Diagnosis, Treatment, and Prevention of Cancer-Associated Thrombosis

Cancer-Associated Venous Thromboembolism Is Common

Since the early observations by Trousseau and others, the strong association between malignancy and development of VTE has been confirmed in multiple retrospective and observational prospective studies. The estimated annual incidence of VTE in patients with cancer is 0.5%, compared with 0.1% in the general population. During the clinical course of cancer, the cumulative incidence of symptomatic VTE has been reported to be approximately 15%, with a range of 3.8% to 30.7%. It has been estimated that up to one-fifth of all patients with VTE have an underlying cancer. As a group, malignancies are associated with a fourfold to sevenfold increase in VTE, with a 28-fold increase in risk of VTE in certain types of cancer. Researchers from the Netherlands observed a sevenfold increase in risk of VTE among patients with malignancy compared with persons without cancer. In a large population-based study, the presence of a malignant neoplasm was associated with an odds ratio (OR) of 6.5 for VTE compared with control subjects without cancer. In addition, incidental asymptomatic VTE is noted on routine imaging in 1.5% to 6.3% of patients.

The incidence of VTE is not uniformly distributed among different cancer types. In general, pancreatic, gastric, brain, ovarian, renal, and hematologic malignancies have been associated with a higher risk of VTE, whereas cancers of the head and neck, breast, prostate, and esophagus are associated with lower VTE rates ( Fig. 33.1 ). However, because of the higher incidence and prevalence of lung, colon, prostate, and breast cancers, these malignancies are associated with the highest absolute number of VTEs. The histologic type and extent of the malignancy also influence VTE risk. For example, lung adenocarcinoma is associated with a greater risk of VTE than is squamous cell carcinoma of the lung. Metastatic cancers are more likely to be associated with VTE than are localized malignancies. For the most common cancer types, the relative risk (RR) of symptomatic VTE has been estimated to be fourfold to more than twentyfold higher for metastatic malignancies. Even within the same patient with cancer, VTE risk varies throughout the course of the person's disease because of effects of other intrinsic and extrinsic factors such as cancer stage, surgery and length of anesthesia, chemotherapeutic and hormonal therapies, age, the presence of indwelling central venous catheters (CVCs), immobilization, inherited thrombophilia, a previous personal history of VTE, and infections. The first year after cancer diagnosis, and especially the first few months, is particularly associated with the development of VTE, although the risk of VTE remains elevated for many years after the initial cancer diagnosis. Blom and colleagues noted an adjusted OR for VTE of 53.5 (95% confidence interval [CI], 8.6–334.3) during the first 3 months after cancer diagnosis, with a subsequent reduction to 14.3 (95% CI, 5.8–35.2) 3 to 12 months after diagnosis, and a reduction to 3.6 (95% CI, 2.0–6.5) 1 to 3 years after diagnosis, but the increased VTE risk persisted for 15 years after cancer diagnosis.

Similar to solid tumors, hematologic malignancies are also associated with an increased incidence of cancer-associated VTE. In fact, in a population-based study, patients with hematologic malignancies had the highest risk of VTE among different tumor types (OR, 28 [95% CI, 4–200]). Among patients with hematologic malignancies, patients with multiple myeloma (MM) are at a higher risk for VTE due to cancer and treatment-related factors. It has been reported that cancer-associated VTE complicates the course of MM in at least 10% of patients. In an analysis from the California Cancer Registry, the 1-year cumulative incidence of VTE in persons with acute myeloid leukemia, lymphoma, chronic lymphocytic leukemia, acute lymphoblastic leukemia, and chronic myeloid leukemia were 3.7%, 2.8%, 2.7%, 2.6%, and 1.5%, respectively.

