Cancer therapy associated thrombosis

Abstract

Cancer patients have an increased risk of both arterial and venous thrombotic events compared to the general population. Both the site and stage of cancer are known to contribute to the increased risk of thrombotic events. In addition, several treatment related factors enhance the risk of thrombosis, including hospitalization, surgery, central venous catheters, radiation, and anti-cancer agents. Chemotherapy serves as a mainstay treatment for a broad range of malignancies. Chemotherapeutic agents typically exert anti-neoplastic effects through either direct cytotoxicity or inhibition of cellular processes necessary for the proliferation of malignant cells. Unfortunately, in addition to targeting malignant cells chemotherapeutic agents are also cytotoxic to non-malignant cells. More recently, targeted agents have been developed that offer improved selectivity towards malignant cells. However, some of these agents are also associated with an increased risk of both arterial and venous thrombosis. Here, we review the association between specific anti-cancer agents and thrombotic events in cancer patients. Despite an established association in most cases the mechanism by which these agents increase thrombosis is incompletely understood. Anti-cancer agents may damage the endothelium, decrease anticoagulants and/or increase procoagulants leading to activation of coagulation, and/or activate platelets. An improved understanding of the mechanisms that drive the increased risk of thrombotic events associated with anti-cancer agents will enable treatment strategies that mitigate this risk.

Introduction

Cancer is a leading cause of mortality worldwide resulting in approximately 9.6 million deaths per annum ¹. Cancer patients have a higher rate of thrombosis compared with the general population that is an important cause of morbidity and mortality ^2–5. The incidence rate of arterial thromboembolism (ATE) and venous thromboembolism (VTE) in cancer patients is estimated to be 2–5% and 4–20%, respectively ^{2, 6–8}. Both the site and stage of cancer have been found to contribute to the risk of thrombotic disease ^{6, 7, 9, 10}. For instance, cancers of the pancreas, brain and stomach are associated with a relatively high risk of VTE, whereas cancers of the breast and prostate are associated with a relatively low risk of VTE (Table 1) ^{2, 11, 12}. In addition, patients with advanced stage cancers, particularly those with metastatic disease, have a relatively high risk of VTE compared to those with early stage localized disease ^{2, 11, 12}.

Table 1:

Incidence of VT by cancer site

Cancer Site	1 year cumulative incidence VTE (%)
Brain	6.9
Pancreas	5.3	High
Stomach	4.5
AML	3.7
Esophagus	3.6
Renal cell	3.5
Ovary	3.3
Lymphoma	2.8	Medium
CLL	2.7
ALL	2.6
Lung	2.4
Colon	2.3
Liver	1.7
Uterus	1.6
CML	1.5
Bladder	1.5	Low
Prostate	0.9
Breast	0.9
Melanoma	0.5

ALL, acute lymphocytic leukemia; CLL, chronic lymphocytic leukemia; CML, chronic myeloid leukemia. Adapted from ¹¹

Treatment related factors also contribute to the increased risk of thrombosis in patients with cancer. These treatment related factors include hospitalization, radiation, surgery, central venous catheters and anti-cancer agents ^13–16. Anti-cancer agents, which includes both chemotherapies and targeted agents, complement other therapeutic strategies, such as radiotherapy and surgery. To date, more than a hundred chemotherapy and targeted therapy drugs have been approved for the treatment of a variety of malignancies ¹⁷. The growth and spread of a tumor depends on the proliferative and invasive capacity of malignant cells as well as the metastatic potential ¹⁸. Chemotherapeutic agents typically exert anti-neoplastic effects through either cytotoxic or cytostatic mechanisms, owing to their ability to kill malignant cells or inhibit their division ¹⁹. However, although chemotherapeutic agents are often highly effective in impairing the replicative capacity of malignant cells they also exert cytotoxicity towards non-malignant cells, which can lead to adverse off-target toxicities ²⁰. It was envisaged that the development of a new generation of targeted therapeutics, that more selectively target malignant cells, would reduce the degree of off-target toxicity; however, a number of these agents have significant toxicity to non-malignant cells ²¹.

A variety of anti-cancer agents have been shown to increase the risk of thrombotic events (TE) in cancer patients (Table 2) ^{15, 16, 22, 23}. However, it has been suggested that there is a significant underestimation of the burden of VTE associated with certain types of chemotherapy regimens due to the method of reporting adverse events in clinical trials in oncology ²⁴. In addition, clinical trials often have strict criteria for the exclusion of patients with a recent history of ATE or VTE. In contrast, there has been concern that a failure to adjust for time-on-treatment may over-estimate the relative risk of thrombosis associated with new therapies because patients receiving these agents may live longer than those in the control arm.

