Introduction

Cancer-associated thrombosis (CAT), most commonly encompassing pulmonary embolism (PE) and both upper and lower extremity deep vein thrombosis (DVT), is a common complication in patients with cancer. Compared with the general population, patients with cancer have a risk of venous thromboembolism (VTE) up to 12-fold higher at 6 months.1 This increased risk can be up to 23-fold higher in patients receiving systemic anticancer therapies.1 Thromboembolism is the second leading cause of death in individuals with cancer, and it also significantly impacts morbidity, prognosis, treatment timelines, health care resource utilization, and overall quality of life for patients and their families.2-4 Over the past 2 decades, CAT incidence has risen at least 3-fold, and it occurs frequently in ambulatory patients.1 Primary thromboprophylaxis, aimed at mitigating the risk of initial VTE events, proves beneficial for selecting high-risk individuals with cancer.5,6 This review aims to delineate common risk factors for VTE in patients with cancer, provide guidance on who should be considered for primary thromboprophylaxis, and highlight important factors to consider when initiating pharmacologic thromboprophylaxis.

Risk factors for cancer-associated thrombosis

Numerous factors contributing to CAT risk can be broadly categorized as patient-related, tumor-related, and treatment-related ones, with the presence of specific biomarkers also playing an important role.2,7,8 An overview of various risk factors is shown in Figure 1.

Figure 1. Risk factors for development of cancer-associated thrombosis (generated with BioRender.com)

a High-risk chemotherapy includes platinum-based therapy, anthracycline-containing therapy.

Abbreviations: NETs, neutrophil extracellular traps; PAI-1, plasminogen activator inhibitor-1; RF, risk factor; VTE, venous thromboembolism

Patient-related factors

Pertinent patient-related risk factors include advanced age, previous VTE, family history of VTE, comorbidities, and cardiovascular risk factors.2,9 Extreme body weight has also been reported to influence VTE risk with both low (<⁠18.5 kg/m2) and high (>35 kg/m2) body mass index (BMI) being associated with an increased incidence of CAT.9 Recently, inherited thrombophilia, such as factor V Leiden, as well as non-O blood type have also been associated with an increased risk of CAT.2,10,11

Tumor-related factors

Tumor type can significantly influence the risk of CAT. Gastrointestinal cancers, such as esophageal, gastric, hepatobiliary, and pancreatic cancers are among the most frequently reported cancers to be associated with CAT.1,2 Other tumor types that could be associated with high risk of CAT include brain, genitourinary, lung, and some hematologic malignancies.1,2 In addition, high histologic tumor grade (ie, grade 3 or 4 as per the American Joint Committee on Cancer group-level grading system), regionally advanced or metastatic cancer, vascular compression from the tumor, and time since diagnosis are associated with a higher risk of CAT.2,8,12,13

Treatment-related factors

Systemic anticancer therapies are an important risk factor for CAT. A population-based study from the United States14 reported that, compared with patients without cancer, patients with cancer not receiving chemotherapy had an odds ratio (OR) of 4.1 (95% CI, 1.9–8.5) for CAT, whereas patients receiving chemotherapy had an OR of 6.5 (95% CI, 2.1–20.2). Platinum-based chemotherapy and antiangiogenesis treatments (eg, bevacizumab) are frequently associated with VTE.2,15,16 Other treatments associated with VTE in patients with cancer include immunomodulatory agents (eg, thalidomide, lenalidomide), immune checkpoint inhibitors, hormonal therapy, and erythropoiesis-stimulating agents.2,17 Major cancer surgeries are also associated with an increased risk of VTE. Additionally, some evidence suggests an association between radiation therapy and VTE in patients with cancer.2

