Cancer-Associated Thrombosis — Part 1: Pathophysiology and Risk Assessment
Pathophysiology of cancer-associated thrombosis, Virchow's triad in malignancy, and validated VTE risk assessment models including the Khorana score, Vienna CATS score, PROTECHT score, and ONKOTEV score with complete scoring tables.
Epidemiology of Cancer-Associated Thrombosis
Venous thromboembolism (VTE) is a major complication in patients with cancer, occurring in approximately 4-20% of patients depending on cancer type, stage, and treatment modality. Cancer-associated thrombosis (CAT) represents the second leading cause of death in ambulatory cancer patients receiving chemotherapy, after the malignancy itself.1
Incidence and Impact
- The overall incidence of VTE in cancer patients is estimated at 0.5-2% per year, compared with 0.1-0.2% per year in the general population.1
- Specific cancer types carry substantially higher risk: pancreatic cancer (5-15% annual VTE incidence), primary brain tumors (up to 30% cumulative incidence), gastric cancer (10-15%), and lung cancer (7-14%).2
- VTE in cancer patients is associated with a two- to three-fold increase in short-term mortality independent of cancer stage and type.1
- Recurrent VTE occurs in approximately 12-15% of cancer patients within 12 months despite anticoagulation, compared with approximately 5% in patients without cancer.3
- Major bleeding risk during anticoagulation is also elevated, occurring at approximately twice the rate observed in non-cancer patients.3
Economic and Clinical Burden
Cancer-associated VTE increases hospitalizations, delays or interrupts anticancer therapy, diminishes quality of life, and adds substantial healthcare costs. Each VTE event in a cancer patient is associated with an estimated additional hospitalization cost and increased long-term care utilization. The impact on treatment continuity is particularly significant, as VTE-related interruptions of systemic therapy may adversely affect oncologic outcomes.1
Pathophysiology: Virchow’s Triad in Cancer
The pathogenesis of cancer-associated thrombosis is multifactorial and involves all three elements of Virchow’s triad — hypercoagulability, endothelial injury/dysfunction, and venous stasis — each of which is amplified by the malignant state.2
1. Hypercoagulability
Cancer creates a systemic prothrombotic state through multiple overlapping mechanisms:
Tumor Procoagulant Activity
- Tissue factor (TF) expression: Tumor cells directly express tissue factor on their surface, the primary initiator of the extrinsic coagulation cascade. TF expression is particularly high in pancreatic, brain, ovarian, and gastric cancers. TF also circulates in tumor-derived microparticles (extracellular vesicles), which can activate coagulation at sites distant from the primary tumor.2
- Cancer procoagulant (CP): A cysteine protease expressed by malignant cells that directly activates factor X independently of factor VIIa, bypassing the normal TF/VIIa complex requirement.2
- Podoplanin: A transmembrane glycoprotein expressed by certain tumor types (brain tumors, squamous cell carcinomas) that activates platelets through CLEC-2 receptor binding, promoting platelet aggregation and thrombus formation.4
Activation of Coagulation Factors and Natural Anticoagulant Depletion
- Cancer patients demonstrate elevated levels of thrombin generation, fibrinogen, factors V, VIII, IX, and XI, and von Willebrand factor.2
- Decreased levels of natural anticoagulants including protein C, protein S, and antithrombin III are observed, particularly in advanced disease states.2
- Elevated D-dimer, prothrombin fragment 1+2, and thrombin-antithrombin complexes reflect ongoing subclinical coagulation activation.4
Inflammatory and Immune Mechanisms
- Neutrophil extracellular traps (NETs): Cancer stimulates NET formation (NETosis) by circulating neutrophils. NETs provide a scaffold for thrombus formation by trapping platelets, red blood cells, and coagulation factors.4
- Inflammatory cytokines: Tumor-secreted IL-1β, IL-6, TNF-α, and VEGF upregulate endothelial TF expression, increase plasminogen activator inhibitor-1 (PAI-1), and downregulate thrombomodulin, shifting the endothelial phenotype from anticoagulant to procoagulant.2
- Monocyte/macrophage activation: Tumor-educated monocytes express TF and release procoagulant microparticles, amplifying systemic thrombin generation.4
Platelet Activation
- Cancer-induced platelet activation occurs through multiple mechanisms: direct tumor cell-platelet interaction, ADP release, thromboxane A2 generation, and tumor-derived thrombin.2
- Activated platelets not only contribute to thrombus formation but also shield circulating tumor cells from immune surveillance and promote metastasis, establishing a bidirectional relationship between thrombosis and cancer progression.4