Table 2:

Anti-cancer agents and thrombosis

Agent	VT	AT
Platinum-based agents
Cisplatin	++	−
Carboplatin	++	−
Oxaliplatin	+	−
Anthracyclines
Doxorubicin	+	NR
Daunorubicin	NR	NR
Epirubicin	NR	NR
Pyrimidine antagonists
5-Fluorouracil	−	−
Gemcitabine	−	−
L-asparaginase	NR	NR
Tamoxifen	+	+
Immunomodulatory agents
Thalidomide	++	NR
Lenalidomide	++	NR
Pomalidomide	+	NR
Anti-EGFR antibodies
Cetuximab	+	−
Panitumumab	+	−
Necitumumab	+	−
VEGF targeted molecules
Bevacizumab	−	++
Aflibercept	−	NR
VEGFR RTKI
Sunitinib	−	++
Sorafenib	−	++
Axitinib	−	++
Pazopanib	−	+
Vandetinib	−	+
Lenvatinib	NR	NR
Cabozantinib	NR	NR
BCR-ABL RTKI
Imatinib	−	−
Dasatinib	++	++
Nilotinib	++	++
Ponatinib	++	++
Bosatinib	NR	NR
CDK inhibitors
Palbociclib	+	NR
Abemaciclib	++	NR
Ribociclib	+	NR

+, associated; −, not associated; ?, unclear; NR, not reported

The mechanisms that drive cancer therapy associated thrombosis are not fully understood. A primary mechanism may be the activation or disruption of the endothelium by the anti-cancer agents (Figure 1). In addition, these agents may decrease anticoagulants and increase procoagulants, such as tissue factor, leading to activation of coagulation (Figure 1). Finally, anti-cancer drugs may directly or indirectly activate platelets. (Figure 1).

Figure 1: Potential mechanisms of cancer-therapy associated thrombosis.

Anti-cancer agents may induce a prothrombotic state through a number of potential mechanisms including reduced levels of endogenous anticoagulants, induction of tissue factor procoagulant activity, activation of the endothelial, endothelial cytotoxicity and activation of platelets. Created with BioRender.com

It is important to note that it is difficult to separate the prothrombotic activity of the tumor itself from the prothrombotic activity of different anti-cancer treatments and their effect on non-malignant cells. Moreover, cancer patients typically receive multiple treatments, including combinations of cytotoxic drugs. This can make it challenging to determine the relative contribution of each agent to the thrombotic risk. In other cases, an anti-cancer drug, such as thalidomide, may not be prothrombotic when administered alone but can be prothrombotic when combined with other drugs, such as dexamethasone, as has been observed in patients with multiple myeloma (MM).

Here, we review the clinical evidence of an association between anti-cancer agents and the risk of ATE and VTE in cancer patients. In addition, we discuss potential mechanisms for thrombosis if known with the different anti-cancer agents.

Platinum-based agents

Platinum-based agents are a class of cytotoxic drugs that induce formation of toxic platinum-DNA adducts that can induce tumor cell apoptosis ²⁵. Platinum-based agents have proven to be effective in the treatment of a wide range of solid tumors, including testicular, ovarian, bladder, stomach and prostate cancer ²⁵. The first-generation agent, cisplatin, and to a lesser extent next generation agents, such as carboplatin and oxaliplatin, have been associated with a range of dose limiting off-target toxicities ²⁵.

In a study of patients with germ cell cancer treatment with a cisplatin-containing combination regimen was associated with a significantly increased incidence of VTE compared to a regimen without cisplatin (8.4% versus 3.4%) ²⁶. A similar finding was made in a complementary study ²⁷. A meta-analysis showed that cisplatin-based therapy had a significantly increased risk of VTE compared with patients receiving non-cisplatin-based therapy, 1.92% versus 0.79% (relative risk [RR] 1.67) ²⁸. In a complementary meta-analysis derived from the same series of trials no significant difference in the relative risk of ATE was observed between cisplatin and non-cisplatin-based therapies ²⁹. Together, this suggests that cisplatin containing regimens increase the risk of VTE but not ATE.

Interestingly, carboplatin does not offer an improved toxicity profile with respect to TE when compared to cisplatin. In patients with lung cancer treated with either cisplatin or carboplatin no significant difference in the incidence of TE was observed between the two agents ^{30, 31}. Consistent with these findings another study of patients with urothelial cell carcinoma found no significant difference in the incidence of TE between carboplatin and cisplatin treatment ³². In contrast, oxaliplatin-based regimens appear to be associated with a lower incidence of TE compared to cisplatin-based regimens. In a phase III trial of patients with metastatic gastroesophageal adenocarcinoma treatment with a combination therapy (5-Fluorouracil (5-FU), leucovorin), the arm receiving oxaliplatin was found to have a markedly lower incidence of TE than the cisplatin arm, 0.9% versus 7.9%, respectively ³³. In another phase III trial patients receiving a cisplatin-based regimen with epirubicin and 5-FU or capecitabine exhibited an incidence of VTE of 15.1% whereas patients receiving an oxaliplatin-based regimen with epirubicin and 5-FU or capecitabine had an incidence of VTE of 7.6% ³⁴.

The primary mechanism by which platinum-based agents induce a prothrombotic state has yet to be determined. In vitro studies showed that cisplatin increased tissue factor activity of mouse peritoneal exudate cells treated with LPS and induced apoptosis of endothelial cells and release of extracellular vesicles ^35,36. In addition, cancer cells exposed to cisplatin or carboplatin released increased numbers of extracellular vesicles ^{37, 38}. Surprisingly, levels of extracellular vesicle tissue factor activity were either unchanged after treatment of patients with metastatic testicular cancer with cytolytic chemotherapy or decreased in MM patients or pancreatic cancer patients treated with chemotherapy ^39–41. At present, the role of extracellular vesicles in the prothrombotic effect of platinum-based agents is unclear.

In a small study of cancer patients, cisplatin-based combination chemotherapy increased circulating von Willebrand Factor in a subset of patients ⁴². Another study found that administration of cisplatin and gemcitabine to patients with non-small cell lung cancer activated the coagulation system, measured by increased levels of thrombin-antithrombin complexes and D-dimer, increased levels of platelet-derived extracellular vesicles and activated platelets ⁴³.

