Antiplatelet therapy and the effects of B vitamins in patients with previous stroke or transient ischaemic attack: a post-hoc subanalysis of VITATOPS, a randomised, placebo-controlled trial

Summary Background Previous studies have suggested that any benefits of folic acid-based therapy to lower serum homocysteine in prevention of cardiovascular events might be offset by concomitant use of antiplatelet therapy. We aimed to establish whether there is an interaction between antiplatelet therapy and the effects of folic acid-based homocysteine-lowering therapy on major vascular events in patients with stroke or transient ischaemic attack enrolled in the vitamins to prevent stroke (VITATOPS) trial. Methods In the VITATOPS trial, 8164 patients with recent stroke or transient ischaemic attack were randomly allocated to double-blind treatment with one tablet daily of placebo or B vitamins (2 mg folic acid, 25 mg vitamin B6, and 500 μg vitamin B12) and followed up for a median 3·4 years (IQR 2·0–5·5) for the primary composite outcome of stroke, myocardial infarction, or death from vascular causes. In our post-hoc analysis of the interaction between antiplatelet therapy and the effects of treatment with B vitamins on the primary outcome, we used Cox proportional hazards regression before and after adjusting for imbalances in baseline prognostic factors in participants who were and were not taking antiplatelet drugs at baseline and in participants assigned to receive B vitamins or placebo. We also assessed the interaction in different subgroups of patients and different secondary outcomes. The VITATOPS trial is registered with ClinicalTrials.gov, number NCT00097669, and Current Controlled Trials, number ISRCTN74743444. Findings At baseline, 6609 patients were taking antiplatelet therapy and 1463 were not. Patients not receiving antiplatelet therapy were more likely to be younger, east Asian, and disabled, to have a haemorrhagic stroke or cardioembolic ischaemic stroke, and to have a history of hypertension or atrial fibrillation. They were less likely to be smokers and to have a history of peripheral artery disease, hypercholesterolaemia, diabetes, ischaemic heart disease, and a revascularisation procedure. Of the participants taking antiplatelet drugs at baseline, B vitamins had no significant effect on the primary outcome (488 patients in the B-vitamins group [15%] vs 519 in the placebo group [16%]; hazard ratio [HR] 0·94, 95% CI 0·83–1·07). By contrast, of the participants not taking antiplatelet drugs at baseline, B vitamins had a significant effect on the primary outcome (123 in the B-vitamins group [17%] vs 153 in the placebo group [21%]; HR 0·76, 0·60–0·96). The interaction between antiplatelet therapy and the effect of B vitamins on the primary outcome was significant after adjusting for imbalance in the baseline variables (adjusted p for interaction=0·0204). Interpretation Our findings support the hypothesis that antiplatelet therapy modifies the potential benefits of lowering homocysteine with B-vitamin supplementation in the secondary prevention of major vascular events. If validated, B vitamins might have a role in the prevention of ischaemic events in high-risk individuals with an allergy, intolerance, or lack of indication for antiplatelet therapy. Funding Australia National Health and Medical Research Council, UK Medical Research Council, Singapore Biomedical Research Council, and Singapore National Medical Research Council.


Summary
Background Previous studies have suggested that any benefi ts of folic acid-based therapy to lower serum homocysteine in prevention of cardiovascular events might be off set by concomitant use of antiplatelet therapy. We aimed to establish whether there is an interaction between antiplatelet therapy and the eff ects of folic acid-based homocysteinelowering therapy on major vascular events in patients with stroke or transient ischaemic attack enrolled in the vitamins to prevent stroke (VITATOPS) trial.

