current epidemic of both arterial and venous thrombosis^1,3. Possible mechanisms include cumulative effects of risk factors on the arterial wall, decreased regular exercise, increasing immobility resulting in venous stasis, and increasing systemic activation of blood coagulation^4,5. Plasma concentrations of some coagulation factors (factors V, VII, VIII, and IX, fibrinogen) increase progressively with age^6,7. The same is true for von Willebrand factor (vWF), a key protein in platelet-vessel wall interactions⁸. For instance, the Framingham study showed that plasma levels of fibrinogen increased from a mean value of 280 mg/dL in individuals aged 47–54 years to more than 300 mg/dL in those aged 65–79 years⁹, with an increase of 10 mg/dL for each decade of age. High plasma levels of fibrinogen may play a causative role in the high incidence of cardiovascular events observed in elderly people, perhaps by enhancing the bridging of platelets via their glycoprotein IIb-IIIa receptor, by serving as a direct substrate of the clot and/or by increasing blood viscosity¹⁰. Alternatively, high fibrinogen levels may simply be a marker of the chronic inflammatory state typical of aging, without directly contributing to the risk¹⁰. A similar trend was shown for another acute phase protein, coagulation factor VIII, which increases progressively with age, up to more than 200 U/dL in the seventh decade of life⁴. Coagulation factor VII, both as a zymogen and as the activated protease, also increases with age¹¹. The role of tissue factor (TF) and factor VII as key components of blood coagulation and thrombus formation is well established (Figure 1). TF, a protein localised in the membrane of vascular cells, monocytes and circulating microparticles, is considered a key initiator of blood coagulation. When it is exposed in its active form at the vessel wall (e.g. after endothelial activation or during chronic inflammation, both conditions typical of aging), TF activates factor VII. This complex produces small amounts of thrombin and promotes thrombus formation through the activation of coagulation reactions on the membrane surfaces of activated platelets and microparticles¹². During aging, an increasing number of individuals develop a laboratory picture of enhanced activity of coagulation enzymes, expressed by high levels of the activation peptides that are cleaved from prothrombin, factor IX, factor X and fibrinogen [prothrombin fragment 1+2 (F1+2), thrombin-antithrombin complex (TAT), factor IX activation peptide, factor X activation peptide, fibrinopeptide A] when these zymogens are converted into their corresponding active enzymes^13,14. An impairment of fibrinolytic activity also occurs with aging. There is an increase of plasminogen activator inhibitor type 1 (PAI-1), the major inhibitor of fibrinolysis¹⁵, and a corresponding age-dependent decrease in fibrinolytic activity¹⁶. An increase in platelet reactivity with aging has also been demonstrated, and activated platelets greatly accelerate thrombin generation. Platelets of 60-year-old or older individuals aggregate more in response to adenosine diphosphate (ADP) and collagen than platelets from younger individuals¹⁷. Furthermore, a positive correlation has been observed between age and markers of platelet activation such as plasma β-thromboglobulin (a protein stored in the α granules of platelets) and platelet membrane phospholipids¹⁸. Because the vascular endothelium plays an important role in the normal process of haemostasis, any structural or functional change in the vascular wall (involving the extracellular matrix, vascular smooth muscle or endothelium) that occurs during aging may contribute to the increased risk of thrombosis in the elderly, particularly atherothrombosis. Advanced age is characterised by stiffness and dilation of the arteries, due to degeneration of elastic fibres and an increase in collagen and calcium content, and by a decrease in prostacyclin and nitric oxide with a related reduction in endothelium-dependent dilation¹⁹. There is also increased binding of platelet-derived growth factor to arteries, caused by changes in the glycosaminoglycan content of the vessel wall, which enhances the progression of atherosclerosis and indirectly contributes to atherothrombosis²⁰. In conclusion, there are several alterations of the haemostatic system in the elderly. A causal association between these alterations and thrombosis is likely but has not been formally proven, because of the lack of prospective studies demonstrating the development of clinical manifestations of thrombosis in comparison with aged healthy individuals.

Figure 1.

Role of tissue factor (TF) and coagulation factor VII in the activation of the coagulation cascade leading to thrombin formation.

TAFI = thrombin activatable fibrinolysis inhibitor; “a” = “activated”.

Thrombophilia abnormalities

Normally, the coagulation process is under the control of several inhibitors that limit clot formation near the damaged vessel wall, thus avoiding thrombus propagation (Figure 2). This delicate balance can be interrupted whenever the procoagulant activity of one of the coagulation factors is increased or the activity of one of the naturally occurring inhibitors decreases, leading to thrombus formation. This occurs with inherited deficiencies of natural inhibitors, as well as with inherited gain-of-function mutations of some coagulation factors²¹ (Table III). Inherited antithrombin, protein C and protein S deficiencies are rare but strong risk factors for venous thrombosis; they have little or no effect on arterial thrombosis. Antithrombin directly inhibits several activated coagulation factors, particularly thrombin and activated factor X, and the inhibitory effect is amplified by its binding to glycosaminoglycans of the endothelial surface which carry heparin-like activity. Antithrombin deficiency results in significantly reduced inhibition of thrombin and activated factor X and an increased tendency to clot formation, particularly in the venous system where the coagulation pathway (as distinct from platelets) plays a major role in thrombus formation²¹. The protein C anticoagulant pathway, localised on the surface of the endothelium, is essential in the down-regulation of thrombin generation. Thrombin activates protein C; the presence of thrombomodulin, together with endothelial protein C receptor (EPCR), accelerates the catalytic efficiency of this activation. Activated protein C proteolytically inactivates factor Va and factor VIIIa, the two most important activated co-factors of the coagulation cascade, dramatically slowing the rate of thrombin and fibrin formation. The inhibitory effect of activated protein C is accelerated by its main co-factor, protein S²². The inherited deficiency of one of these inhibitors leads to a critical reduction of the natural anticoagulant system and enhances thrombin generation, increasing susceptibility to VTE²¹.

