The development of both idiopathic and situational VTE remains multifactorial [1]. Advanced age [2], malignancy [1] and smoking [3], in addition to both genetic and biochemical abnormalities within the coagulation and endovascular systems [4], predispose to VTE. Furthermore, women are perceived to be at greater risk for VTE [2,5], principally due to the presence of both endogenous and exogenous oestrogen sources [6,7,8,9].

Published data implicate both the guanine-to-adenine nucleotide 20210 (G20210A) prothrombin and guanine-to-adenine nucleotide 1691 (G1691A) factor V gene polymorphisms in the development of VTE during pregnancy and the puerperium [10,11]. Few studies have assessed these and other genetic risk factors, as well as hyperhomocysteinaemia, among nonpregnant women, while also considering such elements as smoking [3], malignancy [1], oestrogen use [6] and recent immobilization [12].

In order to better define the role of thrombophilic defects and other risk factors in the development of VTE in women, we conducted the present study. In addition to probable genetic risk factors [10,11] and total homocysteine (tHcy) levels, we examined the relationship between VTE and four polymorphisms involved in tHcy remethylation to methionine. The first two, the adenine-to-cytosine (A1298C) and thymine-to-cytosine (T1317C) [13], which are MTHFR polymorphisms, participate in the cycling of folate coenzyme by the MTHFR enzyme, in addition to the classic MTHFR cytosine-to-thymine (C677T) polymorphism [14]. We also studied the effect of the novel adenine-to-guanine polymorphism (A66G) in MTRR, the cobalamin-dependent enzyme that is responsible for folate-dependent tHcy remethylation [14].


We conducted a retrospective case-control study. Eligible cases included all women aged 16-79 years with objectively confirmed VTE seen at the University Health Network in Toronto from April 1996 to October 1999. Objective diagnostic testing comprised compression ultra-sonography or venography for deep vein thrombosis of the limb, ventilation/perfusion or contrast enhanced spiral computed tomography imaging for pulmonary embolism, and magnetic resonance or computed tomography imaging for intra-abdominal or intracranial venous thrombosis. We excluded those women whose VTE was related to a central venous catheter [15], as well as those with a history of coronary artery, cerebrovascular or peripheral vascular disease [16], or a previously identified thrombophilia defect. After obtaining written informed consent, one female control individual (within 10 years of age) was matched to every case; these control individuals were derived from a cohort of healthy women recruited through the Family Medicine Unit at the Toronto University Health Network. Thirty cases (approximately 25% of the study population) had been enrolled in a previous study that examined the relationship between hyperhomocysteinaemia and VTE [17], and none of the control individuals had participated in a previous study.

At the initial visit, a clinical nurse specialist interviewed cases using a standardized data collection form. An assessment for VTE risk factors included age at first VTE, history of recurrent VTE, history of malignancy within the past 5 years, current cigarette consumption, and immobilization or surgery within 3 months before the diagnosis of VTE. Presence of an 'oestrogen exposure state' was defined as either current use of an oral contraceptive agent, hormone replacement therapy or tamoxifen, or being pregnant or within 6 weeks postpartum. Information was also collected about the anatomical site of the VTE and the method of detection. Control group data were obtained using a standard questionnaire, administered by a trained research assistant during a face-to-face interview. The same demographic information was collected for cases and control individuals, with the exception of a history of recent immobilization or surgery, which was not assessed in the control group.

All laboratory analyses were performed with the investigator blinded to each woman's clinical diagnosis. Participants were instructed to fast for at least 8 h before plasma tHcy specimens were collected. Neither cases nor control individuals were known to be taking folate or vitamin B12 supplements at the time of blood specimen collection. Plasma tHcy was analyzed using high-performance liquid chromatography as previously described [18], and red-cell folate was measured using a radioimmunoassay method (Quantaphase II; Bio-Rad Laboratories Incorporated, Toronto, Ontario, Canada). DNA was extracted from peripheral blood cells [19]. A single multiplex polymerase chain reaction assay was designed to identify simultaneously the prothrombin, factor V and MTHFR genotypes [20,21,22]. The MTRR A66G genotype was analyzed as previously described [23]. Cases (but not control individuals) were evaluated for the presence of anticardiolipin antibodies, the lupus-like anticoagulant, and deficiencies in protein C, protein S and antithrombin.