Methods
In the VITATOPS trial, 8164 patients with recent stroke or transient ischaemic attack were randomly allocated to double-blind treatment with one tablet daily of placebo or B vitamins (2 mg folic acid, 25 mg vitamin B 6 , and 500 μg vitamin B 12 ) and followed up for a median 3·4 years (IQR 2·0-5·5) for the primary composite outcome of stroke, myocardial infarction, or death from vascular causes. In our post-hoc analysis of the interaction between antiplatelet therapy and the eff ects of treatment with B vitamins on the primary outcome, we used Cox proportional hazards regression before and after adjusting for imbalances in baseline prognostic factors in participants who were and were not taking antiplatelet drugs at baseline and in participants assigned to receive B vitamins or placebo. We also assessed the interaction in diff erent subgroups of patients and diff erent secondary outcomes. The VITATOPS trial is registered with ClinicalTrials.gov, number NCT00097669, and Current Controlled Trials, number ISRCTN74743444.
Findings At baseline, 6609 patients were taking antiplatelet therapy and 1463 were not. Patients not receiving antiplatelet therapy were more likely to be younger, east Asian, and disabled, to have a haemorrhagic stroke or cardioembolic ischaemic stroke, and to have a history of hypertension or atrial fi brillation. They were less likely to be smokers and to have a history of peripheral artery disease, hypercholesterolaemia, diabetes, ischaemic heart disease, and a revascularisation procedure. Of

Introduction
Observational studies show a strong, positive, and doserelated association between serum concen trations of homocysteine and the risk of stroke, which is independent of other vascular risk factors and biologically plausible. 1,2 Homocysteine can be lowered by a mean of 25% (95% CI 23-28) with folic acid supplementation. 3 A metaanalysis of eight randomised, placebo-controlled trials of folic acid supplementation in 37 485 patients 4 showed that, despite yielding an average 25% reduction in homo cysteine, folic acid had no signifi cant eff ect on the rate of fi rst stroke (rate ratio 0·96, 95% CI 0·87-1·06) over a median followup of 5 years. However, the role of homocysteine-lowering in stroke prevention might be complex. 5 A meta-analysis of 237 genetic epidemiological studies, 6 in which homo cysteine and the presence of the methylene tetrahydro folate reductase C677T poly morphism in 60 000 individuals were correlated with 20 885 subsequent stroke events, suggested that established or increasing dietary folate intake in the countries where the trials were undertaken might have modifi ed the eff ect of lowering homocysteine on risk of stroke. 6 Antiplatelet therapy might also modify the eff ect of lowering homocysteine on the risk of stroke and ischaemic heart disease events. [7][8][9] An exploratory analysis of trials of lowering homocysteine 7 suggested an interaction between antiplatelet therapy and the eff ect of lowering homocysteine on risk of ischaemic heart disease events: in the fi ve trials with the lowest prevalence of antiplatelet therapy (mean 60%, usually aspirin), the relative risk was 0·93 (95% CI 0·84-1·05) and in the fi ve trials with the highest prevalence (mean 91%) the relative risk was 1·09 (1·00-1·19), p for interaction=0·037. In another analysis of trials of the eff ects of lowering homocysteine on the risk of stroke events, 8 the eff ect was greater in the four trials that enrolled patients with renal disease and oesophageal dysplasia (who were not likely to be taking antiplatelet therapy) compared with the trials that enrolled patients with previous vascular disease. The Heart Outcomes Prevention Evaluation 2 (HOPE 2) trial 9 subsequently reported a non-signifi cant trend towards a greater eff ect of folic acid-based vitamin B sup plementation, compared with placebo, in reducing stroke in patients with known cardiovascular disease who were not taking antiplatelet therapy at enrolment compared with patients who were (p for interaction=0·25). The biological plausibility of these fi ndings is supported by the recognised potential for antiplatelet therapy to modify any antithrombotic or other antiatherogenic eff ects of lowering homo cysteine. [10][11][12][13] These analyses prompted us to undertake a post-hoc subanalysis of the vitamins to prevent stroke (VITATOPS) trial. We aimed to explore the hypothesis that there is an interaction between antiplatelet therapy and the eff ect of folic acid-based vitamin B supplementation on major vascular events in the VITATOPS trial population of patients with previous stroke or transient ischaemic attack. 14