Biomarkers

Improved understanding of the mechanisms of CAT has led to identification of several biomarkers that correlate with the occurrence of VTE in cancer.18 These mechanisms can be categorized into 1) tumor expression of proteins that can alter host systems and 2) tumor expression of procoagulant proteins that are released into circulation and activate platelets and / or the coagulation cascade.18 For example, different tumor types can increase the platelet and leukocyte count in the circulation leading to the formation of neutrophil extracellular traps (NETs), which promote VTE by trapping platelets, red blood cells, and extracellular vesicles (EVs) with tissue factor (TF) activity.19,20 The examples of procoagulant proteins released by tumors include EVs containing TF, polyphosphate, or podoplanin (PDPN).18,19 Extracellular vesicles containing TF can activate the clotting cascade, polyphosphate-containing EVs can activate factor XII (FXII) and platelets, and PDPN-containing EVs can activate platelets. Certain tumors can also release plasminogen activator inhibitor 1 (PAI-1), which in turn inhibits fibrinolysis.18,19

Hematologic parameters, such as increased leukocyte and / or platelet counts as well as decreased hemoglobin have been shown to be reliable predictors of VTE risk in patients with cancer, and have been integrated into risk prediction models to identify patients with cancer who are at an increased risk of VTE.21-24 Other biomarkers, such as elevated concentrations of prothrombin fragment 1+2, soluble P-selectin, NETs, TF, PDPN, PAI-1, FVIII, and D-dimer have also been identified as predictors of increased VTE risk in individuals with cancer.19,25,26,27 However, the use of these biomarkers in clinical practice remains limited as most of them are not routinely available.

Identifying patients with cancer who are at high risk of cancer-associated venous thromboembolism

The high rate of VTE in patients with cancer and its significant impact on morbidity and mortality have sparked interest in the use of anticoagulants as primary VTE prophylaxis. The first few randomized controlled trials (RCTs) evaluating thromboprophylaxis in ambulatory patients with cancer compared heparin derivatives with placebo in patients with a range of solid tumor types (eg, lung, breast, ovarian, head and neck, gastrointestinal, pancreatic), and specifically in patients with advanced pancreatic cancer.28-32 These studies reported that primary thromboprophylaxis with heparin derivatives significantly reduced the risk of VTE as compared with placebo, but there was an insignificantly increased risk of major bleeding.28-32 In a recent systematic review and meta-analysis, compared with no thromboprophylaxis, low-molecular-weight heparin (LMWH) was found to reduce the incidence of symptomatic VTE (relative risk [RR], 0.62; 95% CI, 0.46–0.83) but increase the risk of major bleeding events (RR, 1.63; 95% CI, 1.12–2.35).33 Despite the decreased rate of VTE with primary LMWH thromboprophylaxis, its routine use in ambulatory cancer patients was not widely adopted or recommended in guidelines from major societies and organizations.34-39 Frequently cited reasons included a modest reduction of the absolute VTE risk (ie, 0.8%–2.4%), resulting in a relatively high number needed to treat (NNT >100), the concerns of increased risks of bleeding, as well as the inconvenience and potentially high costs of daily subcutaneous injections.8

To improve upon the abovementioned limitations, various risk prediction models have been proposed to identify patients with cancer who are at higher risk of VTE, as targeting high-risk patients may improve the risk-benefit balance. These risk models incorporate various patient-, tumor-, and treatment-related factors, and use selected biomarkers to predict the risk of VTE and stratify patients according to their risk of future VTE event.21-23,40,41 The most widely used and validated risk prediction model is the Khorana risk assessment model.21,42 The Khorana score incorporates 5 parameters: site of primary cancer, BMI equal to or above 35 kg/m2, prechemotherapy platelet count equal to or above 350 × 109/l, prechemotherapy leukocyte count above 11 × 109/l, and hemoglobin level below 10 g/dl or the use of an erythrocyte stimulating agent (Table 1).21 Each parameter is assigned 1 point, except for the site of primary cancer, where very high-risk cancers, such as the stomach and pancreatic cancers, are assigned a score of 2, and other high-risk cancers such as lung, lymphoma, gynecological, bladder, or testicular cancers are assigned a score of 1.21 In the original cohort, a score of 0 indicated a low risk of VTE, a score of 1–2 indicated intermediate VTE risk, and a score of 3 or more indicated a high risk of VTE, corresponding to rates of VTE over a median follow-up of 2.5 months being 0.8%, 1.8%, and 7.1%, respectively.21 Similar rates of VTE were observed in the validation cohort.21