Bidirectional Relationship Between Cancer and Venous Thromboembolism

The relationship between cancer and VTE is bidirectional; VTE, especially idiopathic VTE, can be a harbinger of an occult malignancy. Population-based observational studies have documented an increased risk of malignancy after a first episode of idiopathic VTE. In an analysis of the Swedish Cancer Registry, Baron and colleagues found that at the time of VTE diagnosis or during the first year of follow-up, there was a large increase in the risk for virtually all cancers (standardized incidence ratio of 4.4 [95% CI, 4.2–4.6]) The increased risk of future cancer diagnosis in patients with VTE compared with those without VTE persisted for 2 through 25 years after admission to the hospital with the index VTE. In a large analysis of the Danish Cancer Registry, an increased standardized incidence ratio of 1.3 (95% CI, 1.21–1.33) for cancer diagnosis was found for patients with a VTE episode. The risk was substantially elevated only during the first 6 months of follow-up and declined rapidly thereafter to a constant level slightly above 1.0 a year after the VTE episode. The risk of cancer was twofold higher for patients diagnosed with idiopathic versus triggered VTE. The association was most pronounced for cancers of the pancreas, ovary, liver, and brain. Unfortunately, extensive cancer screening at the time of VTE diagnosis is not associated with improved outcome because most cancers associated with VTE are metastatic at the time of VTE diagnosis. Consequently, the possibility of an underlying cancer among patients with an idiopathic VTE always should be taken into consideration, but cancer screening should be limited to age-appropriate screening procedures in the absence of obvious signs of an underlying malignancy.

Venous Thromboembolism Is Associated With Worse Outcomes in Patients With Cancer

Studies have shown that diagnosis of VTE in patients with malignancy is associated with worse outcomes and shortened survival. VTE is the second most common cause of mortality after cancer itself among patients with malignancy. In fact, 15% of deaths occurring in hospitalized patients with cancer are attributable to PE. Chew and colleagues analyzed approximately 235,000 patients with cancer in the California Cancer Registry who were linked to the California Patient Discharge Data Set and found in a multivariate analysis that diagnosis of VTE during the first year of follow-up was a significant predictor of death and decreased survival for most cancer types and stages. The 1-year survival of patients with cancer who were diagnosed with VTE is one-third (12% versus 36%) that of patients with cancer who do not have VTE. Levitan and colleagues noted a threefold increase in 6-month mortality among patients with cancer who had VTE compared with those without VTE.

In addition to symptomatic VTE, unsuspected asymptomatic PE found on routine cancer staging computed tomography (CT) was found to adversely affect survival in patients with cancer. In a retrospective matched cohort study, the hazard ratio (HR) for death among patients who had cancer with unsuspected PE detected on staging CT was 1.51 (95% CI, 1.01–2.27), with the risk attributable to proximal rather than subsegmental unsuspected PE. Another retrospective study showed that patients with cancer who were diagnosed with and treated for incidental PE had similar rates of recurrent VTE, bleeding, and mortality compared with patients with cancer who had symptomatic PE. Moreover, several studies in patients with cancer have correlated increases in biomarkers of thrombin generation, even in absence of documented VTE, with more aggressive cancer biology and worse outcomes. Despite ongoing anticoagulation therapy, patients with cancer have a 3.2-fold increased risk of recurrent VTE compared with patients without cancer (12-month cumulative VTE incidence 20.7% versus 6.8%). In addition, the incidence of major bleeding is 2.2-fold higher in patients with cancer compared with patients without cancer (12.4% versus 4.9%). The risks of recurrent VTE and bleeding appear to correlate with the extent of the malignancy. As expected, the development of VTE in patients with cancer is associated with a significant increase in the consumption of health care resources.

Tumor-Specific Factors

Malignancy is characterized by a bidirectional interrelationship connecting cancer growth, progression, and metastasis with activation of the coagulation cascade and subsequent thrombin generation and inflammation. As discussed earlier, tumor-associated factors such as site, histology, and stage all influence the risk of thrombosis. Cancer cells can disrupt the hemostatic balance via several different pathways, including production of procoagulant; profibrinolytic, proproteolytic, and proaggregating activities; expression of adhesion molecules that mediate direct interactions with host vascular and blood cells; and secretion of proinflammatory and proangiogenic cytokines. Fig. 33.2 depicts some of the tumor-specific factors that contribute to the hypercoagulable state of cancer. Cancer procoagulant is a cysteine protease expressed only by malignant cells that can directly activate factor X independent of factor VII. In addition, cancer procoagulant has been demonstrated to activate platelets, further adding to its prothrombotic potential. Cancer procoagulant has been identified in a wide variety of cancer types and has been noted to have increased activity at disease onset with a subsequent slow decline.

