Hypercoagulability preceding cancer M. NIERODZIK and S. KARPATKIN
Does hypercoagulability awaken dormant tumor cells in the host?
The association of venous thrombosis and cancer has been recognized for over 40 years and its description first attributed to Armand Trousseau [
1], who later made the diagnosis on himself. This observation has been confirmed in 13 cohort studies reporting venous thromboembolism (VTE) in association with cancer with a prevalence range of 10-20%[
2]. Of particular interest, however, is the association of occult cancer with VTE. In a series of 14 cohort reports the prevalence of occult cancer with idiopathic VTE in individuals aged 40 years or over was ~4-10%[
2].
However, a cause and effect relationship between
in vivo thrombin generation, thrombosis and cancer was not rigorously established until ~13 years ago when Nierodzik
et al. first reported enhanced murine experimental pulmonary metastasis following the intravenous injection of minute concentrations of thrombin (titered not to reduce the platelet count). This group also demonstrated enhanced adhesion of tumor cells to thrombin-treated platelets with as little as 0.001-0.01 U mL-1 of thrombin exposure for 1 h on a microtiter plate [
3]. Platelet-tumor interaction is required for experimental pulmonary metastasis [
4,5]. In ensuing publications they then observed that thrombin had a protease activated receptor (PAR)-1 binding site on tumor cells [
6] and that thrombin could activate tumor cells to adhere more avidly to platelets, fibronectin and von Willebrand factor [
7] and endothelial cells [
8]
in vitro. These same thrombin-treated tumor cells could undergo marked enhancement in experimental pulmonary metastasis [
7]. These observations were confirmed by Wojtukiewicz
et al. [
9]. Most tumor cells studied contain the thrombin PAR-1 [
6,10], which is rate-limiting for tumor adhesion to fibronectin [
11] as well as experimental metastasis: as demonstrated by transfection experiments in B16F10 melanoma cells [
11]. Other studies have revealed that thrombin can induce tumor growth
in vitro as well as
in vivo[
11,12]. Indeed, thrombin can act as a mitogenic agent for mesenchymal tissue such as fibroblasts, endothelial cells and smooth muscle cells [
13-18].
The requirement of angiogenesis for tumor growth and metastasis is well recognized. Recent studies have revealed that thrombin has a significant stimulatory effect on angiogenesis in that it can induce vascular growth factors, VEGF [
19], its receptor KDR [
20,21] and angiopoietin-2 [
22]. Thrombin can also stimulate the release of VEGF [
23] and angiopoietin-1 from platelets [
24], as well as induce tube formation of endothelial cells in a matrigel membrane system [
25]. Indeed, direct application of as little as 0.05-0.1 U mL-1 of thrombin to a chorioallantoic chick membrane can induce angiogenesis 2-3-fold over a 72-h period. This could be blocked with hirudin, as well as a VEGF receptor inhibitor (KDR-Fc) and an angiopoietin-1 receptor inhibitor (Tie2-Fc), and could be initiated by PAR-1-specific peptide (TRAP) [
26].
However, these animal data in which tumor cells were treated with exogenous thrombin do not represent the true pathophysiological situation, because: (i) these studies give no information of endogenous thrombin production/concentration at the tumor-host interface; (ii) experimental tail vein tumor metastasis is a highly artificial situation in which 1 × 105-1 × 106 tumor cells are injected as a bolus into the tail vein of a mouse and primary pulmonary entrapment and tumor nodule appearance monitored. However, spontaneous metastasis in the host represents the release of ~10 000-fold fewer cells into the circulation, with yet an unknown fraction of these cells actually implanting and growing. However experiments which measured the effect of hirudin (an exquisitely potent inhibitor of thrombin) on spontaneous metastasis of a highly aggressive breast cancer (4T1) tumor cell line following implantation of tumor into the flank demonstrated that endogenous thrombin significantly contributed to tumor implantation, growth, seeding, spontaneous metastasis and death [
27].