Table 1. Overview of published risk assessment models for cancer-associated thrombosis (adapted from55)

Parameter

Khorana (2008)21

Vienna CATS (2010)22

PROTECHT (2012)23

CONKO (2013)40

ONKOTEV (2017)41

Very high-risk tumors (pancreatic, gastric)

+ 2

+ 2

+ 2

+ 2

Khorana score >2

High-risk tumors (lung, gynecologic, lymphoma, bladder, testicular)

+ 1

+ 1

+ 1

+ 1

Prechemotherapy hemoglobin <⁠10 g/dl

+ 1

+ 1

+ 1

+ 1

Prechemotherapy leukocyte count ≥11 × 109/l

+ 1

+ 1

+ 1

+ 1

Prechemotherapy platelet count ≥350 × 109/l

+ 1

+ 1

+ 1

+ 1

BMI ≥35 kg/m2

+ 1

+ 1

+ 1

–

D-Dimer >1.44 µg/l

–

+ 1

–

–

–

Soluble P-selectin >53.1 ng/l

–

+ 1

–

–

–

Gemcitabine chemotherapy

–

–

+ 1

–

–

Platinum-based chemotherapy

–

–

+ 1

–

–

ECOG performance status ≥2

–

–

–

+ 1

–

Metastatic disease

–

–

–

–

+ 1

Previous VTE

–

–

–

–

+ 1

Vascular / lymphatic macroscopic compression

–

–

–

–

+ 1

Proposed cutoff

≥2 / ≥3

≥3

≥3

≥3

NR

External validationa

+++

ND

+

+

+

a The + signs refer to whether an external validation was done or not, +++ signifies several external validations were done.

SI conversion factors: to convert hemoglobin to g/l, multiply by 10.

Abbreviations: BMI, body mass index; ECOG, Eastern Cooperative Oncology Group; ND not done; NR, not reported; others, see Figure 1

Since publication, the Khorana score has been retrospectively and prospectively validated in over 100 000 patients in various oncologic settings.43,44 A prospective validation study conducted by the Vienna group found that patients with cancer and the Khorana score of 2 had a 6-month cumulative incidence of VTE of almost 10%.22 Therefore, it has been proposed that a score of 2 or higher may define patients with cancer at high risk of VTE,22,42 and was further used to define the study population in 2 RCTs evaluating the use of prophylactic direct oral anticoagulants (DOACs) as risk-stratified primary thromboprophylaxis in patients with cancer (AVERT and Cassini trials).5,6

Various modifications to the Khorana score, such as the Vienna CATS,22 PROTECHT,23 CONKO,40 and ONKOTEV41 scores, have been proposed to improve the discriminatory performance.22,40 These scores include some or all the parameters of the Khorana score, with the addition of other parameters, such as biomarkers (eg, D-dimer, soluble P-selectin) and clinical factors (eg, platinum or gemcitabine chemotherapy, performance status)22,23,40,41 (Table 1). Most of these newer scores have not been extensively externally validated, and thus far have limited clinical utility.5,6,45-47