Tissue factor (TF) is the principal initiator of the coagulation protease cascade and is normally expressed as a transmembrane glycoprotein on vascular subendothelial cells. Therefore TF is not normally in contact with blood unless vascular endothelial integrity is compromised or if its expression is induced on endothelial cells by inflammatory stimuli. Many types of cancer cells such as pancreatic adenocarcinoma and malignant glioma express high levels of TF, which can subsequently lead to activation of both factor X and factor IX, thrombin formation, and ultimately, fibrin clot formation. TF, which can also be produced by cells in the cancer microenvironment depending on cancer type and context, has been associated with cancer initiation, metastasis, progression, and angiogenesis, in addition to activation of coagulation. Laboratory studies have demonstrated a correlation between increasing TF expression in glioma cells and histologic grade, progression, and the risk of intravascular thrombosis. In pancreatic adenocarcinoma, it has been shown that fibrinogen and plasminogen activator inhibitors (PAIs) 1, 2, and 3 exist throughout the tumor stroma and that tumor cells stain for TF, PAI, prothrombin, and several other coagulation factors. In another study, expression of TF was found to correlate strongly with the degree of histologic differentiation in pancreatic adenocarcinoma. Tumors with stronger immunoreactivity for TF were more poorly differentiated. Based on these observations, it has been theorized that local coagulation activation may regulate growth and progression of pancreatic adenocarcinoma. TF can drive cancer progression by coagulation-dependent and coagulation-independent mechanisms. The circulating form of TF has been proposed to contribute more significantly to the pathogenesis of cancer-associated VTE than TF expressed on primary cancer cells. In a retrospective study, patients with cancer who had circulating TF-expressing microparticles had a 1-year cumulative incidence of VTE of 34.8% versus 0% in those without detectable TF-bearing microparticles. This and other similar observations suggest that cancer-derived TF-bearing microparticles are thrombogenic in vivo and are likely to play a central role in the pathogenesis of cancer-associated VTE. The clinical utility of this assay has not been demonstrated yet, and it remains an investigational tool at the current time.

Most patients with cancer have been found to exhibit increased levels of coagulation factors V, VIII, IX, and XI, and biomarkers of thrombin generation such as prothrombin fragment 1+2 and D dimer. As mentioned previously, elevated biomarkers of thrombin generation have been associated with aggressive cancer biology and worse clinical outcomes. The fibrinolytic system can also be significantly dysregulated in patients with cancer. Malignant cells can express different proteins in the fibrinolytic system, including urokinase-type and tissue-type plasminogen activators (UPA and TPA, respectively) and PAIs. For example, acute promyelocytic leukemia is often associated at onset with a life-threatening coagulopathy associated with hypofibrinogenemia, increased thrombin generation and fibrin degradation, and prolonged prothrombin and thrombin times. The leukemic cells in acute promyelocytic leukemia can express the UPA receptor on their surface, which activates UPA and TPA, contributing to the fibrinolytic state characteristic of the disease. Other leukemic cells are also capable of expressing various fibrinolytic and proteolytic enzymes that can mediate the bleeding complications seen in some patients with acute leukemia. Similarly, plasminogen activators, PAIs, and other proteins that regulate the fibrinolytic system are also expressed by solid tumor cells, and the resulting imbalance in fibrinolysis may contribute to the hypofibrinolytic and procoagulant state seen in some affected patients. In addition, cancers can be associated with deficiencies in the natural anticoagulants, antithrombin, protein C, and protein S, further promoting the cancer-associated thrombogenic state. The degree of activation of the coagulation cascade and fibrinolysis differs among various tumor types. As noted earlier, some patients with cancer will exhibit clinically evident manifestations of activated coagulation such as DIC and/or VTE or arterial thromboembolism, whereas many other patients with cancer will only have laboratory markers of a procoagulant state such as elevated D dimer.