Statistical analysis

In the primary analysis, the association between the presence of VTE and each thrombophilic risk factor was evaluated using conditional logistic regression analysis. Crude and adjusted odds ratios (ORs) were estimated. The ORs were adjusted for specific covariates, which were defined a priori, including current cigarette smoking habits, oestro-gen exposure state and recent history of cancer. Hyperhomocysteinaemia was defined as the tHcy concentration above the 95th centile value in the control group [24]. Interaction terms between the presence of hyperhomocysteinaemia and the homozygous state for any MTHFR or MTRR polymorphism were also evaluated [25].

A predefined subgroup analysis was performed excluding cases and their respective control individuals with either immobilization or surgery within the past 3 months, and any case-control pairs in which either had had a malignancy within the past 5 years. The potential presence of another identifiable thrombophilia defect among the cases (eg protein S, protein C or antithrombin deficiency) was not accounted for in any of the analyses.

Genotype distributions and allele frequencies and equilibria were analyzed using Genetic Data Analysis version 1.0 (PO Lewis and D Zaykin, 2000, Sinauer Associates, Incorporated, Sunderland, MA, USA). Probabilities for allele frequencies were calculated using Fisher's exact test. For each disequilibrium (Hardy-Weinberg or pairwise linkage), the data program estimated the probability of obtaining the observed genotype distribution using 3600 permuted simulations generated by an allele-shuffling algorithm.

Baseline characteristics of cases and control individuals were compared using one-way analysis of variance for continuous variables, or the Χ2 test for categorical data. All P values were two-sided, and the significance level chosen was 0.05. Statistical analyses were performed using SAS Version 6.12 (SAS Institute Incorporated, Cary, NC, USA). All data were collected in an anonymous manner, and approval to conduct the study was obtained through the University of Toronto Research Ethics Committee.


General findings

After initial review, 54 potential cases were excluded. Reasons for exclusion included the presence of cardiovascular disease (29 women), VTE related to central venous catheters (seven women), recurrent pregnancy loss, heparin-induced thrombocytopenia, superficial phlebitis or cellulitis (eight women), and lack of objective evidence to confirm the diagnosis of VTE (10 women). Thus, 129 cases, along with 129 age-matched control individuals, were included.

Table 1 presents the characteristics of the 129 cases with VTE and their respective control individuals. Of all case-control pairs, 120 (93%) were matched within 7 years of age, and 82 (64%) were within 5 years. The mean age of the cases and controls was 45.8 years and 40.9 years, respectively (P = 0.03). Cases were more likely to have a history of malignancy (24.0% versus 6.2%; P = 0.0001) and fewer were current smokers (13.2% versus 23.3%; P = 0.2).

Table 1 Characteristics of 129 women with VTE and age-matched control individuals

A nonstudy thrombophilia defect was identified in nine cases (7%). Four had positive anticardiolipin antibodies, and a fifth had a positive lupus anticoagulant. Of these five women, one was homozygous for both MTHFR C677T and A1298C, and heterozygous for the factor V polymorphism; one woman was homozygous for MTHFR A1298C; one woman was homozygous for MTRR A66G; and the remaining two women had no other identifiable defects. Three other women displayed mild functional protein S deficiency, of which one demonstrated homozy-gosity for both the MTHFR A1298C and MTRR A66G polymorphisms; one was homozygous for the MTHFR A1298C polymorphism; and the third was homozygous for the MTRR A66G polymorphism. One other woman was found to have antithrombin deficiency, but had no other detectable abnormalities.

Gene frequency

Of the six loci examined (Table 2), the population-dependency of allele frequencies was known for the prothrombin, factor V and MTHFR C677T genes. The frequencies of these variant alleles among the control individuals (1.2, 1.9 and 27.0%, respectively) were within the range of values reported for predominantly Caucasian populations. The genotype distributions at all six loci showed no departures from those expected under the Hardy-Weinberg equilibrium. Comparison of frequency distributions for each allele pair showed significant differences for the prothrombin, factor V and MTHFR C677T polymorphisms (Fisher's exact test P < 0.05). As expected, the linkage disequilibrium was significant for each pairwise comparison among the three MTHFR intragenic loci (data not shown), but not for any other pair of loci (Table 2).