There are also novel risk assessment models that have been developed with an aim to better predict the overall risk of VTE in patients with cancer24,48-50 (Table 2). The COMPASS-CAT score, prospectively developed in a cohort of patients with breast, ovarian, lung, and colorectal cancers, includes clinical characteristics, comorbidities, and treatment parameters that are different than those in the Khorana score.24 On the other hand, the CATSCORE by Pabinger et al48 is a nomogram that uses tumor site and D-dimer levels to estimate an individual’s risk of VTE, whereas the Tic-ONCO and ONCOTHROMB risk assessment models incorporate clinical variables and genetic variants (single nucleotide polymorphism alleles) to determine the risk of VTE.48-50 There have also been efforts to develop cancer-specific risk assessment tools, such as ROADMAP-CAT for lung adenocarcinoma, THROMBOGYN for gynecologic cancer, THROLY for lymphoma, and more; however, none had been sufficiently validated nor used in clinical practice.24,44,51-53 A biomarker-based risk assessment model that uses fibrinogen and D-dimer levels has also recently been evaluated for targeted thromboprophylaxis in the TARGET-TP trial.54 In this RCT, patients (n = 328) with lung or gastrointestinal cancer deemed at high risk of VTE (based on fibrinogen and D-dimer levels before treatment or at 1 month after treatment) were randomized to thromboprophylaxis with enoxaparin vs no thromboprophylaxis. The risk stratification after 1 month of treatment is a novel and innovative use of a risk assessment model, as it has the potential to capture the prothrombotic effects of systemic therapies.54 Although, as mentioned above, D-dimer is a commonly available biomarker included in various risk prediction models, currently it is not routinely assessed in clinical practice to guide decisions on primary thromboprophylaxis in patients with cancer. This may be due to many factors, including the need for more external validation of the risk prediction models as well as the lack of a unified measurement methodology of D-dimer, which could lead to confusion.

Table 2. Overview of novel risk assessment models for cancer-associated thrombosis

Parameter

COMPASS-CAT (2017)24

Pabinger et al (2018)48

TIC-Onco (2018)49

ONCOTHROMB (2023)50

Comments

Only for use in breast, colorectal, lung, and ovarian cancer

Clinical parameters and treatment parameters different from the Khorana score

Nomogram

Adds genetic risk factors

Adds genetic risk factors

Personal history of VTE

+ 1

–

–

–

Cardiovascular risk factora

+ 5

–

–

–

Recent hospitalization for acute medical illness

+ 5

–

–

–

BMI >25 kg/m2

–

–

+

+

Family history of VTE

–

–

+

–

Tumor type

–

+

+

+

Advanced cancer stage

+ 2

–

+

+

Time since cancer diagnosis ≤6 months

+ 4

–

–

–

Antihormonal therapy in hormone-positive breast cancer or anthracycline chemotherapy

+ 6

–

–

–

Central venous catheter use

+ 3

–

–

–

Platelet count ≥350 × 109/l

+ 2

–

–

–

D-Dimer (continuous)

–

+

–

–

Genetic SNPs

–

–

+

+

Proposed cutoff

≥ 7

Personalized risk prediction

Different cutoffs based on sensitivity

Youden J statistic as the point that maximizes the Youden index

External validation

+

+

+

–

a At least 2 of the following: personal history of peripheral artery disease, ischemic stroke, coronary artery disease, hypertension, hyperlipidemia, diabetes, or obesity

Abbreviations: SNP, single nucleotide polymorphism; others, see Table 1 and Figure 1

There remain areas for improvement and knowledge gaps in risk stratification for CAT. A significant limitation of the most current risk assessment models is their reliance on a single measurement at a specific time point (commonly upon cancer diagnosis prior to start of anticancer therapies), failing to account for the dynamic nature of the patient- and treatment-related VTE risk factors over the course of a patient’s cancer journey.46,47 Another issue with the current risk assessment models is that many VTE events still occur in patients classified as low- or intermediate-risk, indicating that the predictive power is far from perfect, and other predictive markers may need to be elucidated.55 Finally, it is also important to have tools that can help better understand the risk of bleeding in patients with cancer at the same time, as the risk of bleeding is the major concern associated with thromboprophylaxis, and the risk-benefit balance is critical in the decision-making process.43