Figure 2.

Anticoagulant mechanisms of blood coagulation. Antithrombin (AT) inhibits mainly activated factors II (IIa) and X (Xa) through its binding to glycosaminoglycans (GAG); protein C (PC), with its co-factor protein S (PS), is activated by thrombomodulin (TM) and inhibits activated factors V (Va) and VIII (VIIIa) through its binding to endothelial protein C receptor (EPCR).

TFPI = tissue factor pathway inhibitor; “a” = “activated”.

Table III.

Inherited, acquired and mixed coagulation or metabolic risk factors for thrombosis.

Inherited	Acquired	Mixed
Antithrombin deficiency	Antiphospholipid syndrome	Hyperhomocysteinaemia
Protein C deficiency		Increased fibrinogen levels
Protein S deficiency		Increased factor VIII levels
Factor V Leiden		Increased factor IX levels
Prothrombin G20210A		Increased factor XI levels

The two most common genetic risk factors for VTE are the G1691A mutation in the factor V gene (factor V Leiden) and the G20210A mutation in the prothrombin gene. The factor V Leiden gain-of-function mutation consists of the substitution of an arginine by glutamine at position 506 of coagulation factor V (R506Q), which is the cleavage site for activated protein C in the factor V molecule²³. Mutant factor V is partially resistant to inactivation by activated protein C, leading to a hypercoagulable state. Factor V Leiden explains more than 90% of cases of activated protein C resistance²⁴. The G20210A mutation in the prothrombin gene is a G to A transition at nucleotide position 20210 in the 3′-untranslated region of the coagulation factor II (prothrombin) gene, which increases plasma prothrombin levels²⁵. These two mutations also increase the risk of atherothrombosis, but to a lesser extent²⁶. The prevalences of inherited thrombophilia in the general population and in patients with VTE are shown in Table IV.

Table IV.

Prevalence (%) of inherited risk factors for VTE in the general population and in patients.

Abnormality	General population	Patients with VTE	Patients with recurrent VTE or age < 45 years
Antithrombin deficiency	0.02 - 0.17	1.1	0.5 - 4.9
Protein C deficiency	0.14 - 0.5	3.2	1.4 - 8.6
Protein S deficiency	?	2.2	1.4 - 7.5
Heterozygous factor V Leiden	3.6 - 6.0	21.0	10 - 64
Heterozygous prothrombin G20210A	1.7 - 3.0	6.2	18

The antiphospholipid antibody syndrome is one of the most important acquired risk factors for thrombosis. Characterised by the presence of circulating antiphospholipid antibodies in plasma, it is associated with arterial or venous thrombosis and/or pregnancy complications, including foetal loss. The clinically relevant antiphospolipid antibodies include lupus anticoagulant, anticardiolipin and anti-β₂-glycoprotein I antibodies. The term “antiphospolipid antibodies” is widely used even if it is not correct, because antibodies are not directed against phospholipids per se, but against a wide variety of protein co-factors acting on phospholipid membrane surfaces (β₂-glycoprotein I, prothrombin, protein C, protein S, annexin V, coagulation factor XII and others). The resulting complexes interact with several cell types, including endothelial cells, monocytes and platelets, all of which play important roles in haemostasis and thrombogenesis. The indirect activation of these cells results in the release of prothrombotic and pro-inflammatory mediators (e.g. TF-bearing microparticles, interleukin-6, proteins of the complement system), leading to the activation of platelet and coagulation pathways²⁷. Recent observations show that antiphospholipid antibodies interact directly with vessel wall and cause alterations of plasma lipoprotein [i.e. high density lipoprotein (HDL)] function leading to increased atherothrombotic risk²⁸.

Hyperhomocysteinaemia is a mild risk factor for thrombosis due to an impairment of the metabolic pathway that transforms the amino acid methionine into cysteine, leading to an abnormal elevation of plasma concentrations of homocysteine, an intermediate product of this pathway. Genetic factors (e.g., gene mutations in methylenetetrahydrofolate reductase and cystathionine β-synthase) and acquired factors (e.g., deficiencies of folate, vitamin B12 or vitamin B6, advanced age, chronic renal failure, and the use of anti-folic drugs) interact to determine plasma homocysteine concentrations, so that hyperhomocysteinaemia is a “mixed” (i.e., genetic and/or acquired) risk factor for both arterial and venous thrombosis²⁹. The possible mechanisms by which hyperhomocysteinaemia contributes to thrombosis are multiple and still under study; they include a toxic effect on endothelial cells, smooth-muscle-cell proliferation and intimal thickening, impaired generation of nitric oxide and prostacyclin, increased platelet adhesion, activation of factor V, interference with protein C activation and thrombomodulin expression, induction of tissue factor activity and inhibition of tissue plasminogen activator (t-PA)³⁰.

An association between increased plasma levels of some coagulation factors (VIII, IX, XI, and fibrinogen) and an increased risk of VTE has been demonstrated ³¹. The plasma levels of these factors are influenced by age and inflammation, but are also under genetic control. The mechanisms by which increased coagulation factors plasma levels enhance the risk of thrombosis are unknown, but a shift in the balance of the coagulation process towards a procoagulant state is plausible. High levels of fibrinogen are associated with an increased risk of atherothrombosis, while the effect of factor VIII is dependent on vWF, which plays the most important role in the increased risk of thrombosis associated with the factor VIII/vWF complex.

