Patients with malignancies have an increased risk for thromboembolic events due to the release of tissue factor by the tumor, damage to the vessel wall, and immobilization. Moreover, tumors may improve their growth and metastatic spread by utilizing the coagulation system. To date, no information is available on the additional role of prothrombotic mutations in these patients.
The prevalence of the factor V Leiden mutation (FVL) and the prothrombin G20210A mutation and of homozygosity for the methylenetetrahydrofolate reductase (MTHFR) C677T substitution has been analyzed in a cohort of 175 patients with gastrointestinal adenocarcinoma by the polymerase chain reaction-restriction fragment length polymorphism technique.
6.9% of the patients were heterozygous for FVL, 5.7% were heterozygous for the prothrombin mutation, and 9.7% were homozygous for the MTHFR C677T mutation was detected in 9.7% of patients. Compared with the normal population, we found an increased prevalence of the prothrombin G20210A substitution (5.7% vs. 0.8%, P = 0.028). Thromboses were absent in 147 patients (Group A), whereas 28 of the patients suffered from thromboses during the period following tumor diagnosis (Group B). In Group A, 6.8% of the patients and 21.4% of the patients in Group B had a thrombosis before the diagnosis of cancer (P = 0.025, odds ratio [OR] 3.7). Heterozygous FVL was present in 4.8% of the patients in Group A and in 17.9% of the patients in Group B (P = 0.026, OR 4.4). In patients with thromboses before the detection of the tumor, the risk was elevated 6.3-fold (25.0% vs 5.0%, P = 0.015). Heterozygosity for the prothrombin mutation and homozygosity for the MTHFR C677T substitution did not increase the incidence of thromboses.
Since the publication about 130 years ago of Trousseau's famous observation linking the coincidence of tumor and thromboses, the close correlation between malignant disease and thromboembolism (TE) has been well documented. The pathomechanism of thrombophilia and thrombogenesis in patients with malignancies, however, is less obvious.1 Tumor cells generate a prothrombotic state by inducing fibrin formation and activating platelets. Direct damage to the vessel wall by the invading tumor may play an additional role. The intrinsic thrombophilia of tumor patients may be aggravated by diagnostic and therapeutic procedures and by immobilization, leading to the common phenomenon of tumor-associated thromboembolism. However, thrombophilia is not only a side effect of malignant diseases. It may also be related closely to cancer pathogenesis, as tumor cells can improve their own survival and metastatic potential by expressing tissue factor and cancer procoagulant in vitro.2 Blocking this pathomechanism in cancer patients might be beneficial and treatment with anticoagulants as warfarin or fractionated heparin may improve survival of patients with malignant diseases.3, 4
Several genetic risk factors for thrombophilia have been described and they occur in about 40% of patients with recurrent venous thrombosis. The most common genetic defect within the Caucasian population is the factor V Leiden (FVL) mutation, an arginine to glutamine substitution at amino acid position 506 of coagulation factor V, which interferes with the cleavage and inactivation of factor V (FV), thus leading to a reduced clearance of factor Va. Heterozygosity for the FVL mutation is found in 5–10% of the Caucasian population and increases the risk of thrombosis about sevenfold.5 Another common thrombophilic mutation is the G→A substitution at nucleotide position 20210 in the 3′-untranslated region of the prothrombin gene. This point mutation increases the risk of thrombosis about twofold by elevating prothrombin levels in plasma and has also been associated with myocardial infarction and cerebral vein thrombosis.6, 7 Conversely, homozygosity for the cytosine to thymine substitution at position 677 in the 5,10-methylenetetrahydrofolate reductase (MTHFR) cDNA results in a thermolabile variant of the enzyme (TL-MTHFR) with reduced catalytic activity and in elevated plasma levels of homocysteine,8 which increase the risk of venous and arterial thromboembolism in some case–control studies.9, 10
We tested whether the inherited thrombotic risk factors FVL, prothrombin G20210A, and MTHFR C677T play a role in thrombosis pathogenesis in a population of patients with gastrointestinal carcinoma, thereby causing a higher incidence of tumor-associated thromboembolisms and a poorer disease outcome.
MATERIALS AND METHODS
Study Subjects and Tumor Data
We studied all patients with an adenocarcinoma of the stomach, colon, or rectum, who had been admitted between January 1, 1999 and January 1, 2000 to two cancer rehabilitation hospitals associated with the tumor center of the University of Munich. A total of 175 patients were included in the study. All patients were Caucasians. After giving written informed consent, the patients were questioned carefully with regard to a history of thrombosis and the presence of risk factors for thromboembolism (family history, age, smoking, adiposity, central line, varicosis, estrogens, chronic inflammation, and immobilization). Their complete oncologic history was registered.
