Homocysteine-Lowering Therapy and Risk for Venous Thromboembolism

Methods

Design

The design of HOPE-2, a large randomized, placebo-controlled clinical trial, is described elsewhere (7, 8). Soon after HOPE-2 began, a decision was made to include venous thromboembolism as a study outcome. A diagnosis of venous thromboembolism was based on prespecified, accepted criteria (7).

An independent data and safety monitoring board evaluated the safety of the participants and the overall quality and scientific integrity of HOPE-2. The research ethics review board of each participating center approved the trial, and all participants provided written informed consent.

Participants

The HOPE-2 included 5522 participants 55 years of age or older who had a history of coronary, cerebrovascular, or peripheral vascular disease; diabetes mellitus; and at least 1 additional risk factor for cardiovascular disease, regardless of baseline homocysteine level (7, 8). Persons taking daily vitamin supplementation that contained more than 0.2 mg of folic acid were excluded. A history of venous thromboembolism or the presence or absence of risk factors for venous thromboembolism did not affect eligibility. A complete list of inclusion and exclusion criteria appears elsewhere (7, 8). Data on all persons who were enrolled in HOPE-2 are included in the current report.

Centers

Individuals were recruited from 145 centers in 13 countries, including Canada (n = 3568), the United States (n = 414), Brazil (n = 265), western European countries (n = 426), and Slovakia (n = 849).

Intervention and Randomization

Between January 2000 and December 2000, participants were randomly assigned to receive a once-daily supplement containing 2.5 mg of folic acid, 50 mg of vitamin B₆, and 1 mg of vitamin B₁₂, or matching placebo. Randomization was computer generated, with a block size of 4; was stratified by clinical center; and was performed by having clinical centers call an automated centralized system. Information about block size and whether it was random or fixed was kept confidential for all study investigators. The randomization sequence was concealed, and all study personnel and study participants were masked to treatment allocation. The vitamin and placebo pill formulations were indistinguishable by size, color, weight, taste, or dissolution in water. Changes in blood levels of folate, vitamins B₆ and B₁₂, and homocysteine, which are affected by the study intervention, are not commonly measured in clinical practice, and the results of any such measurements that were performed as part of the study were kept confidential. No request was made to unmask treatment allocation for a participant.

Baseline Measurements and Follow-up

Baseline demographic data; medical history; and medication use, including current anticoagulant therapy, were recorded for all participants at study entry. History of venous thromboembolism was not documented. Baseline homocysteine levels were obtained in 3306 randomly selected participants (60% of total) who had fasted overnight. Stratified random sampling was used to achieve proportional representation of a subset of participants in countries with folic acid food fortification (Canada and the United States) and countries without this standard (all other countries with participating centers). Homocysteine was measured by using a fluorescence polarization immunoassay (Abbott IMx, Abbott Laboratories, Abbott Park, Illinois). The distribution of homocysteine was statistically significantly skewed; thus, these measures were log-transformed and inverse transformations were used to generate geometric mean values.

The first evaluation for venous thromboembolism occurred 18 months after randomization; at this visit, all participants were assessed for any venous thromboembolism event arising between trial entry and the 18-month visit. Thereafter, venous thromboembolism was assessed routinely every 6 months, to an average follow-up of 5 years. The trial used simple case report forms, which were faxed toll-free to the study coordinating office and were entered into a database by using optical character recognition (DataFax, Clinical DataFax Systems, Hamilton, Ontario, Canada). The database was fit for quality control assessments and statistical analyses. At each 6-month interval, participants were assessed in the study clinics. These assessments were directed primarily to ascertain study end points. Side effects were also evaluated, and adherence to treatment was assessed by interview and pill count. When in-person visits were not possible, participants were contacted by telephone.

Outcomes

The primary outcome in our study was symptomatic venous thromboembolism, which included deep venous thrombosis or pulmonary embolism (or both). In the original HOPE-2 report (8), venous thromboembolism was included under “other outcomes.” Diagnosis of deep venous thrombosis required confirmation with duplex leg ultrasonography or venography. Diagnosis of pulmonary embolism required confirmation with ventilation–perfusion lung scanning, computed tomographic pulmonary angiography, or conventional pulmonary angiography. When diagnostic testing had indeterminate results or was not done, which rarely occurred, we required oral anticoagulant therapy to be initiated at the same time that new-onset venous thromboembolism was recorded on the case report form. A maximum of 1 episode of venous thromboembolism per participant was counted during follow-up. We subcategorized episodes as unprovoked venous thromboembolism if they occurred in participants who did not have cancer at baseline and occurred 90 days or more after a lower limb fracture or 30 days or more after a hospitalization. We ascertained all events with concealment to randomization for study participants and assessors.

Statistical Analysis

Our primary analysis was a comparison of the incidence of venous thromboembolism in the 2 study groups. We prespecified secondary analyses and included comparisons between the groups of rates of venous thromboembolism (including unprovoked events) according to subgroups and strata (Figure 1).

View larger version:

• In this window
• In a new window

Figure 1. *Limited to the 3306 participants for whom homocysteine was measured at baseline. †Homocysteine level >13.8 µmol/L. Risk for venous thromboembolism (VTE) in prespecified subgroups.