Metabolic syndrome and smoking

One of the most widely used definitions of the metabolic syndrome was proposed in 2001 by the National Cholesterol Education Program Adult Treatment Panel III (NCEP ATP III) and is based on the presence of at least three of the following diagnostic criteria: elevated waist circumference (abdominal obesity), elevated triglycerides, reduced HDL cholesterol, elevated blood pressure and elevated fasting glucose³². There is increasing evidence for an association between atherothrombosis and the metabolic syndrome^32–34. The INTERHEART study identified nine risk factors that collectively accounted for more than 90% of the risk of acute myocardial infarction. Risk predictors included life-style factors such as smoking, co-morbidities (hypertension, diabetes, abdominal obesity, abnormal lipid profiles), as well as psychosocial factors³⁵. Meta-analyses of randomised controlled trials on blood pressure³⁶ and cholesterol reduction³⁷, and observational studies on smoking cessation³⁸ confirmed that these three risk factors play causative roles in arterial disease, partly through atherogenesis and partly through a systemic activation of blood coagulation and inflammation³⁹. Available evidence from multiple randomised trials supports life-styles that promote weight reduction, cigarette smoking cessation and regular moderate exercise in order to reduce platelet reactivity and coagulability, and to promote fibrinolysis. The overall effects can be expected to translate into an improved cardiovascular prognosis or other beneficial clinical outcomes in healthy individuals and those with cardiovascular risk factors or established coronary heart disease⁴⁰. Nevertheless, a high residual risk for secondary ischemic events, despite life-style modifications, is evident from the results of recent clinical practice registries, which justifies the need for adding pharmacotherapy.

From a biological point of view, the metabolic syndrome is frequently accompanied by a prothrombotic state. This includes elevated plasma levels of PAI-1, thrombin-activatable fibrinolysis inhibitor (TAFI), vWF, coagulation factors VIII, VII, and XIII and fibrinogen, TF, increased release of endothelial cell microparticles and decreased protein C levels. Moreover, patients with the metabolic syndrome exhibit endothelial dysfunction (mainly decreased production of nitric oxide and prostacyclin) and heightened platelet reactivity³³. The activation of the haemostatic system related to the metabolic syndrome has been mainly attributed to the action of pro-inflammatory and pro-atherogenic mediators (e.g. leptin, tumour necrosis factor-α, interleukin-6) released by adipose cells³³, to a triggering effect of very low density lipoproteins (VLDL) and remnant lipoproteins on platelet activation and PAI-1 gene expression⁴¹, to the adverse effects of chronic hyperglycaemia on fibrin structure and function (generating a clot more resistant to fibrinolysis)⁴² and to increased circulating microparticles that sustain blood coagulation by exposure of anionic phospholipids and TF ⁴³.

This inflammatory and hypercoagulable state may explain the biological role of the major cardiovascular risk factors and could also be involved in VTE. Patients with idiopathic VTE have a higher prevalence of atherosclerosis than patients with VTE secondary to known risk factors and control subjects³⁴. In addition, the long-term incidence of cardiovascular disease is higher in patients with idiopathic VTE than in those with secondary VTE^44,45. These studies support the hypothesis of VTE as the first symptomatic cardiovascular event. Several epidemiological studies showed associations between obesity, metabolic syndrome or type 2 diabetes and VTE^46–50. A meta-analysis⁵⁰ of 63,552 patients showed that the relative risk of VTE was 2.33 (95% CI 1.68–3.24) for obesity, 1.42 (95% CI 1.12–1.77) for diabetes and 1.51 (95% CI 1.23–1.85) for hypertension (Table V). Obesity may confer an increased risk of VTE independently of the metabolic syndrome, because a high body weight can cause mechanical impairment of the valve system in the deep veins of the lower limbs, favouring venous stasis. In the same meta-analysis a non-significant increased risk for smoking was found⁵⁰, while in a large population-based case-control study [Multiple Environmental and Genetic Assessment (MEGA) study], the relative risk of VTE was 1.42 (95% CI 1.28–1.58) in current smokers and 1.23 (95% CI 1.10–1.37) in past smokers, compared to the risk in individuals who had never smoked⁵¹.

Table V.

Associations between classic cardiovascular risk factors and VTE.

Risk factor	OR (95% CI)
Obesity (BMI)	2.33 (1.68–3.24)50
Diabetes	1.42 (1.12–1.77)50
Hypertension	1.51 (1.23–1.85)50
Smoking	1.42 (1.28–1.58)51

BMI = body mass index; OR = odds ratio; CI = confidence intervals.

Finally, also dyslipidaemia may exert a mild influence on the risk of VTE^52,53, as determined by a recent meta-analysis in which patients with VTE had high triglyceride and low HDL cholesterol levels, while no effect of total cholesterolaemia on VTE was seen⁵⁰. Moreover, preliminary evidence shows that statins may be protective against VTE^54,55, supporting the hypothesis of dyslipidaemia influencing the risk of VTE.

In conclusion, despite the discrepancy between the estimated relative risks of VTE and atherothrombosis associated with cardiovascular risk factors, the latter may represent a link between two clinical entities which have classically been considered distinct.