The flowchart of the study is shown in Figure 1. If patients did not suffer from thromboses in the setting of malignancy until December 1, 2000, they were included in Group A (no TE in the setting of malignancy), otherwise they were included in Group B (TE in the setting of malignancy). In December 2000, all family doctors were questioned as to the further course of the disease.
One-hundred forty-seven patients had no evidence of thrombosis during their tumor history and were included in Group A. Twenty-eight patients with malignant tumors who suffered from thromboses in the setting of malignancywere included in Group B. A thrombotic event was defined as thrombosis in the setting of malignancy, if it occurred within 1 year before diagnosis of the carcinoma in patients with active malignancy. A thrombosis was not tumor associated if it occurred more than 1 year before discovery of malignant disease.
The demographic characteristics of the patients in Groups A and group B are shown in Table 1. The groups did not differ significantly with regard to gender, age at tumor diagnosis, and observation time. As shown in Table 2, most patients had adenocarcinoma of the colon. The two study groups were also not significantly different regarding localization of the tumor, International Union Against Cancer stage, histologic grading (Group A: 4.6% Grade 1, 50.4% Grade 2, 45.1% Grade 3; Group B: 46.2% Grade 1, 53.8% Grade 2, 0% Grade 3), and radicality of the resection at the time of tumor diagnosis.
Table 1. Epidemiologic Data of the Study Populationa
Group A (N = 147)
Group B (N = 28)
ns: not significant.
All numbers are median values. Number of patients is shown in parentheses.
Age at tumor diagnosis (yrs)
Dead until December 1, 2000
Cause of death
Observation time (weeks)
Table 2. Localization of Tumor, Stage, and Metastasisa
ns: not significant; UICC: International Union Against Cancer.
All numbers are median values.
At December 1, 2000
The therapeutic procedures are shown in Table 3. Almost all patients had a tumor resection within a median of 2 weeks after tumor diagnosis. Within a median of 11 (Group B: 13) weeks after tumor diagnosis, about three fourths of the patients underwent chemotherapy and about one sixth of the patients underwent irradiation.
Table 3. Therapeutic Procedures and Local Recurrencea
ns: not significant.
All statements except for the last line refer to the timepoint of cancer diagnosis.
Any therapeutic procedure
The FVL, prothrombin G20210A and MTHFR C677T genotype analyses were performed in all patients. Genomic DNA was extracted from white blood cells using the QIAmp DNA blood mini kit (Qiagen, Hilden, Germany) and amplified by the polymerase chain reaction (PCR) with gene-specific primer pairs. Each 50-μL reaction contained 10 mM Tris-HCl, pH 8.3, 50 mM KCl, 1.5 mM MgCl2, 0.01% gelatin, 200 μM of each dNTP, 20 μM of each forward and reverse primer, approximately 200 ng of high molecular weight DNA, and 1.25 U Taq DNA Polymerase (Sigma-Aldrich, St. Louis, MO). After denaturation at 95 °C for 3 minutes, DNA was amplified for 40 cycles at 95 °C for 30 seconds, 60 °C (prothrombin and FVL) or 65°C (MTHFR) for 30 seconds, and 72 °C for 30 seconds. The positive assay control was DNA from a normal (FVL) or homozygous mutant subject (prothrombin, MTHFR), whereas a water blank served as a negative control for each run.
The newly generated products were then digested with the appropriate restriction enzyme (New England BioLabs, Beverly, MA) and electrophoresed in 2–3% low melting point agarose gels (Gibco, Grand Island, NY). Mnl I digestion of the 288-base pairs (bp) PCR product of the FV gene generated fragments of 158, 93, and 37 bp for the normal allele. Digestion products of the mutant allele were 158 and 130 bp in size. The prothrombin fragment of 230 bp was cleaved by Hind III in case the G→A mutation was present and yielded two smaller fragments of 190 and 40 bp. The 219-bp PCR product of the normal MTHFR allele was not digested by Hinf I, whereas digestion of the mutant allele generated two fragments, 176 and 43 in length.
Results of the two groups were compared with the Mann–Whitney U test, the Fisher exact test, and Pearson's chi-square test for categorial variables. Relative risk (RR) and its 95% confidence intervals were calculated. Survival test was done by log rank analysis. All statistical analyses were performed with the Statistical Package for the Social Sciences (SPSS for Windows 10.0, SPSS, Chicago, IL).
Survival and Metastases Data
At the end of the observation period (December 1, 2000), 65.3% of the patients in Group A and 67.9% of the patients in Group B had died (Table 1). At diagnosis, 55.1% of the patients in Group A and 35.7% of the patients in Group B had metastases. Until December 1, 2000, these numbers had increased to 87.8% in Group A and 82.1% in Group B. The local recurrence rates in Groups A and B were 22.4% and 32.1%, respectively, without significant differences between the two groups (Table 3).
