Circulating tumor cells and cancer-associated venous thrombosis- a missing link

In this issue of Arteriosclerosis, Thrombosis, and Vascular Biology, Gi et al¹ characterized thrombi collected post-mortem from venous thromboembolism (VTE) patients with or without cancer. Surprisingly, despite many post-mortem studies there are no studies to date analyzing the composition of the thrombi from cancer patients.

There is a long history of the association between cancer and VTE with the first report by Jean-Baptiste Bouillard in 1823 followed by a more widely quoted description by Armand Trousseau in 1865. Two early autopsy studies revealed a high rate of VTE in pancreatic cancer patients^{2, 3}. In 2006, two large autopsy studies of Japanese and Swedish cancer patient populations showed an overall prevalence of pulmonary embolism (PE) of 2.3% and 23%, respectively^{4, 5}. The different rates in these populations may reflect differences in the ethnic backgrounds. Remarkably, 42% of the Swedish pancreatic cancer patients had a PE⁵. PE was also associated with adenocarcinoma and metastasis⁵. Another study of 537 lung cancer patients in the Netherlands observed a higher frequency of PE in patients with adenocarcinoma (10.5%) compared to those with squamous cell carcinoma (3.8%) (adjusted hazard ratio of 3.1 [95% CI:1.4–6.9)⁶. More recently, autopsy analysis of 9571 cancer patients in the Netherlands found that 12.4% had a PE⁷. VTE is associated with reduced survival in cancer patients^{8, 9}.

There are several factors that contribute to the increased risk of VTE in cancer patients, including cancer type, site and stage¹⁰. For instance, the highest rates of VTE are observed in pancreatic cancer and brain cancer patients with intermediate rates of VTE observed in lung cancer and colon cancer patients¹¹. In addition, treatments, including radiation, chemotherapy and targeted therapy, increase the risk of VTE in cancer patients¹². Certain tumors, such as renal cell carcinoma and hepatocellular cell carcinoma, have a propensity for intravascular extension with subsequent thrombosis (termed tumor thrombosis). One study of 153 patients found that the most common sites of tumor thrombosis were the portal vein, the renal vein and the inferior vena cava¹³.

A prominent pathway that appears to drive VTE in pancreatic cancer patients is expression of the transmembrane receptor tissue factor (TF)¹⁴. TF binds factor VII/VIIa and the TF/factor VIIa complex activates the coagulation cascade. Cancer cells, particularly pancreatic cancer cells, express high levels of TF and release TF-positive extracellular vesicles (EVs) into the circulation that likely enhance systemic coagulation. In brain cancer, a high level of podoplanin (Pdpn) expression by cancer cells is associated with VTE^{15, 16}. Pdpn is a transmembrane protein that is expressed by cancer cells and activates platelets by binding to C-type lectin-like receptor 2 (CLEC-2). Pdpn is also released into the circulation on EVs and activates platelets.

Circulating tumor cells (CTC) were first reported by Ashworth in 1869¹⁷. CTC are rare and are rapidly cleared from the circulation by immune cells, natural killer cells and vascular endothelium. They can be present in the blood before a primary tumor is visible on imaging¹⁸. They are surrogate biomarkers of solid tumors and can be used for diagnosis, prognosis and monitoring efficacy of therapy¹⁸. A recent review concluded that CTC can be used as a prognostic biomarker for relapses and overall survival¹⁹. CTC are detected using epithelial cell molecules, such as EpCAM and cytokeratins. CTC and CTC clusters form micrometastases. Indeed, the level of CTC in the blood correlates with the incidence of metastasis²⁰. CTC especially those derived from adenocarcinomas would be expected to be prothrombotic due to TF expression. In mouse models of hematogenous metastasis, tumor cell TF, fibrinogen and platelets cooperate to increase metastasis in part by limiting the capacity of natural killer cells to clear micrometastases (Figure 1A)^21–23.

Figure 1. Role of circulating tumor cells in metastasis and thrombosis.

(A) Circulating tumor cell clusters covered with platelets and fibrin from micrometastases. (B) Thrombus containing circulating tumor cell clusters may be formed by adhesion of circulating tumor cells to the endothelium. The figure is created with BioRender.com

One study determined if infusion of Panc02 cells, a mouse pancreatic cell line that expresses TF, enhances ferric chloride-induced thrombosis of mesenteric arterioles²⁴. Surprisingly, infusion of cancer cells did not increase thrombosis. However, infusion of TF-positive EVs from the cells into mice increased thrombosis in mesenteric arterioles²⁴.

Gi et al¹ analyzed thrombi from autopsy cases of VTE with (n=114) or without cancer (n=66). There was a frequent use of chemotherapy (55.3%) or radiation therapy 24.6%) in the cancer patients, which may damage the endothelium. Patients had several different primary cancers, including lung (19%), stomach (11%), pancreatic (10%) and colorectal (6%) cancers. The histological cancer type included adenocarcinoma (47%) and squamous cell carcinoma (10%). The most common sites of DVT were the iliac and femoral veins (51.6%) and inferior vena cava (34.4%). The specimens of deep vein thrombosis (DVT) and PE contained parts of the vessel wall.

Levels of erythrocytes, platelets, neutrophils and the neutrophil extracellular trap marker citrullinated histone H3 were evaluated. In addition, thrombi were stained for TF and Pdpn. Finally, the presence of cancer cells within the thrombi from cancer patients was assessed by staining for cytokeratin. The focus of the study was characterizing thrombi from cancer patients. Strikingly, 26.6% of the thrombi from cancer patients contained cancer cells. Similar levels of cancer cells were observed in DVT and PE from cancer patients. Gi et al¹ identified 2 patterns of thrombi: cancer cells invading the venous or pulmonary artery wall and clusters of cancer cells without invasion. Some of the thrombi in the inferior vena cava may be tumor thrombi as described above. Thrombi containing cancer cell clusters may have been triggered by adhesion of CTC to the endothelium (Figure 1B). This type of thrombi was observed in 28.6% of DVT and 66.7% of PE. It would have been interesting to know the location of the thrombi relative to the primary tumor.

Most of the cancer cells in the thrombi expressed TF and to a lesser extent Pdpn and were associated with fibrin and platelets. All histological types of cancer expressed TF but the highest frequency was observed in adenocarcinoma and squamous cell carcinoma. For Pdpn, expression was predominantly by squamous cell carcinoma. As noted above, lung cancer patients with adenocarcinoma have a higher frequency of PE compared to lung cancer patients with squamous carcinoma⁶. We found that cancer patients with adenocarcinoma have high levels of EV TF activity compared to other types of cancer²⁵, which may explain this difference.

Next, Gi et al¹ compared thrombi from patients with or without cancer. Cancer-associated DVT areas but not PE areas were significantly larger (>2.5-fold) than those without cancer. Cancer-associated DVT had a lower degree of neutrophilic infiltration compared to DVT without cancer. This is somewhat surprising because leukocytosis is often observed in cancer patients¹⁴. There were no significant differences in areas of erythrocytes, platelets, fibrin and citrullinated histone H3 between cancer-associated VTE and VTE without cancer. TF but not Pdpn expression in CD163-positive monocyte/macrophages was significantly higher in cancer-associated VTE compared to VTE without cancer.

In conclusion, the study by Gi et al¹ demonstrates that a high number of thrombi in cancer patients contain cancer cell clusters. One can speculate that TF expression by the cancer cell triggered formation of thrombi. It is possible that the level of CTC is a risk factor for VTE.