We used an intention-to-treat analysis to compare the effect of homocysteine-lowering therapy with that of placebo on the subsequent development of venous thromboembolism. We conducted the time-to-event analysis by using a Cox proportional hazards regression model and expressed unadjusted risk as hazard ratios and 95% CIs. We examined the proportional hazards assumption by fitting the models with the interaction terms between time and treatment. We estimated a survival curve according to the Kaplan–Meier procedure and compared treatment groups by using a log-rank test. At each interval clinic visit, follow-up was greater than 99%. In the rare circumstance that an individual could not be assessed at a clinic visit and could not be contacted by telephone, we considered the individual to be free of venous thromboembolism at that point. Individuals who were lost to follow-up were censored at the time of last contact.

The original HOPE-2 was designed to recruit 5000 participants, with a mean of 5 years of follow-up, to detect a relative risk reduction of 17% to 20% and a statistical power of 80% and 90% in the primary composite outcome of cardiovascular death, myocardial infarction, and stroke, given an annual event rate of 4% in the placebo group and a 2-sided P value of 0.05. We did not estimate a formal sample size.

A 2-sided P value less than 0.05 was considered significant for all analyses, which we performed by using SAS, version 9.1 (SAS Institute, Inc., Cary, North Carolina).

Role of the Funding Sources

The study was funded by the Canadian Institutes of Health Research and Jamieson Laboratories. The funding sources had no role in the design, conduct, or reporting of the study or in the decision to submit the manuscript for publication.

Results

Study Participants and Homocysteine Levels

Of the 5522 study participants, 2758 were randomly assigned to receive homocysteine-lowering therapy and 2764 were assigned to receive placebo (Table 1). A total of 3982 participants (72%) were from Canada and the United States, where universal food fortification with folic acid was in place before the start of the trial. Adherence to therapy was similar between the treatment and placebo groups at 1 year (95% vs. 96%), 2 years (94% vs. 93%), 3 years (92% vs. 92%), 4 years (91% vs. 90%), and 5 years (91% vs. 88%). No serious adverse events related to study treatment occurred. The most common reasons for temporary or permanent discontinuation of treatment or placebo use were the participant's decision (11.1% vs. 12.6%), physician advice (1.6% vs. 2.0%), hospitalization (1.0% vs. 0.8%), and general malaise (1.0% vs. 0.7%). Twenty-one participants in the treatment group and 16 in the placebo group were lost to follow-up or withdrew from the study. They had been enrolled for at least 2 years and were included in the final analysis and censored for the duration of observation.

Baseline characteristics were well balanced in the 2 groups (Table 1). The mean participant age was 69 years. Twenty-eight percent of participants were women, 4% had a recent history of cancer, 80% were receiving an antiplatelet agent, and 8% were receiving an oral anticoagulant. Seventeen percent of women were receiving estrogen replacement therapy (Table 1).

The geometric mean homocysteine level at baseline was 11.5 µmol/L in both groups (n = 3306) (Table 1). In the treatment group, the mean homocysteine level was 2.1 μmol/L lower in participants in countries with folic acid food fortification than in those in countries without this fortification. In the placebo group, this difference at baseline was 2.3 μmol/L. At the end of the study, the mean homocysteine level was 9.3 µmol/L in the treatment group (n = 533) (a decrease of 2.2 µmol/L) and 12.3 µmol/L in the placebo group (n = 531) (an increase of 0.80 µmol/L). When we restricted this analysis to the 533 participants in the treatment group and the 531 participants in the placebo group for whom homocysteine was measured at baseline and at the end of the study, the net change was −2.0 μmol/L and 0.80 μmol/L, respectively. Among treatment recipients, the mean homocysteine level decreased by 1.9 µmol/L in those in countries with folic acid food fortification and 4.8 µmol/L in those in countries without fortification. Among placebo recipients, the mean homocysteine level increased by 1.0 µmol/L in those in countries with food fortification and by 0.66 µmol/L in those in countries without fortification.

Venous Thromboembolism

Of the 91 episodes of venous thromboembolism, 3 were not included as events because the method of diagnosis was not indicated and initiation of anticoagulant therapy was not documented. Thus, 88 episodes of venous thromboembolism were included, of which about two thirds were deep venous thrombosis and 47% were unprovoked (Table 2). For 3 (3.4%) of the 88 episodes of venous thromboembolism, the clinical center did not indicate the method of diagnosis but documented initiation of anticoagulation. Seventeen events (19.3%) were recorded in the first 18 months after randomization, and 71 were recorded thereafter.

Forty-four episodes of venous thromboembolism occurred in each group, corresponding to an incidence rate of 0.35 per 100 person-years in each group (hazard ratio, 1.01 [CI, 0.66 to 1.53]; P = 0.97) (Table 2 and Figure 2).

We did not observe a benefit of homocysteine-lowering therapy in any subgroup, including participants receiving an oral anticoagulant at baseline or those in whom outcome events were deep venous thrombosis alone, pulmonary embolism, or provoked or unprovoked venous thromboembolism (Table 2 and Figure 1).