Platinum-based agents are associated with a significantly increased risk of VTE. However, the absolute risk of VTE associated with this class of agent remains low. The cytotoxic effects of these drugs lead to endothelial damage, activation of coagulation and platelets that all likely contribute to thrombosis (Figure 2).

Figure 2: Prothrombotic mechanisms of specific anti-cancer agents.

A number of mechanisms have been identified that may contribute to the observed association between both conventional chemotherapies (blue boxes) or targeted therapies (grey boxes) and an increased risk of thrombotic events. CDK, cyclin dependent kinase; EGFR, epidermal growth factor receptor, RTKI, receptor tyrosine kinase inhibitor; VEGF, vascular endothelial growth factor; VEGFR, vascular endothelial growth factor receptor.

Anthracyclines

Anthracyclines, including doxorubicin (also called adriamycin), daunorubicin, and epirubicin, are a family of cytotoxic chemotherapy agents that interrupt topoisomerase mediated ligation of DNA during replication and induce apoptotic cell death. The use of the topoisomerase II inhibitor doxorubicin is associated with both early and late cardiomyopathy ⁴⁴. Doxorubicin is also associated with an increased risk of VTE. In patients with non-Hodgkin lymphoma doxorubicin treatment was associated with a significantly increased risk of VTE (hazard ratio [HR] 1.96) ⁴⁵. A prospective phase III trial of MM patients compared VTE rates in two cohorts that received a combination chemotherapy and thalidomide that differed only by the inclusion of doxorubicin ⁴⁶. The regimen with doxorubicin was associated with a significantly increased risk of VTE (16.0% versus 2.5%) ⁴⁶. There have been relatively few studies investigating the effect of other topoisomerase inhibitors on VTE risk.

Doxorubicin and epirubicin have been shown to increase the activity of pre-existing tissue factor on cultured endothelial cells by inducing the surface exposure of phosphatidylserine ⁴⁷. Similarly, doxorubicin and daunorubicin have been found to increase tissue factor activity in a monocytic leukemia cell line (THP-1) ⁴⁸. Doxorubicin has also been found to induce release of extracellular vesicles that may support activation of coagulation ^{49, 50}. In addition, doxorubicin disrupted the anticoagulant protein C pathway in cultured endothelial cells by reducing levels of endothelial protein C receptor ⁵¹.

Doxorubicin and cyclophosphamide have been found to increase plasma thrombin-antithrombin complexes and D-dimer in a small study of breast and lung cancer patients ⁵². Another study with breast cancer patients treated with epirubicin or doxorubicin showed modestly increased levels of thrombin-antithrombin complexes, protein C and activated protein C, but not other markers, including extracellular vesicle tissue factor activity ⁵³. Doxorubicin has also been found to induce endothelial injury in pediatric patients ⁵⁴.

Preclinical studies demonstrate that doxorubicin is prothrombotic. Administration of doxorubicin to mice led to platelet activation and enhanced platelet binding to the endothelium ^{55, 56}. Further, doxorubicin and epirubicin administration increased plasma thrombin-antithrombin complexes in mice ⁵⁷. Doxorubicin also increased thrombosis in a rat aortic loop model ⁵⁸. In a rat inferior vena cava (IVC) stasis model doxorubicin significantly increased thrombus weight ⁵⁹. Interestingly, treatment with the platelet inhibitors aspirin and clopidogrel and the thrombin inhibitor hirudin, but not the anticoagulant heparin, significantly reduced the doxorubicin dependent increase in thrombus weight in this model ⁵⁹. Similarly, in a rat IVC ferric chloride injury model of venous thrombosis doxorubicin dose-dependently increased thrombus burden ⁶⁰.

Anthracyclines, such as doxorubicin, appear to induce a prothrombotic state by inducing endothelial injury, reducing the protein C pathway, increasing tissue factor procoagulant activity and activating platelets (Figure 2).

Pyrimidine antagonists

Pyrimidine antagonists are a class of anti-metabolic chemotherapy agents that disrupt synthesis of pyrimidine containing nucleotides cytosine, uracil and thymine. Pyrimidine antagonists are all prodrugs that require metabolism by endogenous pyrimidine biosynthetic pathways. Metabolism of pyrimidine antagonist prodrugs leads to the generation of cytotoxic nucleosides and nucleotides. Pyrimidine antagonists, such as 5-FU and gemcitabine, are used to treat a wide range of malignancies, including colorectal, pancreatic, breast, head and neck, and non-small cell lung cancers.

5-FU was developed in the late 1950s and as a result there is a relatively limited amount of controlled clinical data regarding the effect of this agent on the risk of TEs. This is further complicated by the frequent use of 5-FU as part of combination chemotherapy regimens. In a phase I trial of patients with metastatic gastrointestinal cancer treated with 5-FU and leucovorin the rate of VTE was 17% ⁶¹. In a phase I trial of metastatic renal cell carcinoma treated with a combination of gemcitabine and 5-FU the rate of VTE was 43% ⁶². However, in a meta-analysis there was a non-significant increase in VTE and ATE for gemcitabine containing regimens compared to non-gemcitabine containing regimens ⁶². Similarly, a recent study showed that gemcitabine treatment was not associated with a significant increase in VTE ²⁷. Gemcitabine-associated thrombotic microangiopathy has been reported but this is rare ⁶³.