Primary thromboprophylaxis in ambulatory patients with cancer

While initial RCTs, including the PROTECHT31 and SAVE-ONCO,32 showed that thromboprophylaxis with LMWH significantly reduced the risk of VTE in unselected cancer patients across a spectrum of solid tumors, the absolute risk reduction was thought to be modest, and these results did not translate into routine clinical practice. Subsequent trials aimed to target high-risk patients by using risk prediction models, mainly the Khorana score, to improve the risk-benefit ratio. The first was the PHACS trial56 that evaluated the benefit of outpatient thromboprophylaxis with dalteparin in high-risk patients with cancer (ie, Khorana score ≥3). Although the PHACS trial did not reach the accrual target, 98 patients were randomized (50 to dalteparin and 48 to observation). Dalteparin was associated with a trend of lower risk of VTE (although not statistically significant), 12% vs 21% (dalteparin vs observation), with hazard ratio (HR) of 0.69 (95% CI, 0.23–1.89).56 The primary safety outcome of clinically-relevant bleeding events occurred in 7 patients in the dalteparin arm and 1 in the observation arm (HR, 7.02; 95% CI, 1.24–131.6). However, major bleeding only occurred in 1 patient in each arm. An individual patient-level meta-analysis of RCTs57 similarly found that LMWH decreased the risk of VTE by 64%, as compared with placebo or observations (OR, 0.36; 95% CI, 0.22–0.58) in patients with the Khorana score of 3 or higher.

More recently, 2 RCTs, CASSINI6 and AVERT,5 set out to evaluate low-dose DOACs, rivaroxaban and apixaban, respectively, as primary thromboprophylaxis in ambulatory patients with cancer and intermediate-to-high risk of VTE (defined by the Khorana score ≥2).

The CASSINI trial was an international, double-blind, placebo-controlled, randomized, superiority trial.6 Patients underwent lower-extremity compressive sonography to screen for pre-existing DVT prior to randomization. Participants without DVT were randomized to rivaroxaban 10 mg daily or placebo for up to 6 months. Repeat compressive ultrasonography was planned every 8 weeks. The primary efficacy outcome was a composite of incidental or symptomatic proximal lower-extremity DVT or PE, symptomatic upper-extremity DVT or lower-extremity distal DVT, or VTE-related death during 6-month follow-up. The primary safety outcome was major bleeding as defined by the International Society on Thrombosis and Haemostasis (ISTH). A total of 49 patients (4.5%) were diagnosed with DVT at the first screening ultrasound and excluded from randomization. The intention-to-treat (ITT) analysis included 841 patients (420 in the rivaroxaban group and 421 in the placebo group). Overall, 32.6% of the patients had pancreatic cancer, 20.9% had gastric or gastroesophageal junction cancer, and 15.9% had lung cancer. In the ITT analysis, the primary efficacy outcome occurred in 6% of the patients in the rivaroxaban group, insignificantly lower as compared with 8.8% in the placebo group (HR, 0.66; 95% CI, 0.4–1.09; P = 0.1). In a prespecified on-treatment analysis (only including outcomes occurring on or until 2 days after administration of the study drugs), the primary efficacy outcome occurred in 2.6% of the patients in the rivaroxaban group, which was significantly lower than 6.4% in the placebo group (HR, 0.4; 95% CI, 0.2–0.8). Major bleeding occurred in 2% of the patients receiving rivaroxaban and 1% of the patients receiving placebo (HR, 1.96; 95% CI, 0.59–6.49). Clinically relevant nonmajor bleeding (CRNMB) was ascertained in 2.7% of the patients receiving rivaroxaban and 2% of those receiving placebo (HR, 1.34; 95% CI, 0.54–3.32).6