The risk factors for thromboembolism are shown in Table 4. They did not differ significantly between Groups A and B. In Group B, 44.4% of the patients were immobilized at the time of the tumor-associated thrombosis. All patients had been receiving thrombosis prophylaxis with unfractionated or low molecular weight heparin only during the short postoperative stay (median 2 weeks).
Table 4. Risk Factors for Thrombosis
BMI: body mass index; ne: not evaluable; ns: not significant.
Risk factors for thrombosis
Age (> 60 yrs)
Adiposity (BMI > 26)
Immobilization at time of thrombosis
Thromboembolism independent of malignancy
In Group A, 6.8% of the patients and 21.4% of the patients in Group B had a thrombosis 2–56 years before diagnosis of the malignant disease (Table 5). The difference is statistically significant (P = 0.025) with a RR of 3.7 (1.2–11.3).
Table 5. Thromboembolic Events (TE) before Diagnosis of Malignancy
We detected 36 tumor-associated thrombotic events (Table 6), occurring between 8 weeks before the diagnosis of malignancy and 482 weeks thereafter (median 22 weeks). Two events occurred 6 and 2 weeks, respectively, before the detection of the tumor and led to a cancer diagnosis. The most frequent thromboembolic events were pulmonary embolisms, followed by thromboses in the thigh. Two Group B patients died of pulmonary embolisms and both had a tumor-associated thrombosis several months before the fatal event.
Table 6. Thromboembolic Events (TE) in the Setting of Malignancy
The highest rate of tumor-associated thromboses was observed during the first 4 months after diagnosis of the malignant disease (Fig. 2) and 78.6% of the thrombotic events occurred during the first year of malignancy. The association with tumor diagnosis, surgery, chemotherapy, and death is shown in Figure 3. Only 8.3% of the patients had a thromboembolic event within 2 weeks after surgical intervention (surgery related). 13.8% of the thromboses were registered in the first 2 weeks following cancer diagnosis, even before surgery (diagnosis related), and 25.0% occurred during chemotherapy (chemotherapy related). The chemotherapy-related events included one arm vein thrombosis (patient with indwelling central line), three pulmonary embolisms, and five thromboses of the leg/pelvis. 13.8% of thromboembolic events occurred during the week before death (death related), two of those even caused death due to pulmonary embolism. 38.8% of the tumor-associated thrombotic events could not be related to any clinical data (no association). 44.4% of patients were immobilized at the time of thrombosis.
We did not detect any patient homozygous for the FVL or prothrombin mutation (Table 7). In the entire study population, we found heterozygosity for FVL in 6.9% of the cancer patients and for the prothrombin G20210A substitution in 5.7% and homozygosity for the MTHFR C677T mutation in 9.7%. Compared with the prevalence in normal populations reported by Poort et al.6 (P = 0.020), Rosendaal11 (P = 0.001), and Pihusch et al.12 (P = 0.028), the prevalence of the prothrombin mutation was significantly increased in the cancer population.
Table 7. Prevalence of the Heterozygous Factor V Leiden (FVL) Mutation, the Heterozygous Prothrombin G20210A Mutation, and the Homozygous MTHFR C677T Mutation
MTHFR: methylenetetrahydrofolate reductase; ns: not significant.
The prevalence of the FVL mutation in patients with a thrombosis before tumor disease was 25% compared with 5.0% in patients without a thrombosis before tumor disease. The difference is statistically significant (P = 0,0015) with an RR of 6.3. We found similar results in patients with thromboembolism in the setting of malignancy (17.9% vs. 4.8%, P = 0.026, RR 4.4). Only three patients (1.7%) in Group B with a negative thrombosis history were FVL heterozygous, whereas 19 were homozygous for the wild-type allele.
The results regarding the prothrombin mutation were not significant because of the much lower prevalence of this form of genetic thrombophilia, but they show a trend toward a 2.4-fold increased risk of thrombosis in heterozygous carriers. One of the two patients suffering from a fatal pulmonary embolism was a double heterozygote for the FVL and prothrombin mutations. We did not detect an influence of the FVL and prothrombin mutations as well as of homozygosity for the MTHFR C677T mutation or of a positive thrombosis history on the survival of patients or on the time lag between diagnosis and metastasis of the tumor.
In this case–control study, heterozygous FVL increased the risk of thrombosis in the setting of malignancy by 4.4-fold, which underscores the importance of FVL in the pathogenesis of thromboembolism in cancer patients. Although not significant, the RR of a tumor-associated thrombosis was lower than the risk of a thrombosis not linked to malignancy, which was increased 6.3-fold in our study and is in accordance with published data.5, 10 This remarkable finding may indicate that, in tumor-associated thrombosis, the risk induced by FVL is partially overcome by nongenetic influences. The same might be true for the prothrombin G20210A substitution, which failed to reach statistical significance in our study.