Administration of 5-FU to rabbits caused disruption of the endothelial monolayer, exposure of the subendothelial matrix and accumulation of platelet aggregates ^{64, 65}. Exposure of cultured endothelial cells to 5-FU resulted in increased reactive oxygen species generation, reduced prostacyclin release, and reduced expression of endothelial nitric oxide synthetase ^66–68. 5-FU has also been associated with coronary vasospasm ⁶⁹. Administration of 5-FU to mice increased levels of thrombin-antithrombin complexes ⁵⁷. The effect of 5-FU and other pyrimidine antagonists has yet to be assessed in preclinical models of thrombosis.

Even though pyrimidine antagonists are cytotoxic to the endothelium in preclinical models, there is no clear association between pyrimidine antagonist treatment and an increased risk of TEs in patients with cancer.

L- Asparaginase

L- Asparaginase is an enzyme that catalyzes the hydrolysis of the amino acid asparagine and is used as a chemotherapeutic agent in the treatment of acute lymphocytic leukemia (ALL) ⁷⁰. The anti-neoplastic potential of L-asparaginase in ALL stems from the dependence of leukemic lymphoblasts on asparagine present in the blood ⁷⁰. Loss of the ability to synthesize asparagine is a hallmark of leukemic lymphoblasts ⁷⁰.

A challenge in determining the thrombogenic potential of L-asparaginase is that the rate of thrombosis in pediatric ALL patients without L-asparaginase treatment is not clear. A study of pediatric ALL patients observed a TE rate of 1–2%, most of which were intracranial thrombi ⁷¹. In a meta-analysis of prospective studies, pediatric ALL patients receiving L-asparaginase had a TE rate of 5.2% ⁷². In one study of adult ALL patients the rate of TE acutely after L-asparaginase and combination chemotherapy treatment was 4.2% ⁷³. In a multicenter study, the incidence of VTE was 10% for adult ALL patients treated with L-asparaginase ⁷⁴. In a recent large prospective study the incidence of VTE in adult ALL patients treated with L-asparaginase was 16% despite antithrombin supplementation in the majority of the patients ⁷⁵. In another study of adult acute leukemia patients a 4.9 fold increase in TE was observed in patients treated with L-asparaginase compared to those not receiving L-asparaginase (11.1% versus 2.0%) ⁷⁶. Interestingly, the type of VTE associated with L-asparaginase treatment is atypical, frequently involving cerebral venous thrombosis in both pediatric and adult ALL patients ^{77, 78}.

L-asparaginase treatment is associated with the induction of a procoagulant state ⁷⁹. Treatment of ALL patients with L-asparaginase leads to decreased plasma levels of several coagulation proteins, including both procoagulants (FIX, FX and prothrombin) and anticoagulants (protein C, protein S, and antithrombin), as well as acute hypofibrinogenemia ^80–85. The initial reduction in levels of protein C, protein S and antithrombin is thought to induce a temporal procoagulant state ^{82, 83}. Interestingly, levels of procoagulant and anticoagulants decreased in proportion to their half-lives suggesting that L-asparaginase decreased synthesis of these proteins ⁸². L-asparaginase did not affect vitamin K-dependent carboxylation of the Gla domains because there was no change in the antigen/activity ratio of the clotting proteins ⁸². In addition, L-asparaginase induced tissue factor expression in endothelial cells but not monocytes, and one small study showed that L-asparaginase treatment was associated with increased tissue factor activity in plasma ⁸⁶. Consistent with activation of coagulation, L-asparaginase treatment increased levels of thrombin-antithrombin complexes and D-dimer in patients with ALL and acute lymphoblastic lymphoma ⁸⁷.

Attempts to reduce VTE in ALL patient treated with L-asparaginase have used various strategies with inconsistent results. Prophylactic fresh frozen plasma reduced the rate of VTE from 19% to 6% in adult ALL patients treated with L-asparaginase⁷⁴. However, a more recent study found that antithrombin supplementation or fresh frozen plasma did not affect the rate of VTE whereas administration of fibrinogen concentrates increased the rate of VTE ⁷⁵. Overall, it appears that L-asparaginase treatment causes a prothrombotic state in ALL patients by decreasing levels of anticoagulants and possibly by increasing TF expression (Figure 2).

Tamoxifen

The estrogen receptor antagonist tamoxifen is used in the treatment of estrogen receptor-positive breast cancer patients ⁸⁸. Early studies with tamoxifen showed that it increased the risk of TE ^{89, 90}. The increased risk of VTE associated with tamoxifen treatment is particularly apparent in older women consistent with age being a contributing factor to the risk of tamoxifen-associated VTE ^{91, 92}. A randomized clinical trial showed more TE in patients receiving tamoxifen compared with those receiving placebo (0.9% versus 0.2%) ⁹³. Another study found that the 5-year risk of VTE was 1.2% in women receiving tamoxifen and 0.5% for women not taking tamoxifen ⁹¹. In addition, in a large clinical trial more TE were observed with tamoxifen compared to anastrozole, which inhibits the synthesis of estrogen ⁹⁰. Finally, ATE and VTE were evaluated in breast cancer patients treated with chemotherapy with or without tamoxifen ⁹⁴. Adjunct therapy with tamoxifen was associated with a significant increase in TE compared with patients on observation (5.4% versus 1.6%) ⁹⁴. Tamoxifen significantly increased the risk of both ATE (1.6% versus 0.0%) and VTE (2.8% versus 0.8%) when administered with chemotherapy in premenopausal patients ⁹⁴. In a population-based study a significant temporal increase in the incidence of VTE was observed for breast cancer patients treated with tamoxifen compared to those not treated with tamoxifen in the first two years (0.45% versus 0.12%, RR 3.5) ⁹¹.