Previous thrombosis

Despite the unambiguous benefit achieved with life-style modification, blood pressure control and the use of statins, angiotensin II-active and antiplatelet agents, the residual risk of recurrent acute events in patients with established atherothrombotic disease remains substantial^56,57. The residual risk is likely related to progression of atherosclerosis despite the use of statins and angiotensin II-active agents, to insufficient inhibition of platelet activation by aspirin and thienopyridines, and to other factors not yet identified⁵⁸. Studies that investigated the effect of novel approaches with or beyond statins on progression of atherosclerosis showed inconsistent results^59–64. Response variability to antiplatelet therapy may contribute to the residual risk of thrombotic events. Sufficient evidence supports the hypothesis that a persistent enhanced platelet reactivity despite the use of aspirin⁶⁵ and/or clopidogrel⁶⁶ is associated with adverse clinical outcomes. A number of factors may influence response to antiplatelet therapy, including drug dose and absorption, patient’s compliance and genetic polymorphisms⁶⁶. Novel approaches to limit platelet-mediated thrombosis, such as prasugrel, ticagrelor and cangrelor, appear promising. Nevertheless, current antiplatelet agents and those still in development, inhibit the thromboxane A₂ or ADP platelet activation pathways but do not interfere with a number of other platelet activation pathways that may contribute to thrombotic events. In addition, dual antiplatelet therapy is associated with increased bleeding. These considerations underline the urgent need for novel therapies that provide more complete platelet inhibition and, therefore, greater protection against thrombotic events, possibly without increasing the bleeding risk.

The presence of a residual thrombus after a first episode of DVT is an independent risk factor for recurrence⁶⁷. After a first episode of VTE, patients are 40 times more likely to develop a recurrent event compared to previously unaffected individuals⁶⁸. Previous VTE represents the most important risk factor for recurrence of DVT or PE (OR 15.5; 95% CI 6.77–35.99) and the risk is higher in individuals with previous idiopathic VTE than in those with secondary VTE⁶⁹. The risk of recurrence varies over time, being higher during the first 6–12 months after the index event⁷⁰. In a study involving 355 patients, the incidence of recurrent VTE was 8.6% at 6 months and 17.5% after 2 years⁷¹. After 8 years, the rate of recurrence was as high as 30.3%⁷¹. Moreover, recurrent DVT or PE is associated with an increased risk of post-thrombotic syndrome and chronic thromboembolic pulmonary hypertension⁷². Secondary prevention of VTE is, therefore, crucial to reduce the burden of these diseases significantly and to date the most effective strategy is represented by anticoagulant therapy.

A potential mechanism by which the residual thrombus increases the risk of recurrence is impaired venous outflow, resulting in blood stasis and clot formation. However, because some patients develop recurrent thrombosis in the initially unaffected leg and others develop isolated PE, other mechanisms must be implicated. Residual thrombosis is perhaps a marker for a more generalized procoagulant diathesis. Indeed, elevated plasma D-dimer levels after withdrawal of oral anticoagulation (a marker of hypercoagulability) are an independent risk factor for recurrent venous thrombosis^73,74.

Trauma, surgery and immobilisation

These transient conditions are associated with an increased risk of venous thrombosis. The incidence of DVT associated with major trauma is up to 58%, PE occurs in 2% of these individuals and is the third cause of death among patients who survive the first 24 hours after trauma^75,76. Considering minor trauma, the incidence of DVT is 28% when lower limbs are involved and the risk is higher in proximal than in distal fractures⁷⁷. In a large, population-based, case-control study the relative risk of VTE associated with previous minor injury was 3.1 (95% CI 2.5–3.8)⁷⁸. Minor injuries in a leg were strongly associated with VTE (OR 5.1; 95% CI 3.9–6.7), whereas minor injuries in other parts of the body were not⁷⁸. The presence of factor V Leiden in patients with a leg injury increased the risk up to 50-fold⁷⁸. The risk of thrombosis associated with surgery varies from 15% to 60% in the case of a laparotomic approach⁷⁹, while the risk associated with laparoscopic and arthroscopic surgery is less defined. Major orthopaedic surgery involving the lower extremity is a major risk factor for VTE. Rates of DVT without prophylaxis range from 40% to 60% in the 2 weeks after major orthopaedic surgery⁸⁰. Patients undergoing total hip arthroplasty are in the highest risk category for developing post-operative VTE. Evaluation of the natural history of this population of patients reveals that the incidence of DVT without appropriate prophylaxis is as high as 32% to 60%, and that of PE is as high as 16%, with a 0.3% to 3.4% occurrence of fatal PE⁸¹. Current data suggest an overall DVT rate of 9.9% and a proximal DVT rate of 2.1% after knee arthroscopy without thromboprophylaxis⁸². Even with appropriate thromboprophylaxis, total hip or knee replacement will lead to symptomatic VTE in 1% to 3% of patients⁸³. The mechanism by which these conditions lead to VTE is a combination of stasis and local accumulation of TF (i.e., hypercoagulability). Blood flow is relatively static in the pockets of venous valves, particularly those of the lower limbs. This effect is accentuated by immobilization. Stasis locally concentrates haemostasis activation factors (cytokines and other mediators of inflammation), favours cellular margination and interaction of circulating blood cells with endothelium, and is responsible for local hypoxia which is one of the principal mechanisms of endothelial activation⁸⁴. Studies in animals have shown that stasis alone does not provoke thrombosis⁸⁵. TF is expressed by cells in the subendothelial compartment. Thus, physical disruption of the endothelium, as occurs in trauma or surgery, may lead to exposure of blood to extravascular TF. However, the vast majority of venous thrombi occur in the context of an intact endothelium. In these cases, TF may be expressed on the surface of activated endothelial cells and/or mononuclear cells which have been stimulated by any number of inflammatory mediators including cytokines, chemokines (interleukins 1, 6 and 8, tumour necrosis factor-α, monocyte chemoattractant protein-1), vascular endothelial growth factor, factors derived by complement activation (C5a and complex of membrane attachment), immunocomplexes and antibodies, P-selectin, haemodynamic stress, hypoxia, and cell-cell interactions^84,86. In addition to expressing TF on their cell surface, activated cells (e.g., endothelial cells, monocytes, leucocytes and platelets) may release TF- and phospholipid-rich microparticles that circulate in the bloodstream⁸⁷. Microparticles can interact with other cells through the action of adhesive proteins. For example, P-selectin glycoprotein ligand 1 facilitates the transfer of P-selectin from platelets or endothelial cells to microparticles of monocyte origin⁸⁸. These properties may facilitate thrombus propagation and activate coagulation in various sites. Finally, leucocytes and platelets can further enhance thrombosis through their expression of TF under inflammatory stimuli (C5a, bacterial formylated peptides, P–selectin) and platelet agonists (ADP, collagen, thrombin), respectively⁸⁹.