Participants
The methods and primary results of the VITATOPS trial have been reported. 14 Briefl y, the VITATOPS trial was a randomised, double-blind, parallel, placebo-controlled trial in which 8164 patients were recruited from 123 centres in 20 countries of four continents, and randomly assigned to take one tablet daily of placebo or B vitamins (2 mg folic acid, 25 mg vitamin B 6 , 500 μg vitamin B 12 ). Patients were eligible for inclusion if they had a stroke (ischaemic or haemorrhagic) or transient ischaemic attack (eye or brain) within the past 7 months.
Patients were excluded if they were taking folic acid, vitamin B 6 , vitamin B 12 , or a folate antagonist (eg, methotrexate), if they were pregnant or were women of Data are mean (SD) or n (%). *History of hypertension or treated hypertension at randomisation. †History of hypercholesterolaemia (>6·5 mmol/L) or treated hypercholesterolaemia at randomisation. At enrolment, participants were asked if they were taking antiplatelet drugs (eg, aspirin, clopidogrel, dipyridamole). The trial received ethical approval from national (India, New Zealand, and the UK) and local research ethics committees and all patients provided written informed consent before enrolment.

Procedures
Patients were randomly assigned (1:1) to receive either B vitamins or matching placebo by means of a central 24 h telephone service or an interactive website in which random permuted blocks were stratifi ed by hospital. Treatment groups were masked from patients and investigators. Randomisation was not stratifi ed in accordance with the presence or absence of antiplatelet therapy. The primary outcome was the composite of any stroke, myocardial infarction, or death from vascular causes.

Statistical analysis
We tabulated baseline characteristics and laboratory data in accordance with the presence or absence of antiplatelet therapy at baseline and in accordance with the assigned treatment groups, and expressed them as proportions for categorical variables and means for continuous variables. We compared categorical variables in each group with the χ² test, and continuous variables with the t test. We calculated event rates as the number of events during the follow-up period divided by the total number of patients that entered randomisation.
We constructed Kaplan-Meier curves to show the cumulative eff ects of B vitamins compared with placebo on the primary outcome in participants who were and were not taking antiplatelet therapy at baseline.
We assessed the interaction between antiplatelet therapy and the eff ects of treatment with B vitamins on the primary outcome by means of Cox proportional hazards regression before and after adjusting for imbalances in important baseline prognostic factors in participants who were and were not taking antiplatelet drugs at baseline, and in participants assigned B vitamins or placebo. We also assessed the consistency of the interaction eff ect in diff erent subgroups of patients, and in diff erent secondary outcome events including ischaemic stroke, haemorrhagic stroke, myocardial infarction, and death from vascular causes.
We adjusted for certain variables in our models: age, sex, ethnic origin, pathological and causal subtypes of stroke and transient ischaemic attack, stroke severity as measured by the Oxford handicap score, smoking, treated and untreated hypercholesterolaemia, and history of stroke, myocardial infarction, ischaemic heart disease, peripheral arterial disease, atrial fi brillation, and diabetes.
We compared the mean serum concentrations of homocysteine and vitamin B 12 and mean red-cell Data are mean (SD) or n (%). *History of hypertension or treated hypertension at randomisation. †History of hypercholesterolaemia (>6·5 mmol/L) or treated hypercholesterolaemia at randomisation. con centration of folate, which were measured at both baseline and follow-up in the same individual, with a paired t test. We calculated the diff erence between baseline and follow-up measures, and tested the interaction eff ect between antiplatelet use at baseline and treatment allocation with a linear regression model. We used two-sided signifi cance tests throughout and we deemed a two-sided p value of less than 0·05 to be signifi cant. The VITATOPS trial is registered with ClinicalTrials.gov, number NCT00097669, and Current Controlled Trials, number ISRCTN74743444.