The AVERT trial5 was a double-blind, placebo-controlled randomized trial comparing the use of apixaban 2.5 mg twice daily and placebo for thromboprophylaxis of intermediate-to-high-risk ambulatory cancer patients receiving chemotherapy. Eligible patients were randomized in a 1:1 ratio to receive apixaban or placebo. Unlike in the CASSINI trial, no screening ultrasonography was done at baseline or throughout the 6 months of follow-up. The primary efficacy outcome was a composite of symptomatic or incidental proximal upper-extremity or lower-extremity DVT, symptomatic or incidental PE, or PE-related death. The main safety outcome was major bleeding by the ISTH criteria. A total of 574 patients were randomized, and 563 were included in the modified ITT analysis. The most common cancer types were gynecologic cancer (25.8%), lymphoma (25.3%), and pancreatic cancer (13.6%). The primary efficacy outcome occurred in 4.2% of the patients in the apixaban group, significantly less than 10.2% of the patients in the placebo group (HR, 0.41; 95% CI, 0.26–0.65; P = 0.001). Major bleeding occurred in 3.5% of the patients receiving apixaban and 1.8% of those receiving placebo (HR, 2; 95% CI, 1.01–3.95; P = 0.046). CRNMB was noted in 7.3% of the patients receiving apixaban and 5.5% of those receiving placebo (HR, 1.28; 95% CI, 0.89–1.84). In the secondary on-treatment analysis, the primary efficacy outcome occurred in 1% of the patients in the apixaban group and 7.3% in the placebo group (HR, 0.14; 95% CI, 0.05–0.42), and major bleeding occurred in 2.1% individuals in the apixaban group and 1.1% of those in the placebo group (HR, 1.89; 95% CI, 0.39–9.24).5

When combining the results of the AVERT and CASSINI trials in an ITT, the RR of VTE at 6 months was 0.56 (95% CI, 0.38–0.83) (DOAC vs placebo), with the NNT of 24.58 The relative risk of major bleeding was 1.96 (95% CI, 0.88–4.33) for a number needed-to-harm of 77.58 Furthermore, the risk of death from any cause was unchanged with the use of thromboprophylaxis (RR, 0.92; 95% CI, 0.73–1.16).58 While combining the results of the AVERT and CASSINI trials offers a better understanding of the VTE burden and efficacy of thromboprophylaxis in patients with cancer and the Khorana score of 2 or higher, there are important distinctions between the 2 studies. Key differences are as follows: patients with primary brain cancer or cerebral metastases were excluded from the CASSINI but not the AVERT study; the CASSINI trial included a large proportion of pancreatic and gastric or gastroesophageal cancer patients, whereas the AVERT trial had more hematologic and gynecologic cancer patients; the CASSINI study focused on recruiting patients with locally-advanced or metastatic cancer, whereas the AVERT study included all patients with newly-diagnosed or progressive cancer starting chemotherapy; the CASSINI trial required all patients to undergo a screening ultrasound before randomization, while no screening ultrasound was employed in the AVERT trial.5,6,8 Despite the differences in the trial designs and patient populations included in the AVERT and CASSINI trials, international guidelines and guidance statements have been updated to incorporate these results as supporting evidence for consideration of primary thromboprophylaxis in patients with cancer at intermediate-to-high risk of VTE.34-39

Most recently, the abovementioned TARGET-TP trial54 also showed that thromboprophylaxis with enoxaparin significantly reduced the risk of CAT in the patients identified as high-risk based on a different risk assessment approach using fibrinogen and D-dimer levels (enoxaparin vs placebo, 8% vs 23%; HR, 0.31; 95% CI, 0.15–0.7; P = 0.005; NNT = 6.7). There was no increased risk of major bleeding associated with enoxaparin in this trial (1% vs 2%, P = 0.88).54

Multiple myeloma

Patients with multiple myeloma (MM) are at high risk of developing VTE due to unique disease-related complications and treatment-related toxicities, especially from immunomodulatory drugs (IMiDs).59 Patients with MM were under-represented in the original Khorana cohort, and thus many MM-specific risk factors for VTE were not included in the Khorana score. Furthermore, the Khorana score has been shown to have poor prediction performance for the risk of VTE in patients with MM.60 Four myeloma-specific risk assessment models have been proposed to predict the risk of VTE (Figure 2). The first model was based on expert consensus proposed by the International Myeloma Working Group (IMWG) and later adopted by the National Comprehensive Cancer Network (NCCN).61,62 Despite the incorporation of the IMWG/NCCN thromboprophylaxis guidelines, the cumulative incidence of VTE in the first year after MM diagnosis remained substantially high, and attempts at external validation of the IMWG/NCCN risk stratification model were unsuccessful.63,64 The SAVED and IMPEDE-VTE risk prediction models were then specifically derived and externally validated to estimate the risk of VTE in patients with newly-diagnosed MM63,65 (Figure 2). Most recently, the PRISM score was derived, which included patient-, disease-, and treatment-specific factors in the context of modern antimyeloma therapy.66 External prospective validation in a larger population is still needed before these scores can be incorporated into clinical practice.66 Overall, guidelines recommend that thromboprophylaxis should be considered in patients with MM assessed at high risk for VTE, especially newly-diagnosed patients receiving IMiD-based combination therapies.67

