Associations of dietary folate, vitamin B6 and B12 intake with cardiovascular outcomes in 115664 participants: a large UK population-based cohort

The evidence of relationship between dietary intake of folate, vitamin B6 and vitamin B12 and cardiovascular diseases (CVD) in UK populations is limited. We aimed to analyze the association of dietary intake of folate, vitamin B6, and vitamin B12 with CVD events [stroke, myocardial infarction (MI)] and CVD mortality. We included 115,664 participants, aged 40–70 years, with no CVD events or cancer at baseline, enrolled between 2006 and 2010 and followed up to the end of 2018. Dietary intake was measured with an online 24-h dietary assessment. Multivariable Cox proportional hazards models were used to estimate hazard ratios (HR) and 95% confidence intervals (CI) for the associations. After multivariate adjustment, higher dietary folate intake was inversely associated with CVDs with hazard ratios of 0.99, 0.92, and 0.88 in groups 2–4 compared with group 1 (the lowest group); inversely associated with stroke with hazard ratios of 0.94, 0.90, and 0.86 groups 2–4 compared to group 1 (lowest group); inversely associated with MI with hazard ratios of 1.01, 0.90 and 0.86 groups 2–4 compared to group 1 (lowest group); inversely associated with CVD mortality with hazard ratios of 0.95, 0.80 and 0.74 Groups 2–4 compared to group 1 (lowest group). Each tablespoon/day higher intake of raw vegetable intake, pieces/day higher intake of fresh fruit intake bowls/week higher intake of cereal intake, and g/day higher intake of dietary fiber were associated with higher intakes of folate every 0.02,0.06,0.05, and 0.08 SD, respectively. E-value analysis suggested robustness to unmeasured confounding. Each increase in folate intakes was related to 5% lower risks of total CVD events and 10% lower risks of CVD mortality. Our findings support that strengthening dietary folate intake as a primary prevention strategy for CVD events and CVD mortality.


INTRODUCTION
Vitamin B6 (pyridoxine), vitamin B9 (folate), and vitamin B12 (cobalamin) are essential cofactors of homocysteine metabolism in the body [1,2]. Abundant clinical studies have shown that increased accumulation of homocysteine will significantly increase the risk of coronary heart disease, stroke, and peripheral vascular disease [3]. A recent meta-analysis of folate and B-complex vitamins in RCTs demonstrated that folate reduced total CVD and stroke by 17% and 21%, respectively, and B-complex vitamins reduced stroke by 10% [4]. Furthermore, recent epidemiological studies have shown that intake of B vitamins (folate, vitamin B6, and vitamin B12) reduced the risk of CVD [5].
Emerging evidence suggests that folate, vitamin B6 and vitamin B12 may play an important role in reducing the risk of cardiovascular disease (CVD) events [6]. Several cross-sectional and prospective cohort studies showed that both dietary intake of folate and vitamin B6 were associated with lower risk CVD events [7,8]. However, findings from observational studies have been inconsistent or controversial [9]. In addition, studies have shown that B vitamins can reduce blood homocysteine levels, but the combination of folic acid, B6 and B12 did not reduce the risk of CVD [10] Evidence for an association between dietary intake of these B vitamins and the risk of CVD events in different populations remains limited and insufficient [11,12]. In addition, previous publications did not analyze the relationship between dietary factors and dietary intake of B vitamins [6][7][8].
Therefore, in this prospective cohort study, we examined the association between dietary intake of folate, vitamin B6, and vitamin B12 and CVD events [stroke, myocardial infarction (MI)] and CVD mortality in over 500,000 participants in the UK Biobank. We further investigated the relationship between dietary intake of folate, vitamin B6, and vitamin B12 and dietary factors.

Study design and population
The detailed study design and methods of the UK Biobank study have been described elsewhere [13]. During 2006-2010, the UK Biobank recruited more than 500,000 community-dwelling participants aged 40-70 years who were recruited between April 2007 and December 2010. Participants provided relevant health information through touch screen questionnaires and face-toface interviews [14]. Data from 502505 participants can be used in our research. Participants without available measurements of folate, vitamin B6, and vitamin B12 intake were excluded. Participants with CVD or cancer (n = 40,024) at baseline were excluded from the analysis. We further excluded participants whose baseline demographic and lifestyle factors measured data were of poor quality (n = 54,898). Our final analysis included 115,664 participants ( Supplementary Fig. 1).

Ascertainment of exposure
Dietary intake was assessed with a 24-h dietary recall questionnaire that included consumption of approximately 200 commonly consumed foods and drinks. The Oxford Cancer Epidemiology Department of the UK Biobank evaluated the validity of the developed questionnaire. More details can be found elsewhere [15]. The questionnaire contains detailed questions about the food and beverage intake consumed in the previous 24 h. Participants needed to answer the questionnaire regarding what food and drinks they had consumed yesterday. By multiplying the number of servings of each food or beverage by the quantity consumed, the quantity of each food and beverage consumed in the previous 24 h can be calculated. Detailed information on the measurements is provided on the UK Biobank website (https://biobank.ndph.ox.ac.uk/showcase/showcase/docs/ DietWebQ.pdf).
We evaluated several possible confounding variables using the baseline questionnaire, including sociodemographic factors (age, sex, ethnicity, and employment), lifestyle habits (smoking status, physical activity, drinking status), body mass index, and blood samples (HDL, LDL, triglyceride, cholesterol). Baseline disease history (diabetes, hypertension, and cancer) was collected through self-reporting. Details of these assessments were available on the UK Biobank website (www.ukbiobank.ac.uk).
Ascertainment of CVD events and CVD mortality CVD mortality and CVD events (myocardial infarction and stroke) were used as the primary outcomes of this study. We used the reported UK Biobank algorithms to determine CVD events. We collected information about CVD events and CVD mortality through certified death records and cumulative medical records diagnosed by hospitals (until March 2018). Admission and diagnosis data of the records were used to ascertain incidents. CVD events with the ICD-10 (International Classification of Diseases, 10th revision). Stroke was defined as ICD-10 codes I60-I64. Myocardial infarction was defined as codes I21, I22, I23, I24.1, or I25.2. CVD mortality was defined as codes I00-I99.