Although VTE is the most frequent thrombotic complication of surgery, surgical iatrogenic injuries can also lead to arterial occlusion. Moreover, arterial thrombosis secondary to surgery can represent the first manifestation of heparin-induced thrombocytopenia, an autoimmune disease triggered by exposure to the heparin that is commonly given as antithrombotic prophylaxis of post-operative VTE. The clinical picture is characterised by transient thrombocytopenia (in more than 90% of cases the platelet count is >15,000/μL) and both arterial and venous thromboses, especially of the lower limbs, are described in surgical patients⁹⁰.

Cancer

Cancer is one of the most important acquired risk factors for VTE⁹¹. Some authors estimate an annual incidence of VTE of 1 in 200 patients with cancer⁹², and 20% of VTE cases occur in patients with cancer⁹³. Conversely, of all patients with cancer, 15% will develop symptomatic VTE⁹³, 50% asymptomatic VTE⁹⁴, and 50% will have VTE diagnosed at autopsy⁹¹. The risk of VTE is higher at diagnosis (OR 53.5; 95% CI 8.6–334.3) and in patients with distant metastases (OR 19.8; 95% CI 2.6–149.1)⁹⁵. If a patient with cancer survives an initial VTE event, he or she has an increased risk of recurrence (OR 1.72; 95% CI 1.31–2.25) compared with that in a patient without cancer. The cancer patient with VTE also has a significantly increased risk of death (OR 8.1; 95% CI 3.6–18.1), which persists for as long as the malignancy persists⁷¹. In addition, VTE is the second leading cause of death in hospitalised patients with cancer, after infections⁹⁶.

The pathophysiology of VTE in patients with cancer is even more complex than that in patients without. Three sets of factors have been linked to the increased risk of VTE in these patients: those related to the tumour, those related to the host, and those to the therapies the patient is receiving⁹⁷. Tumour mass may create stasis by compression and invasion of vessels. Tumour cells may promote the release of TF from the affected organs during expansion and the metastatic processes. Importantly, cancer cells themselves may release TF-rich microparticles. These microparticles can then adhere to (and be incorporated into) monocytes and other cells, in particular those activated by hypoxia, and promote fibrin formation^98,99. Finally, tumour cell-derived inflammatory and pro-angiogenic cytokines (e.g., tumour necrosis factor-α, interleukin 1 and mostly vascular endothelial growth factor) may induce TF expression as a host response in endothelial cells and monocyte-macrophages¹⁰⁰. Data from several epidemiological investigations demonstrated significant heterogeneity in the risk of VTE according to the different cancer histology. Pancreatic cancer, lymphoma, and brain cancer have a relative risk for VTE greater than 25, whereas the VTE risk associated with cancer of the ovary, stomach, kidney, colon, rectum, and lung is lower (> 17), compared with that of individuals without cancer¹⁰¹. In the MEGA study, patients with haematological malignancies had the highest risk of VTE (OR 28.0; 95% CI 4.0–199.7), followed by that in patients with lung cancer and gastrointestinal cancer⁹⁵.

Considering the therapeutic factors that influence the cancer-related risk of VTE, it has been shown that patients with cancer face double the risk of developing VTE because of oncological surgery, compared with the risk in patients without cancer undergoing similar surgery⁹¹. Chemotherapy increases the risk of thrombosis 6.5-fold¹⁰². The proposed mechanisms for chemotherapy-related risk of VTE probably include both direct drug-induced damage of the endothelium and an increased expression of TF procoagulant activity by macrophages and monocytes, thus inducing a procoagulant response by host cells¹⁰³. Another prothrombotic mechanism of antitumour therapy is likely related to the direct hepatotoxicity of radio- and chemo-therapy, which can cause a reduction in the plasma levels of natural anticoagulant proteins (antithrombin, protein C and protein S)¹⁰⁰. Women who are treated with tamoxifen for breast cancer have a 2- to 5-fold increased risk of VTE, and the risk is even higher after menopause and when tamoxifen is associated with chemotherapy^104,105. Cancer patients with a central venous catheter or transvenous pacemaker have a 6-fold increase in upper-extremity DVT¹⁰⁶. Finally, the presence of factor V Leiden in cancer patients increases the risk of VTE nearly 12-fold compared to those individuals without cancer and who have the wild-type factor V Leiden; similar results have been observed in cancer patients carrying the prothrombin G20210A variant⁹⁵.