It was hypothesized that tamoxifen may decrease levels of anticoagulants. Interestingly, different results have been observed. An early study found no differences in plasma levels of antithrombin, protein C or protein S ⁹⁵. However, another study found that tamoxifen reduced levels of antithrombin and protein S, but not protein C, compared to levels before treatment ⁹⁶. More recently, it was reported that tamoxifen decreased levels of antithrombin, protein S and protein C ⁹⁷. Tamoxifen has also been found to induce activated protein C resistance ^{98, 99}. Finally, tamoxifen has been shown to increase levels of FVIII, FIX and von Willebrand Factor ⁹⁷.

The effect of tamoxifen on platelets is less clear. Tamoxifen metabolites have been shown to increase production of reactive oxygen species by platelets and potentiate agonist-induced platelet aggregation ^{100, 101}. Conversely, tamoxifen treatment has also been found to inhibit platelet adhesion and thrombin or collagen induced platelet aggregation ^{102, 103}. In a preclinical study, tamoxifen was found to inhibit thrombus formation in a mouse mesenteric arteriole ferric chloride model of thrombosis ¹⁰³. The effect of tamoxifen in preclinical models of venous thrombosis has yet to be determined.

Tamoxifen is associated with a significantly increased risk of both ATE and VTE. However, the absolute risk of VTE in the patient population receiving tamoxifen is very low. Mechanistic studies suggest that tamoxifen decreases levels of anticoagulants and increases levels of procoagulants (Figure 2).

Immunomodulatory agents

Immunomodulatory drugs, such as thalidomide, have found considerable utility as treatments for patients with MM. The dose-limiting toxicity observed with thalidomide treatment in patients with MM led to the development of several analogs, including lenalidomide and pomalidomide. The therapeutic effect of lenolidamide and pomalidamide requires the presence of cereblon, an E3 ubiquitin ligase ¹¹⁴. Indeed, the gene expression changes induced by lenalidomide were reduced in the absence of cereblon. Downstream targets of cereblon include several transcription factors, such as interferon regulatory factor 4. More recently, it was shown that lenalidomide provides anti-cancer activity by inducing selective degradation of the B cell specific transcription factors Ikaros family members 1 and 3 ^{115, 116}.

Several clinical studies have demonstrated that patients with MM treated with thalidomide are at an increased risk of VTE. MM patients have an increased risk of VTE (5–10%) compared to the general population ¹⁰⁴. The incidence of VTE associated with thalidomide alone is relatively low (~3%) ¹⁰⁵. However, the addition thalidomide to a multiagent chemotherapy regimen, including dexamethasone, has been found to increase the rate of deep vein thrombosis from 3–4% to 20–28% in MM patients ^{106, 107}. In a meta-analysis, thalidomide in combination with dexamethasone significantly increased the risk of VTE in MM patients compared to dexamethasone alone (12.5% versus 6.5%, odds ratio [OR] 2.6) ¹⁰⁸. At present, it is unclear why combined therapy with thalidomide and dexamethasone confers an increased risk of VTE when neither agent alone is associated with increased risk.

The second-generation thalidomide analogue lenalidomide is also associated with an increased risk of VTE in MM patients. Like results seen with thalidomide, two clinical trials showed that there was a higher rate of VTE on patients receiving lenalidomide together with dexamethasone compared with dexamethasone alone (14.7% versus 3.4%, and 11.4% versus 4.6%) ^{109, 110}. Another study showed a higher rate of VTE in patients receiving lenalidomide together with high-dose dexamethasone compared with patients receiving lenalidomide together with low-dose dexamethasone (26% versus 12%)¹¹¹. In a meta-analysis of randomized controlled trials lenalidomide treatment was associated with a significantly increased risk of VTE compared to non-lenalidomide treated controls, 5.8% versus 2.2% (RR 2.52) ¹¹². In a phase III trial of pomalidomide plus high-dose dexamethasone versus low-dose dexamethasone in MM patients the frequency of VTE events was 2.7% vs 0%, respectively, despite pomalidomide treated patients receiving thromboprophylaxis ¹¹³.

The mechanism by which thalidomide and its analogs combine with dexamethasone to increase thrombosis has not been identified. However, several studies have demonstrated platelet hyperactivation in response to thalidomide treatment (Figure 2). Patients treated with thalidomide were found to have increased platelet P-selectin exposure and an increased abundance of platelet-monocyte aggregates ¹¹⁷. Additionally, platelet function assessed in response to collagen and ADP stimulation was enhanced by thalidomide and dexamethasone treatment ¹¹⁸. Limited evidence also indicates that thalidomide exposure increased apoptosis and reduced cell viability in cultured endothelial cells ¹¹⁹. Further mechanistic studies are warranted to elucidate the prothrombotic potential of thalidomide in combination with dexamethasone.

Due to the increased risk of VTE in MM patients treated with thalidomide and its analogs, patients receive various thromboprophylaxis agents (aspirin, low molecular weight heparin or warfarin) ¹²⁰. The beneficial effects of aspirin are consistent with studies showing that thalidomide activates platelets. A meta-analysis of MM patients with low TE risk found that a prophylactic dose of low molecular weight heparin or therapeutic dose of warfarin was most effective in the reduction of VTE risk associated with thalidomide and lenalidomide ^{108, 121}. However, one review concluded that the type of thromboprophylaxis strategy to recommend is still unclear ¹²¹. Interestingly, in a phase III trial of MM patients receiving thalidomide no significant difference in the risk of VTE was observed between groups receiving aspirin, low molecular weight heparin and warfarin thrombophrophylaxis ¹²².