The role of nongenetic factors in tumor-associated thrombosis is stressed by the striking clustering of thromboembolic events that occur about the same time as the tumor diagnosis. Because many of these thromboses occurred before any therapeutic procedure (surgery, chemotherapy, radiation), the increase in invasive procedures during the first weeks following diagnosis may play a pivotal role as an additional risk factor. We conclude that patients with a high risk of tumor-associated thrombosis should receive thromboprophylaxis beginning on the day of tumor diagnosis.
At least in our elderly patients, it was possible to identify those with a high risk for tumor-associated thrombosis by taking a careful thrombosis history at the time of diagnosis of malignant disease. Patients with a positive thrombosis history had a 3.7-fold increased risk of developing a second thrombosis after tumor diagnosis. Additional routine screening for FVL in all patients with newly diagnosed cancer might not be necessary, as only 1.7% of the tumor-associated thromboembolic events were prevented by genetic analyses in addition to a careful thrombosis history. Immobilization may be another important cofactor in the pathogenesis of tumor-associated thrombosis, as 44% of the patients were immobilized at the time of the thrombotic event. In addition, patients who already had a tumor-associated thrombosis may have a high risk for fatal pulmonary embolism. Both of our patients with fatal pulmonary embolisms had a tumor-associated thrombosis several months before the final event. We conclude that thromboembolism before the diagnosis of cancer predicts the risk of thrombosis after the diagnosis of malignancy and should be the decisive factor in determining prophylactic measures. In addition, immobilization and a history of tumor-associated thrombosis should be taken into consideration.
In our study, most therapy-associated thromboembolic events occurred in close correlation with intravenous chemotherapy. This is in accordance with several trials in which a thrombophilic influence of intravenous chemotherapy was demonstrated,13, 14 possibly due to the stimulation of the plasma coagulation system, a decrease in coagulation inhibitors,15 and/or the activation of the endothelium and platelets.16 Indwelling central venous catheters pose an additional risk factor for axillary/subclavian vein thrombosis in cancer patients.17 In our study, one patient with a central line suffered from axillary vein thrombosis during application of chemotherapy. Due to the high thrombosis risk of cancer patients during chemotherapy, it might be reasonable to implement prophylactic anticoagulation, thus reducing the thrombosis risk. This approach is similar to that used for patients undergoing surgery, for whom a small percentage of events occurs following routinely administered heparin prophylaxis.
As in arterial and venous thrombogenesis in healthy populations,18 the homozygous MTHFR C677T mutation did not increase the risk of tumor-associated thrombosis. In contrast to several studies, which showed a reduced incidence of the T allele in patients with colorectal carcinoma,19, 20 the prevalence of this nucleotide substitution was not reduced in our cancer population. This is in accordance with another large study that was unable to find any correlation between colorectal carcinoma and the MTHFR C677T mutation.21 An effect of this particular MTHFR mutation on gastrointestinal carcinoma pathogenesis is unlikely.
Analysis of our complete patient cohort showed a significantly increased prevalence of the prothrombin mutation when compared with normal controls. It might be that a different ethnic background influences the results. For example, southern European populations have a higher prevalence than northern European populations.10 Most published epidemiologic data are from The Netherlands. However, Pihusch et al.12 reported a prevalence in a healthy population from our geographical area that was similar to that in the Netherlands. Therefore, an ethnic effect is highly unlikely. In addition, a bias in the selection of patients can be excluded, as the increased prevalence is selectively confined to the prothrombin mutation, whereas the incidence of FVL with a much higher thrombosis risk is not increased. Our data suggest that the prothrombin mutation might be a possible risk factor for gastrointestinal carcinoma.
To our knowledge, no information on the prevalence of thrombophilic mutations in cancer patients is available. Multiple interactions between malignancy and coagulation have been reported and thrombophilia may not only be a side effect of malignant diseases, but may be related closely to cancer pathogenesis. As the prothrombin mutation causes only a moderate thrombophilic state, it is currently unknown how the resulting increased prothrombin levels could support cancer pathogenesis. In contrast to FVL, however, the prothrombin mutation has been reported to be associated with coagulation disturbances in both the venous6 and arterial system,7 which may be important for tumor development. In addition, the increased prothrombin levels might not only affect plasma coagulation. Besides generating fibrin, thrombin also activates platelets, smooth muscle cells, fibroblasts, mesangial cells, and macrophages,22 all of which are present within tumor tissues. Thrombin therefore induces a variety of cellular responses such as proliferation and chemotaxis. Increased prothrombin levels might affect tumor growth by influencing pivotal mechanisms such as cell adhesion, cell proliferation, and vasculogenesis.23 Because we did not detect any influence of the prothrombin mutation on the metastasis and survival of cancer patients, this mutation may play a role at a very early stage of cancer pathogenesis. It may be rewarding to study these aspects prospectively in a patients carrying the heterozygous prothrombin G20210A mutation.