Oral contraceptives and hormone therapy

Various clinical studies have investigated the risk of thrombosis with hormone-based oral contraceptives. However, due to the variety of preparations and the heterogeneity of study populations, these studies showed different or even contradictory results. Evidence on oral contraceptive-related risk of VTE has mostly been derived from case-control and nested case-control studies, attributing a relative risk of VTE to oral contraceptive use (compared with non-use) ranging from 1.12 (95% CI 0.4–2.9) to 22.1 (95% CI 5.9–84.2)¹⁰⁷. Overall, a 2- to 6-fold increased relative risk of VTE is observed in oral contraceptive users compared with non-users, for an absolute risk of 1 to 3 cases of VTE per 10,000 women-years¹⁰⁷. An increased risk of VTE was reported among women using third-generation oral contraceptives, i.e., those containing desogestrel or gestodene, compared to those using second-generation products containing levonorgestrel^108–111. A meta-analysis confirmed that third-generation oral contraceptives were associated with a significantly increased risk of VTE (RR 1.9; 95% CI 1.5–2.2) compared with that associated with second-generation oral contraceptives¹¹². Another meta-analysis confirmed an overall adjusted odds ratio for third-versus second-generation oral contraceptives of 1.7 (95% CI 1.4–2.0)¹¹³. Moreover, the odds ratios for short-term users compared with those of longer term users were 2.5 (95% CI 1.6–4.1) and 2.0 (95% CI 1.4–2.7), respectively¹¹³. However, some authors emphasised that the difference in VTE risk according to whether second- or third-generation oral contraceptives are used is minimal, and probably related to underlying congenital or acquired thrombophilic states¹¹⁴. Second- and third-generation oral contraceptives (as well as pregnancy/post-partum) increase the risk of VTE in carriers of factor V Leiden by 3.3- fold and 4.2-fold, respectively, whereas other risk factors have a minor effect¹¹⁵.

An unequivocal mechanism for explaining the thrombogenicity of oral contraceptives (especially oestrogen compounds) has not been identified, as several metabolic abnormalities might be triggered to induce a mild prothrombotic state. Mechanisms include a direct effect of oestrogens on the vascular wall, changes in factors that promote endothelial dysfunction, and changes in coagulation factors. Studies in animals suggest a loss of the normal elastic configuration of the aorta, significant intimal thickening and an increase in endothelial permeability after administration of oral contraceptives¹¹⁶. There are also a few reports of increased venous distensibility and reduced blood flow in women taking oral contraceptives¹¹⁷. A possible explanation might be an oestrogen-induced dose-dependent increase in the expression of matrix metalloproteinases that cleave collagen and elastin in the vascular intima. The loss of venous tone, with the accompanying tendency to venous stasis, increases the risk of venous thrombosis. Oral contraceptives may increase the risk of arterial thrombosis by promoting endothelial dysfunction. However, this is a poorly investigated area and it has not been established how much these changes matter in the pathophysiology of thrombosis in oral contraceptive users¹¹⁸. Another mechanism increasing the risk of thrombosis, particularly that of atherothrombosis in women taking oral contraceptives, is linked to changes in lipids and lipoprotein metabolism. Oral contraceptives increase total cholesterol, mainly by increasing LDL cholesterol. In addition, oestrogens decrease HDL cholesterol and increase triglyceride levels, affect lipoprotein metabolism by increasing the hepatic synthesis of apolipoproteins, and may induce changes in hormones affecting lipoprotein metabolism such as cortisol, thyroxine or growth hormone^119,120. Progestogen-only oral contraceptives have generally no or little effect on plasma lipoprotein levels.

Oral contraceptives modify the plasma levels of several coagulation factors (Table VI). However, these changes are often modest and concentrations of coagulation factors usually remain within the normal range. Oral contraceptive-mediated alterations in coagulation factor levels may result in synergistic or opposing effects on the risk of venous thrombosis. Levels of the anticoagulant proteins antithrombin and protein S decrease during oral contraceptive use, whereas protein C levels may increase^121,122. The greatest effects are seen with preparations containing the highest oestrogen doses. Important effects of oral contraceptives on blood coagulation are an acquired resistance to activated protein C and the reduction of protein S plasma levels¹²³. These and other changes in coagulation factors appear to be more pronounced in women using third-generation compounds than in those using second-generation compounds^124,125, although the difference is debated¹²⁶. Oral contraceptives also affect the fibrinolytic system by reducing t-PA levels and increasing levels of TAFI, PAI-1 and D-dimer. An overall increase in thrombin generation in women on oral contraceptives has recently been demonstrated by means of the endogenous thrombin potential test, i.e., the area under the thrombin generation curve, which is able to identify a global hypercoagulable state and has been found to be higher in oral contraceptive users than in non-users^{118,121,122,127}.

Table VI.

Haemostatic changes during oral contraceptive (OC) use and pregnancy.

	Change during OC use	Change during pregnancy
Procoagulant factors
fibrinogen, V, VII, VIII, IX, X, XII	↑	↑
XI	= or ↑	↓
von Willebrand factor	=	↑
Anticoagulant proteins
antithrombin	↓	=
protein C	= or ↑	= or ↑
protein S	↓	↓
resistance to activated protein C (ratio)	↓	↓
Markers of thrombin formation
F1+2, TAT complexes, fibrinopeptide A	↑	↑
D-dimer	↑	↑
Fibrinolytic factors
TAFI, PAI 1 and 2	↑	↑
t-PA	↓	↓

↑ increase, ↓ decrease, = no change, compared to non-use of oral contraceptives and to the non-pregnant state.