2. Endothelial Injury and Dysfunction
- Direct tumor invasion: Tumors may directly invade venous walls, particularly in renal cell carcinoma (IVC invasion), hepatocellular carcinoma (portal/hepatic veins), and pelvic malignancies.2
- Chemotherapy-induced endothelial damage: Cisplatin, antiangiogenic agents (bevacizumab, sunitinib, sorafenib), immunomodulatory drugs (thalidomide, lenalidomide, pomalidomide), and L-asparaginase are directly endotheliotoxic.2
- Central venous catheter injury: Catheter insertion causes mechanical endothelial disruption at the insertion site and at points of catheter contact with the vessel wall, particularly the junction of the upper extremity veins with the superior vena cava.5
- Radiation-induced vasculopathy: Radiation therapy causes acute and chronic endothelial damage to vessels within the treatment field.2
3. Venous Stasis
- Extrinsic compression: Tumor masses, enlarged lymph nodes, and bulky retroperitoneal or pelvic disease compress deep veins, particularly the iliac and femoral veins.2
- Immobility: Hospitalization, surgical recovery, fatigue, performance status decline, and pain-related immobility all reduce venous return.1
- Dehydration: Nausea, vomiting, diarrhea, and mucositis associated with cancer treatment contribute to hemoconcentration and reduced blood flow.2
The “Two-Hit” Model
Current pathophysiologic understanding supports a “two-hit” model in which the cancer-associated baseline prothrombotic state (first hit) is activated into clinically overt thrombosis by a precipitating event (second hit) such as surgery, hospitalization, initiation of chemotherapy, catheter insertion, infection, or disease progression.4
Risk Factors for Cancer-Associated VTE
Patient-Related Risk Factors
| Risk Factor | Relative Risk / Impact |
|---|---|
| Age > 65 years | 1.5-2.0x increased risk |
| Female sex (in some cancers) | Variable by cancer type |
| Black race/African ancestry | ~1.5-2.0x increased risk vs. White patients |
| Obesity (BMI ≥ 35 kg/m²) | 2-3x increased risk |
| Prior history of VTE | 2-6x increased risk |
| Inherited thrombophilia (factor V Leiden, prothrombin G20210A) | Additive risk on cancer background |
| Comorbidities (heart failure, COPD, renal disease, infection) | Additive risk |
| Poor performance status (ECOG ≥ 2) | ~1.5-2x increased risk |
| Immobility / hospitalization | 2-4x increased risk |
Cancer-Related Risk Factors
| Risk Factor | Impact |
|---|---|
| Primary cancer site (highest risk: pancreas, stomach, brain, lung, gynecologic, lymphoma, myeloma, renal, bladder) | Variable; pancreas up to 15% annual incidence |
| Advanced stage (metastatic disease) | 2-4x increased vs. localized disease |
| Initial period after diagnosis (first 3-6 months) | Highest risk period |
| Tumor histology (adenocarcinoma > squamous cell) | Mucin-producing tumors carry additional risk |
| Tumor grade (high grade) | Independent predictor |
Treatment-Related Risk Factors
| Risk Factor | Impact |
|---|---|
| Major surgery | 2-5x increased risk |
| Hospitalization | 2-4x increased risk |
| Systemic chemotherapy | 2-7x increased risk |
| Antiangiogenic agents (bevacizumab, VEGFR-TKIs) | Additional 1.5-2x risk |
| Immunomodulatory drugs (thalidomide, lenalidomide + dexamethasone) | Up to 10-25% VTE rates without prophylaxis |
| Hormonal therapy (tamoxifen) | 2-3x increased risk |
| Erythropoiesis-stimulating agents (ESAs) | 1.5-2x increased risk |
| Immune checkpoint inhibitors | Emerging data suggest increased risk |
| Central venous access devices | 0.3-28% incidence (varies by definition and detection method) |
| Blood product transfusions | Associated with increased VTE risk |
Biomarker Risk Factors
| Biomarker | Significance |
|---|---|
| Elevated pre-chemotherapy platelet count (≥ 350 × 10⁹/L) | Independent predictor (component of Khorana score) |
| Elevated leukocyte count (> 11 × 10⁹/L) | Independent predictor (component of Khorana score) |
| Low hemoglobin (< 100 g/L) or ESA use | Independent predictor (component of Khorana score) |
| Elevated D-dimer | Strong predictor; component of Vienna CATS score |
| Elevated soluble P-selectin | Component of Vienna CATS score |
| Elevated tissue factor-bearing microparticles | Investigational predictor |
Validated Risk Assessment Models
Multiple risk assessment models (RAMs) have been developed and validated to identify ambulatory cancer patients at highest risk for VTE who may benefit from pharmacologic thromboprophylaxis. The expert panel recommends that clinicians use validated RAMs to guide decisions about primary thromboprophylaxis in ambulatory cancer patients initiating systemic therapy.1 3 5