Role of the funding source
The sponsors of the study had no role in study design, data collection, data analysis, data interpretation, the writing of the report, or in the decision to submit the paper for publication. The corresponding author had full access to all the data in the study and had fi nal responsibility for the decision to submit for publication.

Results
At baseline, 6609 patients (81%) were in receipt of antiplatelet therapy, 1463 (18%) were not, and in 92 (1%) antiplatelet therapy status was not known. The composite primary outcome of stroke, myocardial infarction, or death from vascular causes was recorded in 616 patients (15%) assigned to receive B vitamins and 678 (17%) assigned to receive placebo (risk ratio 0·91, 95% CI 0·82 to 1·00, p=0·05; absolute risk reduction 1·56%, 95% CI -0·01 to 3·16). 14 Compared with patients receiving antiplatelet therapy, patients who were not receiving antiplatelet therapy at baseline were more likely to be younger, east Asian, and disabled, to have a haemorrhagic stroke or cardioembolic ischaemic stroke, and to have a history of hypertension or atrial fi brillation (table 1). They were less likely to be smokers and to have a history of peripheral vascular disease, hypercholesterolaemia, diabetes, ischaemic heart disease, and a revascularisation procedure. Of patients who were or were not receiving antiplatelet therapy at baseline, baseline characteristics were evenly distributed between patients assigned to receive either B vitamins or placebo (table 2).
Baseline antiplatelet therapy was an independent signifi cant predictor of a lower rate of subsequent stroke, myocardial infarction, or death from vascular causes in all patients who entered randomisation (hazard ratio [HR] 0·66, 95% CI 0·55-0·81).
Of the 6609 participants in receipt of antiplatelet drugs at baseline, the primary outcome was recorded in roughly 15% of participants assigned to receive B vitamins or placebo (table 3). By contrast, of the 1463 participants who were not in receipt of antiplatelet drugs at baseline, the primary outcome was recorded in slightly more participants in the placebo group (table 3). After adjusting for the eff ects of imbalance in baseline variables, the HR for the primary outcome for patients assigned B vitamins versus placebo was greater for participants taking antiplatelet therapy than for those who were not (table 3).
The fi gure shows Kaplan-Meier curves of the cumulative probability of the primary outcome event in patients who were and were not taking antiplatelet at the time of randomisation into the VITATOPS trial. In

Table 3: Interaction between B-vitamin supplementation and antiplatelet therapy at baseline on each major vascular outcome
for the primary outcome were consistent for stroke and for vascular death, but not for myocardial infarction. In table 4 we show a signifi cant interaction between antiplatelet use at baseline and the eff ect of B vitamins on recurrent ischaemic stroke after adjustment for baseline factors. The trend was similar, but not signifi cant, for recurrent haemorrhagic stroke.
In table 5 we show that of all the listed subgroups, with the exception of participants with cardioembolic ischaemic stroke, the HR for the eff ect of B vitamins compared with placebo on the primary outcome was lower in patients who were not in receipt of antiplatelet therapy at baseline than in patients who were, but many of the comparisons were not statistically signifi cant.
In table 6 we show that supplementation with B vitamins signifi cantly lowered total homocysteine and increased red cell folate concentration during follow-up in patients who were and were not in receipt of antiplatelet therapy at baseline. Supplementation with B vitamins also signifi cantly increased serum vitamin B 12 con centration during follow-up in patients in receipt of antiplatelet therapy at baseline, but the eff ect was not signifi cant for patients not receiving antiplatelet therapy at baseline. The eff ects of supplementation with B vitamins on lowering total homocysteine and increasing red-cell folate and vitamin B 12 concentration were not signifi cantly diff erent between patients who were and were not in receipt of antiplatelet therapy at baseline. The p for interaction between antiplatelet therapy at baseline and trial treatment was 0·2501 for total homocysteine, 0·8996 for red cell folate, and 0·6591 for vitamin B 12 .
After excluding patients with a qualifying diagnosis of haemorrhagic stroke, the interaction between B vitamins and antiplatelet therapy was not signifi cant (adjusted p=0·1159), but the adjusted HR for B vitamins versus placebo on the primary outcome in participants not in receipt of antiplatelet therapy at baseline was still lower (HR 0·75, 95% CI 0·54-1·03) than in participants who were in receipt of therapy (0·98, 0·86-1·12). We also did a matched paired analysis, and a similar pattern was evident.