EGFR targeted antibodies

Epidermal growth factor (EGF) is a protein that signals through the epidermal growth factor receptor (EGFR) promoting cell growth and differentiation. Mutations in EGFR that lead to overexpression of this receptor, constitutive receptor activation, and excessive EGF mediated cell growth are common in patients with a variety of malignancies, including lung, colorectal and brain cancer ^{123, 124}. This has led to the development of several antibody-based therapies that inhibit EGFR signaling, such as cetuximab, panitumumab, necitumumab. Treatment with EGFR targeted antibodies has been associated with an increased risk of VTE. In a retrospective study of 4 trials of non-small cell lung cancer patients, treatment with necitumumab significantly increased the risk of VTE compared to the control arm (9.9% versus 6.6%) (RR 1.58 95%) ¹²⁵. Two meta-analyses of EGFR targeted antibody treatment, that included cetuximab and panitumumab, showed that treatment was associated with a significantly increased risk of VTE (RR 1.32 and 1.46) ^{126, 127}. Interestingly, no significant association between anti-EGFR antibody therapies and ATE has been observed ¹²⁶.

Clinical data indicates that patients treated with EGFR targeted antibodies are at a moderately increased risk of VTE. Mechanistic studies are required to provide insights into how inhibition of EGFR mediated-signaling may contribute to the observed increased risk of VTEs (Figure 2).

VEGF targeted therapies

Vascular endothelial growth factor (VEGF) signaling through VEGF receptors (VEGFRs) on endothelial cells is essential to tumor vessel angiogenesis. This led to the development of several agents that target this axis. The first of these agents to be used clinically was bevacizumab, a humanized anti-VEGF-A antibody. Bevacizumab is used in the treatment of colorectal, lung, brain, kidney, cervical, and ovarian cancer. In addition, an inhibitory VEGFR1 and VEGR2 fusion protein called aflibercept has been developed and is used in the treatment of colorectal cancer.

Meta-analyses have demonstrated a significant association between bevacizumab treatment and risk of ATE in cancer patients ^128–130. In the first of these meta-analyses, the incidence of ATE was 3.8% for bevacizumab treated patients versus 1.7% for control (HR 2.0) ¹²⁸. Two other meta-analyses found that bevacizumab treatment increased the RR of ATE 1.46 and 2.08 ^{129, 130}. Although it is possible that differences in time on treatment affected the observed association, it was reported that most ATE events occurred acutely after study initiation, potentially limiting the impact of this confounder ¹²⁸. There have also been a number of reports indicating that bevacizumab causes renal thrombotic microangiopathy ¹³¹.

The clinical association between bevacizumab treatment and VTE remains contentious. In an early meta-analysis, treatment of cancer patients with bevacizumab was found to be associated with a significantly increased risk of VTE compared to patients not receiving bevacizumab, 12.8% versus 9.8% (RR 1.33) ¹³². However, there was concern that the observed increased risk may have been attributed to an increased time on treatment owing to prolonged progression free survival in the bevacizumab treatment arm. Two subsequent meta-analyses found no significant association between bevacizumab treatment and VTE risk when adjusted for time on treatment ^{128, 133}. Similarly, a meta-analysis found no association between aflibercept treatment and risk of VTE ¹³⁴.

Preclinical studies support the notion that VEGF targeted antibody therapy is prothrombotic in venous thrombosis models. In a mouse model of lung cancer bevacizumab treatment increased venous thrombus formation in the inferior vena cava stenosis and saphenous vein ferric chloride models ¹³⁵. Treatment with bevacizumab enhanced expression of the anti-fibrinolytic protein plasminogen activator inhibitor 1 that directly contributed to the prothrombotic phenotype associated with bevacizumab ¹³⁵. Deletion of the gene encoding VEGF-A in kidney podocytes was found to induce a thrombotic microangiopathy-like phenotype in mice like that observed in bevacizumab treated patients ¹³⁶. This suggests that VEGF-A expressed by podocytes is critical for endothelial homeostasis in the kidney and that disruption of VEGF signaling in this organ can explain the thrombotic microangiopathy observed with bevacizumab ¹³⁶.

The VEGF targeted antibody bevacizumab is associated with a moderate increased risk ATE and renal thrombotic microangiopathy but not VTE. The mechanism is likely to be due to effects on the endothelium (Figure 2).

VEGFR targeted receptor tyrosine kinase inhibitors

Small molecule inhibitors called receptor tyrosine kinases inhibitors (RTKIs) have been developed that block signaling downstream of the different VEGFR receptors (VEGFR1 -3). Inhibitors in this class have potent anti-angiogenic activity. The number of approved VEGFR-targeted RTKIs has expanded rapidly and includes sorafenib, sunitinib, axitinib, pazopanib, cabozantinib, lenvatinib and vandetanib. It is important to note that these VEGFR targeted RTKIs vary significantly in their selectivity and specificity for various VEGFRs and other cell surface RTKs. This class of agent is primarily used in the treatment of metastatic renal cell carcinoma. Endogenous VEGFR signaling plays an important role in endothelial cell homeostasis with VEGFR targeted RTKIs associated with a number of vascular toxicities ¹³⁷.