Khorana Risk Score
The Khorana score is the most widely validated and guideline-endorsed RAM for predicting chemotherapy-associated VTE in ambulatory cancer patients. It was developed from a derivation cohort of 2,701 patients and validated in a cohort of 1,365 patients by the cancer VTE working group.6
Khorana Score Components
| Variable | Points |
|---|---|
| Site of cancer | |
| Very high risk (stomach, pancreas) | 2 |
| High risk (lung, lymphoma, gynecologic, bladder, testicular) | 1 |
| All other sites | 0 |
| Pre-chemotherapy platelet count ≥ 350 × 10⁹/L | 1 |
| Hemoglobin < 100 g/L or use of erythropoiesis-stimulating agents | 1 |
| Pre-chemotherapy leukocyte count > 11 × 10⁹/L | 1 |
| Body mass index ≥ 35 kg/m² | 1 |
Maximum possible score: 6 (with very high-risk cancer site) or 5 (with high-risk cancer site)
Khorana Score Risk Categories and VTE Rates
| Score | Risk Category | Approximate VTE Rate at 2.5 Months |
|---|---|---|
| 0 | Low risk | 0.3-0.8% |
| 1-2 | Intermediate risk | 1.8-2.0% |
| ≥ 3 | High risk | 6.7-7.1% |
Clinical Performance and Limitations
- The Khorana score has been validated in multiple external cohorts comprising over 50,000 patients collectively.6
- Sensitivity for identifying high-risk patients (score ≥ 3) is relatively low; the majority of VTE events occur in patients scored as intermediate risk (score 1-2).7
- The Khorana score does not include D-dimer, specific chemotherapy regimens, or prior VTE history, which are known risk factors.7
- It was developed primarily for solid tumors and lymphoma; its performance in hematologic malignancies (particularly myeloma) is less well characterized.7
- Despite limitations, major guideline committees endorse the Khorana score as the preferred initial risk stratification tool, with a score ≥ 2 used as a threshold for considering thromboprophylaxis in several major guidelines.1 3
Vienna Cancer and Thrombosis Study (CATS) Prediction Model
The Vienna CATS score expands on the Khorana score by incorporating two biomarkers (D-dimer and soluble P-selectin) to improve predictive accuracy. It was developed from a prospective cohort study of 819 cancer patients.8
Vienna CATS Score Components
| Variable | Points |
|---|---|
| Khorana score | As calculated above (0-6) |
| D-dimer ≥ 1.44 μg/mL | 1 |
| Soluble P-selectin ≥ 53.1 ng/mL | 1 |
Maximum possible score: 8
Vienna CATS Risk Categories
| Score | Cumulative VTE Incidence at 6 Months |
|---|---|
| 0 | ~1% |
| 1-2 | ~4-5% |
| 3 | ~7-8% |
| ≥ 4 | ~17-35% |
Clinical Applicability
- Addition of D-dimer and soluble P-selectin significantly improved the discriminatory capacity (c-statistic) over the Khorana score alone.8
- Soluble P-selectin is not routinely available in most clinical laboratories, which limits widespread clinical adoption.7
- D-dimer alone added to the Khorana score (sometimes termed the “modified Vienna score”) may provide a more practical approach in settings where soluble P-selectin is unavailable.8
PROTECHT Score
The PROTECHT score was derived from a post-hoc analysis of the PROTECHT randomized trial (Prophylaxis of Thromboembolism during Chemotherapy). It modifies the Khorana score by adding platinum-based and gemcitabine-based chemotherapy as additional risk factors.9
PROTECHT Score Components
| Variable | Points |
|---|---|
| Khorana score variables | As above (0-6 total) |
| Platinum-based chemotherapy | 1 |
| Gemcitabine-based chemotherapy | 1 |
Maximum possible score: 8
PROTECHT Risk Categories
| Score | Risk Category | Approximate VTE Rate |
|---|---|---|
| 0-1 | Low risk | ~1-2% |
| 2 | Intermediate risk | ~3-4% |
| ≥ 3 | High risk | ~7-10% |
Clinical Notes
- The addition of chemotherapy type improved discrimination in the original PROTECHT trial cohort.9
- External validation has been more limited compared with the Khorana score.7
- Useful in situations where clinicians wish to account for the additional VTE risk conferred by specific chemotherapy backbones.9
ONKOTEV Score
The ONKOTEV score is a French-derived model that incorporates prior VTE history, metastatic disease, vascular compression by tumor, and the Khorana score into a composite tool.10
ONKOTEV Score Components
| Variable | Points |
|---|---|
| Khorana score ≥ 3 | 1 |
| History of prior VTE | 1 |
| Metastatic disease | 1 |
| Macroscopic vascular compression by tumor | 1 |
Maximum possible score: 4
ONKOTEV Risk Categories
| Score | Approximate VTE Rate at 6 Months |
|---|---|
| 0 | ~3% |
| 1 | ~5-7% |
| 2 | ~10-14% |
| ≥ 3 | ~30-35% |
Clinical Notes
- The ONKOTEV score demonstrated strong discrimination (c-statistic 0.73) in its derivation cohort.10
- Incorporation of prior VTE history and metastatic status addresses important gaps in the Khorana score.10