Discussion
The principal result of the VITATOPS trial was that daily administration of B vitamins to patients with recent stroke or transient ischaemic attack for a median of 3·4 years had no signifi cant eff ect, compared with placebo, on the overall incidence of major vascular events. However, our post-hoc subanalysis supports hypotheses from previous independent trials of lowering total homocysteine on both ischaemic heart disease and stroke outcome events that antiplatelet therapy, which was taken by most patients, might have modifi ed any favourable eff ect of folic acid sup plementation on major vascular events (panel). 7,9 The VITATOPS trial had several strengths: systematic bias in treatment allocation was minimised by the randomisation process; observer bias in the assessment of vascular outcomes was minimised by the masking of treatment allocation from assessors, clinicians, and patients; and random error was reduced by the reasonably large number of outcome events. The strengths of our analysis are that it was based on a pre-existing hypothesis (that antiplatelet therapy might interact with the eff ect of B vitamins on vascular risk), the hypothesis is plausible, the interaction between B-vitamin supplementation and only one subgroup was assessed (antiplatelet use at baseline or not; table 3), the primary trial outcome was the main outcome studied, the distribution of important prognostic factors was reasonably, although not perfectly, balanced between treatment groups within each subgroup (table 2), the analysis was based on appropriate statistical tests of subgroup-treatment eff ect interaction, all subgroup analyses that were undertaken have been reported, and the results have been interpreted cautiously on the premise that subgroup analyses are intrinsically limited. 15   Potential limitations are that, because this substudy was not a primary aim or prespecifi ed analysis of the VITATOPS trial, the type of antiplatelet therapy taken (eg, aspirin, clopidogrel, aspirin combined with dipyridamole) was not recorded, and there was a signifi cant imbalance in baseline characteristics of participants in receipt of antiplatelet therapy compared with participants who were not (table 1), and a mild imbalance in baseline characteristics in participants assigned to receive B vitamins versus placebo (table 2). The more favourable recorded eff ect of B vitamins in participants not in receipt of antiplatelet therapy might have been confounded by the reason they were not in receipt of the therapy-ie, B vitamins might have been more eff ective in patients of east Asian origin or patients with cardioembolic ischaemic stroke or intracerebral haemorrhage (who tend not be given antiplatelet drugs). However, we adjusted for the eff ects of this imbalance on the rates of each vascular outcome in our Cox multiple regression analysis. Through our Cox analysis we identifi ed that, after adjusting for these eff ects, the use of antiplatelet therapy at baseline was a signifi cant, independent predictor of the incidence of major vascular events (p<0·0001) and that there was a signifi cant interaction between anti platelet therapy and treatment with B vitamins on the primary outcome (adjusted p for Small artery disease interaction=0·0204), stroke (adjusted p for inter-action=0·0134), and death from vascular causes (adjusted p for interaction=0·0225). We acknowledge the possibility of residual imbalance in other, unmeasured, prognostic factors at baseline, for which we could not adjust our analysis, and that such residual confounding after adjusting for imbalances in measured prognostic factors (eg, haemorrhagic stroke, cardioembolic ischaemic stroke) could aff ect our results. We also acknowledge that our fi ndings might represent not an interaction of B-vitamin supplementation with antiplatelet therapy but a signifi cant eff ect of lowering homocysteine by B-vitamin supplementation in patients with haemorrhagic stroke or cardioembolic ischaemic stroke.
If our fi ndings are valid, the mechanisms by which raised homocysteine might impair vascular function in the absence of antiplatelet therapy remain to be ascertained. Laboratory investigations suggest several potential mechanisms, including impairment of endothelial function, oxidation of low-density lipids, increased monocyte adhesion to the blood vessel wall, increased lipid uptake and retention, activation of infl ammatory pathways, stimulatory eff ects on smoothmuscle-cell proliferation, and pro thrombotic ten dency mediated by activation of coagulation factors and platelet dys function. [11][12][13] If antiplatelet therapy really does modify the eff ects of lowering homocysteine on vascular outcomes, this might be mediated by direct eff ects of antiplatelet drugs on platelet activation and thrombus formation, or indirect eff ects of antiplatelet drugs, such as aspirin, in reducing vasoconstrictor tone, vascular smooth-muscle-cell proliferation, and release of infl am matory cytokines, oxygen radicals, and growth factors. 10 In conclusion, our fi ndings of a signifi cant interaction between antiplatelet therapy and the eff ect of B vitamins on the primary outcome, in our exploratory analysis of an independent group of patients with previous stroke or transient ischaemic attack, support the hypothesis generated from other studies that antiplatelet therapy might modify any potential benefi ts of lowering homocysteine with folic-acid supplementation in the secondary prevention of major vascular events. Rather than antiplatelet therapy negating all of the eff ects of lowering homocysteine, it is also possible that lowering homocysteine might have a small benefi t independent of antiplatelet therapy and a larger benefi t in the absence of additional prophylactic antiplatelet therapy.
The external validity of our fi ndings can be assessed more reliably by means of a meta-analysis of the relevant data from all individual patients enrolled in