Table 2 Allele frequencies of the prothrombin, factor V, MTHFR and MTRR polymorphisms among 129 women with VTE and 129 matched control individuals

Venous thromboembolism risk among all women

Among the participants, women with VTE were significantly more likely than control individuals to be carriers of the prothrombin polymorphism (9.3% versus 2.3%; adjusted OR 4.3, 95% confidence interval [CI] 1.1-17.4; Table 3). One case and no control individuals were homozygous for the prothrombin gene defect. Similarly, the factor V gene polymorphism was more common among cases than control individuals (16.3% versus 3.9%; adjusted OR 3.9, 95% CI 1.4-10.8). Two cases and no control individuals were homozygous for this defect. Three cases (2.3%) were carriers of both the prothrombin and factor V polymorphisms, for which no further statistical analysis was conducted.

Table 3 Prevalence and adjusted ORs for thrombophilia among women with VTE

The MTRR A66G homozygous state was more common among cases (32.6%) than control individuals (27.1%), but not significantly (adjusted OR 1.1, 95% CI 0.6-2.2). Homozygosity for C677T MTHFR was no more frequent among cases (14.7%) than control individuals (10.1%; adjusted OR 1.4, 95% CI 0.6-3.1), and neither was the homozygous state for A1298C MTHFR (52.7% versus 53.5%; adjusted OR 1.0, 95% CI 0.6-1.7; Table 3). Only one case and one control individual (0.8% each) carried the T1317C MTHFR homozygous variant (adjusted OR 1.0, 95% CI 0.1-16.3). The presence of any MTRR or MTHFR homozygous state was also not a significant risk factor for VTE (adjusted OR 1.1, 95% CI 0.6-2.0).

The mean fasting plasma tHcy was significantly higher among women with VTE (11.1 μmol/l) than control individuals (6.5 μmol/l; unpaired t-test P < 0.0001; Table 1). The 95th centile tHcy value for control individuals was 11.0 μmol/l; using this latter cut-point, hyperhomocysteinaemia was significantly more common among cases (35.6%) than control individuals (4.0%). The adjusted OR was 17.8 (95% CI 4.2-74.9) for VTE in the presence of hyperhomocysteinaemia. No significant interaction (P > 0.50 for each) was observed between the presence of hyperhomocysteinaemia and any MTHFR or MTRR homozygous state.

Venous thromboembolism risk, excluding women with recent malignancy, immobilization or surgery

Upon excluding cases with secondary causes for VTE (malignancy, immobilization or surgery), and their respective control individuals, 81 case-control pairs were available for analysis (Table 3). In the presence of the prothrombin polymorphism, the adjusted OR for VTE was 7.8 (95% CI 1.0-64.2). The risk for VTE was also significant in the presence of the factor V polymorphism (adjusted OR 6.6, 95% CI 1.2-11.4). The risk for VTE among homozygotes for each MTHFR and A66G MTRR polymorphism was no higher in this subgroup, however (P > 0.5 for each). Finally, the adjusted OR for VTE in the presence of fasting hyperhomocysteinaemia decreased somewhat from that observed in the main analysis (adjusted OR 11.6, 95% CI 2.7-50.5; Table 3).


We found that the presence of the prothrombin and factor V gene polymorphisms and hyperhomocysteinaemia significantly increased the risk for VTE among women. When those with recent malignancy, surgery or immobilization were excluded, the risk was even greater for carriers of the prothrombin polymorphism. Although hyperhomocysteinaemia also increased the risk for VTE, this association was not found for the MTRR A66G or any of the three MTHFR gene polymorphisms.

Of the cases studied, 7% were found to have another thrombophilia defect, such as antiphospholipid antibodies or protein S deficiency, which was not controlled for in the present analysis. It is unlikely that the presence of such defects influenced the results, however, because they were uncommon, and eight out of nine of the defects were observed among women with the MTHFR or MTRR gene polymorphisms, or both, none of which independently conferred an increased risk for VTE.

The present study might have been limited by a referral bias among the cases, potentially exaggerating our risk estimates for VTE [26]. Because 25% of cases also had recurrent VTE, these estimates might have been further inflated. In order to evaluate this, we compared the rates of thrombophilia gene defects in the cases and control individuals with those in other published case-control studies (Table 4). The prevalence of the prothrombin gene among the control population studied here (2.3%) was within the general limits of other studies (range 1.0-5.4%), and our case rate (9.3%) was similar to that of other studies (range 4.6-31.0%), albeit slightly lower than those of some studies. Similarities in the frequencies of factor V and C677T MTHFR gene polymorphisms were also observed. Thus, we believe that our case and control rates reflect those observed in other adult populations with and without VTE. Ethnic variation, not assessed herein, might explain some of the differences between studies in the prevalence of these polymorphisms, given the wide diversity of ethnicity according to geographic area [27].