A large body of data from placebo-controlled phase III trials exists for VEGFR targeted RTKIs. Meta-analysis of trials showed a significantly increased risk of ATE with sorafenib and sunitinib compared to controls (2.2% versus 0.7%, RR 3.03) ¹³⁸. A meta-analysis that included a broader range of VEGFR targeted RTKIs (sunitinib, sorafenib, pazopanib, axitinib, vandetanib) showed a significant risk of increased ATE (1.4% versus 0.5%, OR 2.26) ¹³⁹. The most common events were myocardial ischemia/infarction ¹³⁹. Some caution is warranted given that the studies do not adjust for time on treatment that is significantly longer in arms receiving VEGFR targeted RTKIs ¹⁴⁰. In contrast to ATE, meta-analyses indicate that RTKIs did not increased the risk of VTE ^141–144. A challenge in interpreting the contribution of specific VEGFR targeted RTKIs to thrombotic risk is that numerous agents with varying specificities are combined for the meta-analysis. This approach may mask the effect of specific agents on thrombotic risk.

Despite the available clinical evidence indicating that VEGFR targeted RTKIs do not increase the risk of VTE in patients, preclinical studies have shown that defective angiogenesis delays venous thrombus resolution. For instance, deletion of VEGFR2 in endothelial cells was found to decrease venous thrombus resolution in a murine model of IVC stenosis ¹⁴⁵. Impaired venous thrombus resolution was also observed in mice treated with axitinib and subjected to an IVC stenosis model of thrombosis ¹⁴⁶.

The mechanisms driving the observed increased risk of ATE in patients treated with VEGFR targeted RTKIs have yet to be determined but likely involves some form of disruption of the endothelium (Figure 2).

BCR-ABL targeted receptor tyrosine kinase inhibitors

Several BCR-ABL targeted RTKIs have been developed as anti-cancer agents and are used in the treatment of chronic myeloid leukemia (CML) and Philadelphia chromosome-positive acute lymphoblastic leukemia. Imatinib was the first of this family of agents to be used in the treatment of CML. However, the presence of specific point mutations in ABL can lead to imatinib resistance in some CML patients. This led to the development of a second generation of BCR-ABL inhibitors that includes dasatinib, nilotunib, bosutinib and ponatinib. Ponatinib has the widest inhibitory spectrum of CML tyrosine kinases. The first-generation inhibitor, imatinib, has not been associated with an increased risk of TEs. However, second-generation inhibitors have been associated with an increased risk of TEs.

Meta-analysis of trials demonstrated that treatment with second generation BCR-ABL RTKIs was associated with a significantly increased risk of TEs compared to imatinib treatment (5.88% for second generation RTKIs versus 1.04% for imatinib, OR 3.45) ¹⁴⁷. Similar results were observed with an independent meta-analysis where second generation agents were associated with a significantly increased risk of ATE when compared to imatinib (4.78% versus 0.96%, OR 3.26) ¹⁴⁸. The prothrombotic potential of ponatinib has been specifically highlighted in the literature ¹⁴⁹. In a phase II trial of ponatinib in CML patients the frequency of adverse vascular events was 8.9% and 55.9% at 11 and 60 months, respectively ^{150, 151}. The high frequency of adverse vascular events appears to be primarily driven by ATE ¹⁵¹. Indeed, a phase III trial of ponatinib in patients with CML was terminated prematurely citing concerns over the frequency of adverse vascular events ¹⁵². Interestingly, the increased frequency of ATE associated with ponatinib was equally distributed between cardiovascular, cerebrovascular and peripheral vascular events ¹⁵². The second generation RTKIs dasatinib, nilotinib or ponatinib were also individually associated with a significant increased risk of VTE compared to imatinib with OR of 3.86, 3.42 and 3.47, respectively ¹⁴⁷.

Preclinical studies have been used to explore the prothrombotic potential of BCR-ABL RTKIs in models of arterial thrombosis. Administration of nilotinib and ponatinib to mice increased thrombus formation in the mesenteric arteriole and carotid artery models of thrombosis whereas imatinib had no effect on thrombosis ^153–155. Treatment of mice with ponatinib also increased aortic endothelial-associated von Willebrand Factor expression and platelet binding 5 to 6-fold compared with controls as well as global microvascular angiopathy indicating endothelial activation ¹⁵⁶. Surprisingly, despite the increased risk of TEs in dasatinib treated patients in preclinical studies, dasatinib reduced thrombus formation in the mesenteric arteriole and carotid artery ferric chloride mouse models ¹⁵³. Platelet activation appears to be main mechanism of ponatinib associated arterial thrombosis ^{154, 155}. Ponatinib was found to induce platelet hyperactivation in response to the glycoprotein VI agonist collagen related peptide¹⁵⁵. Consistent with this glycoprotein VI -mediated mechanism ponatinib induced increased platelet adhesion to collagen ¹⁵⁴.

Second-generation BCR-ABL RTKIs, particularly ponatinib, are strongly associated with an increased risk of both ATE and VTE. Preclinical studies support the prothrombotic potential of these agents and suggest that ATE is primarily caused by activation of both the endothelium and platelets (Figure 2).