Figure 2. Overview of risk assessment models for venous thromboembolism in newly diagnosed multiple myeloma (generated with BioRender.com)

a Low-dose dexamethasone (<⁠480 mg/mo)

b High-dose dexamethasone (>480 mg/mo)

Abbreviations: CVC, central venous catheter; IMiD, immunomodulatory agent; LMWH, low-molecular-weight heparin; MM, multiple myeloma; mo, month; others, see Table 1 and Figure 1

Factors to consider when initiating primary thromboprophylaxis in ambulatory patients with cancer

While primary thromboprophylaxis is suggested to be considered in patients with cancer at intermediate-to-high risk of VTE, the decision to initiate primary thromboprophylaxis needs to incorporate evaluation of the risk of bleeding, potential drug-drug interactions (DDIs) and organ functions. It is also imperative to discuss patients’ preferences and potential costs associated with primary thromboprophylaxis.

Bleeding risk

Patients with cancer receiving anticoagulation therapy have 2- to 3-fold increased risk of bleeding events as compared with individuals without cancer.68,69 In a systematic review and meta-analysis including ambulatory patients with cancer and the Khorana score equal to or higher than 2 receiving primary thromboprophylaxis with either LMWH or DOACs, the risk of major bleeding with thromboprophylaxis was not significantly increased. However, a subgroup analysis restricted to patients receiving thromboprophylaxis with DOACs (AVERT and CASSINI trials), revealed a relatively high RR of major bleeding of 1.96 (95% CI, 0.69–5.5).70 There is no clear consensus on the best risk assessment model that reliably identifies patients with cancer who are at high risk of bleeding. Hence, bleeding risk is still often determined based on clinical judgment.71

Factors that increase the risk of bleeding include a history of bleeding events, recent surgery, bleeding from unresected primary tumor (ie, gastrointestinal tract, genitourinary tract and gynecologic malignancies, intracranial malignancy), metastatic disease, thrombocytopenia, anemia, liver and renal dysfunction, as well as certain anticancer therapies (eg, antiangiogenesis treatments, tyrosine kinase inhibitors).68,72,73 A secondary analysis of the AVERT trial evaluated specific risk factors associated with an increased risk of bleeding events in patients with cancer receiving apixaban thromboprophylaxis.74 Notably, apixaban thromboprophylaxis was associated with an increased risk of combined major bleeding and clinically relevant nonmajor bleeding in patients aged at least 65 years, weighting below 90 kg, with cancer types other than gastrointestinal / genitourinary / gynecologic, and the Khorana score of 2.74 Surprisingly, renal insufficiency and antiplatelet use were not correlated with bleeding events in this cohort.74 While this analysis provides interesting insights, these identified risk factors are not absolute contraindications to primary thromboprophylaxis with apixaban. Evaluating bleeding risks in cancer patients requires individualized clinical judgment, and reassessment of this risk throughout the patient’s cancer journey is imperative.