Table 4 Published case-control studies of the risk for VTE in the presence of polymorphisms of the prothrombin, factor V and MTHFR C677T genes, as well as fasting hyperhomocysteinaemia

Although the rate of fasting hyperhomocysteinaemia found among our control individuals (4.0%) was consistent with those of other studies (range 4.7-5.0%; Table 4), our case rate (35.6%) was more than twice that found by other investigators. There are several possible reasons for this. First, because plasma tHcy is both inversely related to renal function [28] and directly related to age [29], the higher tHcy levels among the cases might be partly explained by their relatively high serum creatinine concentrations and older age. Second, after omitting women with recent cancer from our subgroup analysis, the adjusted OR dropped from 17.8 to 11.6, suggesting that malignancy may play a role in tHcy metabolism, as proposed by others [30]. Third, by excluding women with a history of cardiovascular disease from the present study, we probably eliminated another potential confounder (cardiovascular disease has been association with hyperhomocysteinaemia) [16]. Finally, although we did not collect data on serum vitamin B12, it has been demonstrated that both measured vitamin B12 [31] and its supplementation [32] probably have little influence on plasma tHcy concentration relative to that of folate, which appeared to be higher among the cases than the control individuals in whom it was measured.

A meta-analysis of 23 observational studies [33] failed to demonstrate an association between MTHFR 677T homozygosity and arterial disease (pooled OR 1.1, 95% CI 0.9-1.4). Similarly, our data and those of others (Table 4) suggest that the MTHFR polymorphisms do not heighten the risk for VTE. Furthermore, we found no significant interaction between MTHFR homozygosity and hyperhomocysteinaemia. Thus, we recommend that clinical testing for currently identified polymorphisms in the MTHFR gene be postponed in the evaluation of VTE risk outside of the research setting. A further understanding of the interaction between the combination of hyperhomocysteinaemia and other gene polymorphisms, such as factor V [34] and prothrombin, is needed.

If hyperhomocysteinaemia is a true causative risk factor for a first [24] or subsequent [35] VTE event, then the mechanisms responsible for tHcy elevation and, hence, venous thrombosis require further elucidation. Newly isolated polymorphisms, such as MTRR A66G, may contribute to the development of VTE, but we found no such association in the group studied here. In the meantime, we expect that clinical trials of tHcy reduction [36], in addition to future work in the field of molecular and genetic epidemiology, are the next logical steps.

Others have evaluated the influence of nongenetic factors on the risk for VTE. For example, in a post-trial analysis by the Heart Estrogen Replacement Study (HERS) investigators [8], the risk for VTE was increased among women with lower-extremity fractures (adjusted relative hazard [RH] 18.1, 95% CI 5.4-60.4] or cancer (RH 3.9, 95% CI 1.6-9.4), or who had had inpatient surgery within the preceding 90 days (RH 4.9, 95% CI 2.4-9.8). Even limited periods of immobilization, such as a short illness or minor surgery or injury, appear to increase the risk for deep vein thrombosis (OR2.9, 95% CI 1.5-5.4) [37]. We therefore conducted a subgroup analysis that excluded women with malignancy, recent surgery, or immobilization. Although we might expect that certain thrombophilia markers confer the greatest relative risk for VTE in the absence of other risk factors [4], as was the case for the prothrombin polymorphism, we cannot explain why this was not so for factor V. Future research should further explore the interplay between hereditary and acquired VTE risk factors.

Clinicians face a dilemma in deciding which women should be screened for these and other thrombophilia markers. It would seem both imprudent and costly to investigate those who develop VTE in the presence of active malignancy, because anticoagulant therapy is likely to be continued indefinitely [38]. However, among women who experience VTE either in the absence of any risk factors, while taking hormone replacement therapy [8,39] or oral contraceptives [6,7], or during pregnancy and the puerperium [10,11], a thrombophilia assessment may be a sensible option. Specifically, others have demonstrated that oestrogen use and activated protein C resistance have a synergistic effect on the risk for VTE (OR 13.3, 95% CI 4.3-41.0) [39]. It is least clear whether screening is beneficial in an otherwise healthy woman who develops VTE after surgery or a fracture of the lower extremity [5,40]. Hopefully these issues can be resolved through ongoing research and expert consensus.