Pregnancy

VTE remains the major cause of maternal mortality world-wide (the rate of maternal deaths from VTE is 0.12 per 10,000 live births and stillbirths)¹²⁸. Results from studies in which either all or most pregnant women underwent accurate diagnostic testing for VTE report an incidence of VTE ranging from 0.6 to 1.3 events per 1,000 deliveries, confirming a 5- to 10-fold increased risk in pregnant women compared to that in non-pregnant women of comparable age¹²⁹. The MEGA study showed that the risk of VTE is nearly 5-fold increased during pregnancy and up to 60-fold during the first 3 months after delivery¹³⁰. A 14-fold increased risk of DVT of the legs and a 6-fold increased risk of PE were reported, with this risk being higher in the third trimester (OR 3.3; 95% CI 2.2–5.0) and during puerperium (OR 11.0; 95% CI 8.1–15.1), and highest in the 2 days before and the day after delivery (OR 77.6; 95% CI 52.4–114.8)¹³¹. The major risk factors for VTE during pregnancy include the presence of thrombophilic abnormalities, Caesarean section, advanced maternal age, obesity and pre-eclampsia, which are identified in nearly 70% of women with pregnancy- or puerperium-related VTE. It has been reported that the risk of pregnancy-related VTE might be 11- to 52-fold increased in factor V Leiden carriers, 3- to 31-fold in carriers of the prothrombin G20210A mutation, and more than 10-fold in those with deficiencies of antithrombin (7-fold for mild deficiency and 64-fold for severe deficiency), protein C (3.6-fold for mild deficiency and 7.2 fold for severe deficiency), or protein S (5-fold for mild deficiency) compared to non-pregnant women without thrombophilia^{130,133–135}. An association between thrombophilia and adverse obstetric outcomes such as recurrent miscarriage, pre-eclampsia, placental abruption, foetal growth retardation, stillbirth and foetal death has been observed^136–138, although it is not certain¹³⁹.

From a biological point of view, normal pregnancy is characterised by a hypercoagulable state. Pregnancy is associated with haemostatic changes that include increased concentrations of most procoagulant factors, decreased concentrations of some of the natural anticoagulants and reduced fibrinolytic activity (Table VI). These changes help to maintain placental function during pregnancy and minimise blood loss at delivery. However, they may also predispose to maternal thrombosis and placental vascular complications. Plasma concentrations of coagulation factors V, VII, VIII, IX, X, and XII, fibrinogen and vWF rise significantly during pregnancy, while factor XI levels tend to decrease. Total and free protein S decrease, whereas protein C and antithrombin remain substantially unchanged^140–141. Activated protein C resistance, likely caused by increasing factors V and VIII and decreasing protein S, is frequently observed in pregnancy¹⁴². The activation of coagulation is demonstrated by increasing levels of F1+2, TAT complexes, fibrinopeptide A and D-dimer^141,142. These changes occur during the whole gestational period but are more pronounced in the third trimester. The fibrinolytic system is also impaired during pregnancy, as shown by increased plasma levels of TAFI, PAI-1 and -2 (the latter of placental origin) and decreased t-PA activity^142,143. TF is largely expressed in the placenta and is markedly increased in the amniotic fluid but not in plasma¹⁴³, and, together with thrombomodulin, is involved not only in haemostasis, but also in the differentiation of placental blood vessels¹⁴⁴. Placental detachment at delivery with the ensuing release of trophoblastic substances at the site of separation is responsible, together with post-partum haemoconcentration, for the particularly high risk of VTE in the post-partum period¹⁴⁵. Three weeks after delivery, blood coagulation and fibrinolysis have generally returned to normal¹⁴⁶.

Air pollution

Air pollution consists of gaseous and particulate-matter pollutants. The former include carbon monoxide (CO), nitrogen dioxide (NO₂), sulphur dioxide (SO₂) and ozone (O₃). The latter include particulate matter (PM) with a cut-off of less than 10 μm in aerodynamic diameter (PM₁₀), fine particles of less than 2.5 μm (PM_2.5) and ultrafine particles of less than 0.1 μm (PM_0.1)¹⁴⁷. As compared with PM₁₀ and PM_2.5, ultrafine particles have a larger total surface area and hence a greater potential for carrying toxic substances, including metals, elemental and organic carbon and others. Because of their small size, ultrafine particles are deposited deep in the lung alveoli and can reach the blood stream. Particulate matter is the type of air pollutant that causes the most numerous and serious effects on human health, because of the broad range of different toxic substances that it contains^148,149. Over the last decade, a growing body of epidemiological and clinical evidence has led to a heightened concern about the deleterious effects of air pollution on the cardiovascular system^150–152. Observations from studies across North America and Europe have shown higher rates of hospital admissions for all cardiovascular causes, and a direct association was also identified with the incidence of ischaemic heart disease and failure¹⁵³. The correlation between PM_2.5 levels and onset of symptoms in 772 patients with myocardial infarction was studied in a case-crossover study: elevated odds ratios were associated with an increase of 25 μg/m³ PM_2.5 during a 2-h period before the event (OR 1.48; 95% CI 1.09–2.02) and an increase of 20 μg/m³ PM_2.5 was observed in the 24-h period before the event (OR 1.69; 95% CI 1.13–2.34)¹⁵⁴. In a study on air pollution and emergency admissions, an association was found between NO₂ (12.7% increase), PM_2.5 (8.6% increase) and the risk of hospitalisation for myocardial infarction¹⁵⁵. In another crossover study, an increase in ambient particulates of 10 μg/m³ was associated with a 4.5% increased risk of acute coronary syndromes (unstable angina and myocardial infarction)¹⁵⁶. The association between traffic-related air pollutants and acute myocardial infarction is also supported by the results of the European HEAPSS (Health Effects of Air Pollution among Susceptible Subpopulations) study¹⁵⁷.