Extrapolating from these experimental data, it is likely that low-grade thrombin generation may be harmful to some patients with malignancies, because it may predispose to enhanced growth and metastatic progression of the lesion. Indeed, there is abundant evidence that many tumor cells have constitutively active tissue factor on their surface which can activate the coagulation system with generation of thrombin [
28-34]. Low-grade intravascular coagulation as diagnosed by increased fibrinogen turnover [
35], increased plasma levels of fibrinogen/fibrin-related antigen [
35] and increased plasma levels of fibrinopeptide A have been observed in most patients with solid tumors [
36-38]. One study noted elevated fibrinopeptide-A levels in 60% of patients at time of disease. Persistent elevation was associated with a poor prognosis [
38]. Thus a ‘vicious’ autocrine cycle is established. It is of interest that tissue factor expression correlates with hematogenous metastasis in melanoma cells [
39-41] and is associated with the leading edge of invasive breast carcinomas [
42]. In addition, enzymatically active thrombin has been reported to be present on surgically removed tumor specimens, including malignant melanoma, by affinity-ligand histochemical analysis [
43], and thrombin-receptor (PAR-1) overexpression has been reported in malignant invasive melanoma and breast tumor cell lines
in vitro, as well as human breast metastatic tissue
in vivo[
44].
Of particular interest are the observations of Shulman and Lindmarker [
45], who followed 854 patients who had been treated for 6 months vs. 6 weeks with coumadin for deep vein thrombosis for the ensuing 6 years. Those patients treated for 6 months developed significantly less cancer during the 6-year observation period than those exposed to anticoagulant for 6 weeks, odds ratio (OR) 1.6 [95% confidence interval (CI) 1.1, 2.4,
P = 0.02], and when confined to 40 patients with urogenital tumors, the OR increased to 2.5 (95% CI 1.3, 5.0). It is intriguing to speculate that these data, as well as recent animal observations [
27], strongly suggest that thrombin nourishes tumor dormancy and growth. Indeed, one can further speculate from these observations that tumor cells lie dormant and/or grow slowly in the systemic organs of many cancer-prone individuals and that this dormancy is maintained by endogenous anticoagulants [antithrombin III, protein C, α2-macroglobulin, thrombomodulin, tissue factor pathway inhibitor (TFPI)], and other undescribed factors, as well as immune surveillance. Indeed it has been estimated that it takes 2-8 years for cancer cells to be detectable after transformation to malignant cells considering the doubling time and volume required for detection [
46].
What is the evidence for tumor dormancy? It is well recognized that many if not all subjects harbor tumor cells at various states of dedifferentiation [
47]. Autopsies of individuals dying from accidents or non-malignant disease often reveal microscopic or
in situ cancer. One-third of women aged 40-50 who did not have cancer detectable during their lifetime are estimated to have
in situ tumors in their breasts. Similar estimates have been made for prostate cancer. All autopsied individuals aged 50-70 have
in situ carcinomas in their thyroid gland despite its rare diagnosis during life. There are normal individuals with the ‘Philadelphia’ chromosome translocation, which is found in 95% of patients with chronic myelocytic leukemia and contributes to its etiology [
48]. Chronic lymphatic leukemia (CLL) clones have been described in ‘normal’ elderly men as well as close degree relatives of patients with CLL [
49]. Elderly individuals with monoclonal non-specific gammopathy of undetermined significance (MGUS) have a 1% conversion/year to develop multiple myeloma or lymphoma [
50].
It is therefore particularly intriguing to comment on The Second Northwick Park Heart Study of Miller
et al. [
51], who prospectively studied hypercoagulability once a year for 4 years with follow-up of 11 years in a cohort of 3052 middle-aged men, clinically free of malignancy, for the development of myocardial infarction or malignancy. Hypercoagulability was defined as 2 yearly consecutive positive measurements of both increased prothrombin activation fragments 1+2 as well as fibrinopeptide A exceeding the upper quartiles of the population. This was found in 111 men. Although no greater incidence of myocardial infarction was noted, total mortality from cancer was higher in those with persistent activation than in the other group, 11.3 vs. 5.1%, respectively, with a relative risk of 1.79 (
P = 0.015). This was due largely to a higher mortality from gastrointestinal (GI) cancer, 6.3 vs. 1.9%, relative risk of 3.26,
P < 0.001. This was also associated with an earlier diagnosis of GI cancer as well as a more rapid course to death. This prospective study is remarkably powerful in supporting the role of thrombin in cancer development and/or progression.
The experimental animal data, as well as the intriguing observation of Shulman and Lindmarker [
45] and the recent report of Miller
et al. [
51], provide compelling evidence for the use of an appropriate antithrombin agent for the adjuvant treatment of patients with newly diagnosed cancer. The particular tumor or dosing regimen and duration of therapy remain to be determined, but it is likely that this approach would be efficacious in patients immediately after the diagnosis of cancer, rather than after the tumor burden becomes too great to be responsive.
Linear Mode