Drug-drug interactions

When considering the use of primary thromboprophylaxis in patients with cancer, it is important to consider the risk of DDIs between the anticoagulation agent of choice (LMWH, DOAC) and anticancer treatments, supportive medications, as well as medications used for comorbid conditions.75,76 Apixaban and rivaroxaban are substrates of P-glycoprotein (P-gp) and cytochrome (CYP) 3A4, and therefore other drugs that significantly affect P-gp or CYP3A4 metabolism could potentially affect the safety or efficacy of DOACs. Previous reviews have extensively summarized many of the common anticancer therapies that affect the P-gp or CYP 3A4 pathways, and are a good resource for clinicians when evaluating for potential DDIs.77-79 LMWH is not metabolized through the CYP 3A4 or P-gp pathways and is therefore the recommended anticoagulant if DDIs are a concern.76 However, recent studies showed potentially important pharmacodynamic interactions between LMWH and vascular endothelial growth factor receptor tyrosine kinase inhibitors suggesting that an increased risk of bleeding may still need to be considered.76 As anticancer therapies continue to evolve rapidly, more potential DDIs could affect the efficacy and safety of anticoagulants. It is also worth noting that many DDIs are theoretical concerns, and more data are needed on their clinical relevance. Clinicians should routinely evaluate potential DDIs with anticoagulation and other medications by using existing evidence and consulting resources such as product labelling, large pharmacology databases with drug interaction information, and / or consulting with the local pharmacists.76

Organ function

Renal and liver dysfunctions are common in patients with cancer. As both LMWH and DOACs rely on the kidney and / or liver for their metabolism and / or elimination, it is important to consider organ function when evaluating the primary thromboprophylaxis.80 Renal and / or liver dysfunction can result in drug accumulation, which consequently leads to higher plasma concentration of an anticoagulant and an increased risk of bleeding.80 In existing trials evaluating primary thromboprophylaxis in patients with cancer, individuals with creatinine clearance below 30 ml/min or with significant liver dysfunction were excluded, and thus the results may not be applicable to patients with significant renal and / or hepatic impairment.5,6,31,32

Central venous catheter

Central venous catheters (CVCs) are frequently placed to facilitate the administration of chemotherapy, blood products, parenteral nutrition, and other supportive therapies in patients with cancer. The presence of a CVC is also an important risk factor for VTE in patients with cancer.81 Even patients with lower-risk tumor types (eg, breast cancer, colorectal cancer) have been reported to have higher risk of VTE because of the CVC.81,82 In a recent systematic review and meta-analysis, primary thromboprophylaxis was associated with a favorable risk-benefit ratio in patients with cancer and a CVC.83 Similar findings were also reported in a subgroup analysis of the AVERT trial.84 These results suggest that there may be a role for primary thromboprophylaxis in patients with cancer and a CVC, but it requires further investigation and has not been recommended by major clinical practice guidelines.35,38,39 A large multicenter RCT comparing primary thromboprophylaxis with rivaroxaban 10 mg once daily with placebo in patients with cancer and a CVC is currently underway (NCT05029063).82

Duration of primary thromboprophylaxis

The optimal duration of primary thromboprophylaxis in ambulatory patients with cancer receiving chemotherapy is not well-established. In both the AVERT and the CASSINI trials, a follow-up period of 6 months was chosen, as it corresponds to the time period of the highest risk of VTE.5,6,85 Six months is also the approximate duration of many chemotherapy regimens. Based on current evidence, if primary thromboprophylaxis is initiated, it should be continued for at least 6 months as per the AVERT and CASSINI trials. It is important to remember that the risk-benefit ratio of primary thromboprophylaxis must be regularly reassessed to determine the appropriateness of the ongoing anticoagulant use. Whether thromboprophylaxis should be continued beyond 6 months can be determined based on a patient’s ongoing risk of VTE, risk of bleeding, cancer status, as well as patient’s preference.

Conclusions

Cancer-associated thrombosis is a common complication among ambulatory patients receiving chemotherapy. Many risk factors including cancer type and stage, patient characteristics, and anticancer treatments influence the risk of CAT. Primary thromboprophylaxis with LMWH or DOACs is an effective and relatively safe option to prevent CAT, and could be considered in patients with cancer who are at intermediate-to-high risk of VTE, after a careful review of the patient’s bleeding risk, organ function, and potential DDIs.