Cancer cells can also modulate the hemostatic balance in an indirect fashion through their interaction with host immune cells such as monocytes and macrophages, leading to activation of platelets and factors X and XII. Tumor cells can directly produce inflammatory cytokines or indirectly stimulate their production by host cells (leukocytes and endothelial cells), promoting a hypercoagulable state. These inflammatory cytokines, such as tumor necrosis factor–α (TNF-α), interleukin-1 (IL-1), and vascular endothelial growth factor (VEGF) can induce TF production by endothelial cells and monocytes, stimulate PAI-1 production, and downregulate expression of the natural anticoagulant protein, thrombomodulin, on endothelial cells. In addition, these cytokines can lead to vascular endothelial cell damage and conversion of vascular lining into a thrombogenic surface. In addition to activation of TF production by endothelial cells and monocytes, the effects of VEGF include induction of angiogenesis and increased local vascular permeability, therefore increasing the exposure of TF and promoting cancer-associated thrombogenesis. The acute-phase reactants induced by inflammatory cytokines in patients with cancer include procoagulants such as von Willebrand factor (vWF), factor VIII, and fibrinogen, therefore favoring a thrombogenic hemostatic milieu. Tumor cells can also promote nonenzymatic activation of factor X through the sialic acid moieties of mucin produced by adenocarcinomas. Although increased levels of factor VIII, vWF, fibrinogen, PAI, and markers of thrombin generation and fibrin degradation have been associated with more advanced cancers and worse outcomes, these biomarkers have not shown any benefit to date in selecting patients with cancer who might benefit from primary anticoagulant VTE prophylaxis.

Tumor cells can also directly aggregate platelets and secrete important platelet aggregation agonists such as thrombin and adenosine diphosphate. In addition to these biochemical procoagulant mechanisms, direct cell-cell interactions and the local mass effect of tumors contribute to VTE pathogenesis in patients with cancer. Vascular invasion and physical compression by the tumor can mechanically obstruct venous blood flow, leading to venous stagnation, endothelial lining damage, and local activation of the coagulation cascade, all of which predispose to VTE. In patients with myeloma, increased plasma viscosity, elevated levels of circulating immunoglobulins, autoantibodies targeting natural anticoagulants, and secretion of inflammatory mediators with procoagulant activity have all been proposed to contribute to the pathogenesis of VTE.

Host-Specific Factors

Studies have shown that some host-specific factors can also modify the risk of VTE. Older age (≥65 years) has been modestly associated with cancer-associated VTE in hospitalized patients. Poor performance status in patients with lung cancer who are receiving chemotherapy, which might be a surrogate for limited mobility, has been prospectively associated with increased VTE risk. Patient race (African Americans have a higher risk, whereas Asians have a lower risk) and comorbidities such obesity and renal disease have also been associated with increased risk of cancer-associated VTE, whereas gender does not significantly influence risk. ABO blood group status, which was found to affect the risk of VTE in the general population, has also been found to modify the risk of VTE in patients with malignant glioma. Although the pathophysiology of the ABO blood group's association with VTE is not fully clarified, ABO blood group status does influence factor VIII and vWF levels, which might mediate this effect.

The presence of prothrombotic genetic alterations has been also shown to influence the risk of VTE in patients with cancer. In a large population-based, case-control study of 3220 consecutive patients aged 18 to 70 years in the Netherlands, patients with cancer who were carriers of factor V Leiden or the prothrombin gene 20210A mutation had a fourfold increased risk of VTE compared with patients with cancer who did not have these mutations. In contrast, a smaller study from Brazil found no significant difference in the prevalence of four thrombophilic genetic mutations/polymorphisms (factor V Leiden, prothrombin gene 20210A mutation, FXIII Val34Leu polymorphism, and methylenetetrahydrofolate reductase [MTHFR] C677T polymorphism) in patients with cancer who did and did not have VTE. A third study in patients with gastrointestinal adenocarcinoma found a significant association between VTE and factor V Leiden mutation but not with prothrombin gene 20210A mutation or the MTHFR C677T polymorphism. A matched nested, case-control study of breast cancer prevention using tamoxifen found no significant association between the factor V Leiden mutation or the prothrombin gene 20210A mutation and VTE risk.

Environmental Factors

Environmental factors significantly modulate VTE risk in patients with cancer. As expected, many modifiable VTE risk factors increase VTE risk in patients with and without cancer.