Potential mechanisms leading to cardiovascular disease include autonomic dysfunction, systemic and local inflammation, endothelial injury, and alterations in the coagulation cascade^150,151. Changes in heart rate and heart-rate variability, arrhythmias, increase in markers of inflammation and tissue damage such as C-reactive protein, cytokines, interleukins and serum lipids are conditions induced by air pollution that affect the cardiovascular system¹⁵¹. Experimental and epidemiological studies evaluating plasma concentrations of coagulation factors in association with air pollution exposure have produced different results. While some studies found increased levels of factor VII, fibrinogen and vWF^158–161, others showed decreased levels or no change¹⁶². More recently, a novel association between air pollution and hypercoagulability was observed both in healthy individuals and in patients with DVT^163,164. Air pollution is associated with a shortened prothrombin time in healthy subjects¹⁶² and increased total plasma homocysteine levels in smokers¹⁶³. A large case-control study¹⁶⁴ showed that high mean PM₁₀ levels in the year before venous thrombosis were associated with a significantly shortened prothrombin time, and that each increase of 10 μg/m³ in PM₁₀ was associated with a 70% increase in risk of VTE. This effect was absent in women who used oral contraceptives. As the aforementioned coagulation changes induced by air pollution are similar in characteristics and degree to those observed in oral contraceptive users, it may be that coagulation is already activated by oral contraceptives so that no further enhancing effect is observed after exposure to PM₁₀.

Travel

Over the past decades, several studies have investigated the relationship between thrombosis and travel, but whether or not long-distance travel and symptomatic VTE are truly associated is still debated, as most travellers who develop DVT or PE also have one or more other predisposing risk factors¹⁶⁵. Considering an analysis of three large case-control studies on patients with clinically suspected DVT and PE, the resulting pooled odds ratio for the association between a median travel time of 7 hours and symptomatic VTE was negligible (OR 0.9; 95% CI 0.6–1.4)¹⁶⁶. However, a further analysis of the duration of travel yielded an increased odds ratio of 2.5 (95% CI 1.0–6.2) in the category of 10 to 15 hours of travel¹⁶⁶. A more recent review that summarised available data on this topic concluded that long-distance travel is associated with an up to 4-fold increased risk of VTE¹⁶⁷. The absolute risk of a symptomatic event within 4 weeks of flights longer than 4 hours was 1 in 4600 flights, whereas the risk of acute PE increased with duration of travel, being up to 4.8 per million in flights longer than 12 hours¹⁶⁷. Taken together, these data are consistent with the hypothesis that medium- to long-distance travellers have a 2- to 4-fold increased relative risk of VTE compared to non-travellers. Among the several plausible explanations for this increased risk are immobilisation and a sitting position. Tall individuals are particularly vulnerable because of cramped seating, and short individuals because their feet do not touch the floor and they, therefore, undergo extra compression of the popliteal veins¹⁶⁸. Thrombin generation among travellers has been evaluated in several studies through measurements of F1+2 and its inhibitor complex TAT. Several studies investigating the effect of prolonged immobilisation on thrombin generation and on the fibrinolytic system have yielded conflicting results¹⁶⁷ and the vast majority of these reports lacked a control group. The only controlled study published to date¹⁶⁹ failed to find a difference in F1+2, TAT and D-dimer between travellers and non travellers. The effect of hypoxia (due to decreased cabin pressure) on coagulation has been investigated in both hypobaric and normobaric conditions. The results during hypobaric, but not normobaric, hypoxia support activation of the coagulation and fibrinolytic systems, reflected in a shortened activated partial thromboplastin time, decreased levels of fibrinogen and factor VIII¹⁷⁰, factor VII antigen and TF pathway inhibitor (TFPI)¹⁷¹, increased levels of D-dimer¹⁷⁰, F1+2, TAT and factor VIIa-TF complex^171,172. However, two other studies found no difference in markers of thrombin generation during hypobaric or normobaric hypoxia^173,174. Fibrinolysis was more activated during air travel than during immobilisation or deambulation, as shown in a crossover study¹⁷⁵.

Conclusions

A large body of evidence over the past 20 years has improved our understanding of the biochemical mechanisms involved in the pathogenesis of thrombus formation, in arteries as well as in veins. We begin to understand that changes in blood coagulation, inflammation, and immune response are intricately linked and interdependent. Heightened generation of thrombin, the ultimate enzyme involved in coagulation, platelet activation and also cell-signalling effector molecules, is crucial not only in the development of VTE, but also of atherothrombosis. On the other hand, traditional cardiovascular risk factors may also play a role in VTE. Thus, pathogenesis of thrombosis has to be considered within a multifaceted perspective, as confirmed by the amount of epidemiological data on both genetic and environmental thrombotic risk factors. Nevertheless, a significant proportion of arterial and venous thrombotic episodes, especially among young individuals, occur without a plausible explanation. Further basic and clinical research is needed to reach a correct identification of new factors associated with VTE and/or arterial thrombosis, in order to assess the individual risk of thrombosis and promote more targeted prophylactic and therapeutic options.