Cyclin dependent kinase inhibitors

Cyclin dependent kinase (CDK) inhibitors have emerged as an important therapy for the treatment of advanced estrogen receptor-positive and HER2-negative breast cancer. This class of agent includes the CDK 4/6 inhibitors palbociclib, abemaciclib and ribociclib. Initial clinical trials with these agents indicated that they increased the risk of VTE ^{157, 158}. In a retrospective study, a VTE incidence of 6.3% was observed in metastatic breast cancer patients receiving CDK 4/6 inhibitors ¹⁵⁹. In a meta-analysis of breast cancer patients receiving endocrine therapy, CDK 4/6 inhibitor treatment was associated with a significantly increased risk of VTE compared to control, 2% versus 0.6% (RR 2.62) ¹⁶⁰. Subgroup analysis indicated that this increased VTE risk was primarily driven by abemaciclib containing regimens ¹⁶⁰. The cellular and molecular mechanism causing the increased risk of VTE in patients treated CDK 4/6 inhibitors has yet to be determined (Figure 2).

Immune checkpoint inhibitors

Immune check point inhibitors (ICIs) impair tumor escape mechanisms by targeting programmed cell death 1, its ligands (PD-1/PD-L1) or cytotoxic T-lymphocyte-associated protein 4¹⁶¹. There are six approved ICIs (atezolizumab, avelumab, ipilimumab, nivolumab, pembrolizumab and atezolizumab). ICIs are associated with various immune-related adverse effects that differ from those associated with chemotherapeutic agents ¹⁶². A recent retrospective, single institution study reported the cumulative incidence of VTE and ATE of 12.9% and 0.6% for patients treated with ICIs over a median 8.5 months follow-up ¹⁶³. However, the study did not have a control group and it is unclear if ICIs contributed to the TE in the cancer patients. A meta-analysis found that the overall rate of VTE and ATE in patients treated with ICIs was 2.7% and 1.1% respectively. In randomized trials, ICIs did not increase the RR of VTE compared with non-ICI containing arms ¹⁶⁴. This suggests that ICIs may not significantly increase the risk of TEs.

Thromboprophylaxis

The American Society of Clinical Oncology has recently updated its guidelines for VTE prophylaxis and treatment in patients with cancer ¹⁶⁵. Routine thromboprophylaxis is not recommended for all outpatients with cancer. However, thromboprophylaxis may be offered to high-risk patients. The only recommendation for thromboprophylaxis associated with a treatment is for MM patients receiving thalidomide- or lenalidomide-based regimes with chemotherapy and/or dexamethasone. It is recommended that lower-risk MM patients receive either aspirin or low molecular weight heparin whereas high-risk MM patients receive low molecular weight heparin.

Future Investigations

There are a number of areas in this field that warrant further investigation. In the majority of cases the underlying mechanism by which these agents drive thrombosis has yet to be elucidated. Further, the prothrombotic mechanisms associated with specific combinations of therapies, such as thalidomide and dexamethasone, requires additional investigation. The arsenal of therapeutics approved for use in the treatment of patients with cancer continues to expand. Evaluation of the vascular toxicity of agents currently undergoing development may help to identify those candidates with an increased prothrombotic potential prior to extensive clinical study. Critically, through a better understanding of the mechanisms that drive cancer therapy associated thrombosis it may be possible to develop optimal anticoagulant regimens that minimize thrombotic risk. In the clinical setting systematic reviews and metanalyses of new agents using data derived from prospective randomized controlled trials is warranted. Single-agent analyses or sub-analyses would also enable the identification of specific agents in a class that might otherwise be masked. Systematic review-based approaches are preferable to the increasing number of retrospective uncontrolled studies being published that do not establish the risk of TE associated with a given agent.

Conclusions

Although cancer type and stage remain the leading drivers of thrombosis in patients with cancer, anti-cancer agents also increase the risk of TE in cancer patients. In general, anti-cancer agents appear to be associated with a more pronounced effect of the incidence of VTE than ATE, likely attributed to the lower absolute risk for ATE in cancer patients. The relative increased risk of TEs varies markedly between agents suggesting that a therapy specific approach to anticoagulation may be required. Further, TEs appear to be particularly prevalent in patients treated with multi-agent regimens where they may be synergistic effects at play. Interestingly, newer more specific targeted agents do not appear to be associated with a reduced association with TEs compared with conventional anti-cancer agents. It is possible that TEs associated with targeted therapies represent pathological “on-target” effects as opposed to “off-target” effects attributed to conventional chemotherapy. The molecular and cellular mechanisms that underpin anti-cancer agent associated thrombosis remain incompletely understood and further investigations are warranted. Critically, with the increasing number of anti-cancer agents, new issues involving adverse TEs are likely to occur.

Highlights.

• Specific cancer therapies are associated with a significantly increased risk of thrombosis

• Cancer therapies can induce a procoagulant state through a number of molecular and cellular mechanisms

• For numerous cancer therapies, however, the mechanism by which they induce thrombotic complications remain incompletely understood

Abbreviations

ALL

Acute lymphocytic leukemia

ATE

Arterial thromboembolism

CDK

Cyclin dependent kinase

CML

Chronic myeloid leukemia

EGF

Epidermal growth factor

F

Factor

HR

Hazard ratio

ICI

Immune checkpoint inhibitor

MM

Multiple myeloma

OR

Odds ratio

RR

Relative risk

RTKI

Receptor tyrosine kinase inhibitor

TE

Thrombotic event

VEGF

Vascular endothelial growth factor

VEGFR

Vascular endothelial growth factor receptor

VTE

Venous thromboembolism

5-FU

5-Fluorouracil