View larger version:

• In this window
• In a new window

Figure 2. Cumulative proportion of participants with venous thromboembolism.

The characteristics of participants who were randomly selected to have plasma homocysteine levels assessed at baseline were fairly similar to those of participants who did not have levels assessed, with some notable exceptions (Appendix Table). Specifically, fewer participants who underwent homocysteine sampling at baseline were women (25% vs. 33%), and more were from North America (82% vs. 57%) and were taking a lipid-lowering drug (67% vs. 50%).

Among the 821 participants whose baseline homocysteine level was in the highest quartile (>13.8 µmol/L), vitamin therapy did not reduce the risk for venous thromboembolism (hazard ratio, 1.71 [CI, 0.48 to 6.06]) (Figure 1). For the 2216 individuals who did not undergo homocysteine sampling at baseline, the hazard ratio was 1.49 (CI, 0.79 to 2.80).

Discussion

In our randomized, placebo-controlled clinical trial, decreasing homocysteine levels with folic acid and vitamin B₆ and B₁₂ supplementation did not decrease symptomatic venous thromboembolism in middle-aged or older adults. This lack of a statistically significant effect was consistent across demographic subgroups, regardless of whether participants were from countries with folic acid food fortification and what participants' baseline plasma homocysteine levels were.

Our study has limitations. First, venous thromboembolism outcomes were not centrally adjudicated. Second, the first recording of venous thromboembolic events occurred 18 months after study enrollment; nonetheless, nearly 20% of all events occurred during this period. Third, the proportion of participants with a previous episode of venous thromboembolism was not known. The criteria that we used to define venous thromboembolism were more sensitive and specific than those used in the original HOPE-2 (8), resulting in an approximately 0.1% higher incidence in our study. Finally, the wide CI of 0.66 to 1.53 for the hazard ratio of deep venous thrombosis may reflect some uncertainty about whether homocysteine treatment is helpful or harmful. Together, these limitations suggest that some participants may have been incorrectly classified as having venous thromboembolism, thereby reducing our ability to detect a true benefit of homocysteine-lowering therapy, especially in the first 18 months after study entry.

The HOPE-2 is the largest randomized clinical trial to evaluate the effect of homocysteine-lowering therapy on venous thromboembolism, and fewer than 1% of participants were lost to follow-up. The study had a placebo-controlled design and a prospective assessment of venous thromboembolism that included objective confirmation. Our analysis greatly expands on the initial HOPE-2 findings (8) by focusing on the anatomical location of venous thromboembolism (that is, deep venous thrombosis and pulmonary embolism) and whether it was provoked, according to important risk factors, such as age and sex, baseline homocysteine level, and the presence of folic acid food fortification.

Although the HOPE-2 participants were not enrolled on the basis of having a high risk for deep venous thrombosis, many deep venous thrombosis events occurred at a rate slightly higher than that seen in other studies (9, 10). Given the low rate of oral anticoagulant use at baseline, it is likely that few participants in our study had a recent history of venous thromboembolism. In the Vitamins and Thrombosis (VITRO) trial (11), the value of homocysteine-lowering therapy was evaluated for prevention of recurrent venous thromboembolism after a first episode of unprovoked proximal deep venous thrombosis or pulmonary embolism. Researchers randomly assigned 360 patients with elevated homocysteine levels and 341 patients with normal homocysteine levels to receive 5 mg of folic acid, 50 mg of vitamin B₆, and 0.4 mg of vitamin B₁₂ or placebo. After a 2.5-year follow-up, no convincing benefit of homocysteine-lowering therapy was found among all participants (hazard ratio, 0.84 [CI, 0.65 to 1.98]) or among those with elevated homocysteine levels (hazard ratio, 1.14 [CI, 0.65 to 1.98]). Therefore, our findings are similar to those of the VITRO study and do not support the hypothesis that homocysteine-lowering therapy decreases the risk for venous thromboembolism.

Our study was a secondary analysis of the HOPE-2 trial, which included older adults at high risk for atherosclerosis but not those specifically at high risk for venous thromboembolism. Only one half of events were unprovoked. The upper-quartile homocysteine level of 13.8 µmol/L that we used was slightly lower than the 95th percentile cut-points used to define hyperhomocystinemia in observational studies of risk for venous thromboembolism (1–4). These and the other limitations of our study preclude us from applying these data to younger persons; persons at high risk for venous thromboembolism (especially unprovoked events); and, perhaps, to persons with moderate hyperhomocystinemia. However, to our knowledge, no randomized clinical trial has demonstrated a treatment benefit in these groups.

Current evidence suggests that homocysteine-lowering therapy cannot be recommended for prevention of first or recurrent episodes of venous thromboembolism. Because the lack of efficacy of homocysteine-lowering therapy seems to be independent of plasma homocysteine level, measurement of homocysteine in adults with provoked or unprovoked venous thromboembolism does not seem to be justified (2–4). Among children and young adults with venous thromboembolism or arterial thrombosis in whom homocystinuria is suspected, measurement of plasma homocysteine may still be indicated (12).