Systematic review
We searched PubMed with the terms "homocysteine", "folic acid", "vitamins", "antiplatelet", "aspirin", "clopidogrel", "dipyridamole", "cilostazol", "stroke", "ischaemic heart disease", "major vascular events", "interaction", "randomised trial", and "clinical trial" for reports of an interaction between antiplatelet therapy and treatments that lower homocysteine in the prevention of stroke and other major vascular events. We searched for work published before March, 2012. The quality of evidence we required was a randomised, controlled trial or meta-analysis of such trials. We identifi ed the Heart Outcomes Prevention Evaluation 2 trial 9 and the meta-analysis of randomised trials of lowering total homocysteine on risk of ischaemic heart disease events 7 as directly relevant, and a further meta-analysis 8 as indirectly relevant.

Interpretation
The results of our exploratory analyses of the VITATOPS trial support previous hypotheses that antiplatelet therapy, which was taken by most patients, might modify any favourable eff ect of folic acid supplementation on major vascular events. 7,9 If our fi nding are validated in independent studies, B vitamins might have a role in the prevention of vascular events in high-risk individuals with an allergy, intolerance, or lack of indication for antiplatelet therapy, such as those with haemorrhagic stroke. Data are mean (SD) unless otherwise stated. *Comparison between baseline and during the follow-up was undertaken with a paired t test. Some of the follow-up measures were taken during follow-up (eg, at the regular follow-up assessments every 6 months) and some at the end of follow-up. trials of B vitamins to prevent both stroke and ischaemic heart disease events. If validated, the implications of the fi ndings for clinicians are that B vitamins might have a role in the prevention of vascular events in individuals at high risk but who have an allergy to, intolerance of, or lack of indication for antiplatelet therapy, such as those who are also at risk of bleeding events (eg, haemorrhagic stroke).

Contributors
GJH initiated the analysis for this substudy and wrote the fi rst and fi nal drafts of the report. QY did all the analyses. JWE, KRL, CC, DX, JCN, UKR, WU, SR, JG, and RS contributed to each draft of the report. All authors were members of the International Steering Committee of the VITATOPS trial.

Confl icts of interest
We declare that we have no confl icts of interest.