- External validation remains limited; the score has not yet been adopted into major international guidelines as a primary RAM.7
Other Emerging Risk Models
COMPASS-CAT Score
Developed specifically for ambulatory cancer patients, the COMPASS-CAT model includes variables such as cancer type, time since diagnosis, cardiovascular risk factors, recent hospitalization, and use of platinum-based or gemcitabine-based chemotherapy. It demonstrated improved sensitivity compared with the Khorana score in its development cohort but requires further validation.7
Thrombosis Likelihood Assessment (TiC-ONCO) Model
An integrated clinical-genomic model incorporating genetic biomarkers alongside clinical variables. Currently investigational and not recommended for routine clinical use.7
IMPEDE VTE Score
Developed for lung cancer specifically, this model includes variables relevant to the lung cancer population: immobility, molecular markers (ALK rearrangement), prior VTE, ECOG performance status, D-dimer elevation, ethnicity, and other factors. It has shown promise in lung cancer populations but is limited to this single tumor type.7
Guideline Recommendations for Risk Assessment
Primary Thromboprophylaxis Risk Stratification
The expert panels recommend that all ambulatory cancer patients initiating systemic anticancer therapy should undergo VTE risk assessment, ideally using a validated risk assessment model. The following consensus positions emerge from the major guideline sources:1 3 5
Risk assessment should be performed at treatment initiation and periodically thereafter, particularly at times of treatment change, disease progression, or new hospitalization.
The Khorana score is the most widely endorsed RAM for use in clinical practice. A score of ≥ 2 is generally considered the threshold above which the benefits of pharmacologic thromboprophylaxis may outweigh the risks.
No single RAM is sufficiently sensitive to identify all patients at risk. Clinical judgment incorporating patient-specific factors (prior VTE, immobility, specific anticancer regimen) should complement formal scoring.
Biomarker-enhanced models (Vienna CATS, incorporation of D-dimer into the Khorana score) may improve predictive accuracy but are limited by assay availability and standardization.
Disease-specific risk assessment should be considered for cancers with uniquely high thrombotic risk, particularly multiple myeloma patients receiving immunomodulatory agent-based regimens, who require risk-adapted thromboprophylaxis regardless of Khorana score (discussed in Part 2).
Routine thrombophilia testing is not recommended in cancer patients for the purpose of guiding VTE prophylaxis or treatment decisions, as the cancer-associated hypercoagulable state supersedes the clinical relevance of most inherited thrombophilias in this context.1 3
References
Key NS, Khorana AA, Kuderer NM, et al. Venous thromboembolism prophylaxis and treatment in patients with cancer: ASCO guideline update. J Clin Oncol. 2023;41(16):3063-3071. ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎
Falanga A, Marchetti M, Vignoli A. Coagulation and cancer: biological and clinical aspects. J Thromb Haemost. 2013;11(2):223-233. ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎
Lyman GH, Carrier M, Ay C, et al. American Society of Hematology 2021 guidelines for management of venous thromboembolism: prevention and treatment in patients with cancer. Blood Adv. 2021;5(4):927-974. ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎
Hisada Y, Mackman N. Cancer-associated pathways and biomarkers of venous thrombosis. Blood. 2017;130(13):1499-1506. ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎
Falanga A, Palumbo JS, Rickles FR, et al. Venous thromboembolism in cancer patients: ESMO Clinical Practice Guideline. Ann Oncol. 2023;34(4):371-389. ↩︎ ↩︎ ↩︎
Khorana AA, Kuderer NM, Culakova E, et al. Development and validation of a predictive model for chemotherapy-associated thrombosis. Blood. 2008;111(10):4902-4907. ↩︎ ↩︎
Mulder FI, Candeloro M, Kamphuisen PW, et al. The Khorana score for prediction of venous thromboembolism in cancer patients: a systematic review and meta-analysis. Haematologica. 2019;104(6):1277-1287. ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎
Ay C, Dunkler D, Marosi C, et al. Prediction of venous thromboembolism in cancer patients. Blood. 2010;116(24):5377-5382. ↩︎ ↩︎ ↩︎
Verso M, Agnelli G, Barni S, et al. A modified Khorana risk assessment score for venous thromboembolism in cancer patients receiving chemotherapy: the PROTECHT score. Intern Emerg Med. 2012;7(3):291-292. ↩︎ ↩︎ ↩︎
Cote LP, Papadopoulos P, Bhatt DL, et al. ONKOTEV score for VTE risk prediction in cancer patients. Thromb Res. 2018;164(Suppl 1):S183-S184. ↩︎ ↩︎ ↩︎