Statistical analysis
Baseline characteristics are presented as the mean ± SD (standard deviation) for continuous variables, medians (interquartile range) for asymmetrical continuous variables and percentages for categorical variables. Participants were divided into different groups according to their dietary intake of folate, vitamin B6 and vitamin B12. T tests and chi-square tests were used to compare baseline characteristics between participants in the lowest to highest quartiles of vitamin intake.
Linear regression was used to estimate β coefficients and 95% (confidence interval, CI) the relationship between dietary factors and dietary intake of folate, vitamin B6, and vitamin B12. Model adjusted for demographic, lifestyle dietary factors (age, sex, BMI, physical activity, smoking status, employment, drinking status), and dietary factors (intake of total energy, cooked vegetable, raw vegetable, fresh fruit, bread, cereal, oily fish, non-oily fish, beef, processed meat, lamb/mutton, pork, dietary fiber, saturated fat, and polyunsaturated fat).
The proportional hazard assumption was evaluated by tests based on Schoenfeld residuals [16]. We used a Cox proportional hazard model to estimate the hazard ratio (HR) and 95% CI for incident CVD, stroke, and MI. Vitamins were compared by quartile in four groups (from the lowest group 1 to the highest group 4) and compared for each one standard deviation. We established three sets of models, adjusting for potential confounders: model 1 adjusted for age (years) and sex (male or female), and ethnic(white, black, Asian, mixed, or other ethnic groups); model 2, further adjusted for physical activity (<250 min/week, 250-550 min/week, >550 min/week), smoking status (never, former, current), drinking status (never, former, current), employment (no, yes), HDL cholesterol (continuous), LDL cholesterol (continuous) and total cholesterol; and model 3, further adjusted body index based on model2. Based on model 3, we used Cox proportional hazards models with restricted cubic splines (three knots) to evaluate and visualize the relationship of the dietary intake of folate with CVD events and CVD mortality [17].
According to sex (male or female), we performed a stratified analysis to estimate sex differences effects. We also conducted sensitivity analyses to confirm the robustness of our results. First, participants who had CVD events within 2 years of follow-up were excluded. Second, we considered other potential confounding factors, including waist-to-hip ratio, intake of vitamin D, vitamin supplements, diabetes, and hypertension. Third, we removed participants with baseline diabetes and hypertension. Finally, we calculated the E-values to explore the possibility of unmeasured confounding between dietary intake of folate, vitamin B6 and vitamin B12 and CVD events [18]. All statistical analyses were performed using SAS statistical software version 9.1 (SAS Institute, Inc., Cary, NC, USA). Table 1 shows the baseline characteristics of the study population stratified by quartiles of dietary intake of folate, vitamin B6, and vitamin B12. Baseline mean dietary intake of folate, vitamin B6, and vitamin B12 among participants without CVD events was 310.65 (126.73) ug/d, 2.21 (0.84) mg/d, and 6.65(5.55) ug/d, respectively, compared with 312.13 (129.73) ug/d, 2.25(0.90) mg/d and 6.76(6.02) ug/d in participants who developed CVD. Participants with higher folate intake and higher vitamin B6 intake were more likely to be male, slightly older, had lower body mass index, were more likely to be physically active, were less likely to be current smokers, but more likely to be current drinkers, and had lower HDL cholesterol. Participants with higher vitamin B12 intake were more likely to have higher HDL cholesterol and higher serum total cholesterol. (Supplementary Tables S1-S3) After stratification for potential CVD risk factors, dietary folic acid intake was observed in participants with CVD versus those without CVD. Baseline mean dietary intake of folic acid was higher in participants without CVD events than in participants with CVD in both men and women, respectively (Supplementary Table S4). The same results were also observed in different age groups.

Baseline characteristics
Association of dietary intake of folate, vitamin B6, and vitamin B12 with dietary factors In our analyses, after adjusting for all confounding factors, we found a positive correlation between sex (female compared with male), physical activity (compared with <250 min/week), and dietary intake of folate, vitamin B6, and vitamin B12, while employment (yes compared with no) and current drinking status (compared with never drinking) were inversely associated with dietary intake of folate, vitamin B6, and vitamin B12 (Supplementary Table S5).
Dietary factors of cooked vegetable intake and fresh fruit intake, raw vegetable intake, cereal intake, and oily fish intake were all positively associated with intake of folate, vitamin B6, and vitamin . For CVD mortality, the inverse association with dietary folate intake was still significant in models 1, 2, and 3. The HR for each SD was 0.90(0.83,0.97) (model 3). However, in the multivariate model, intake of vitamin b6 and vitamin B12 was not significantly associated with the risk of CVD events, stroke, myocardial infarction, or CVD mortality (Table 2).
We used restricted cubic splines to model the relationship between dietary folate intake and CVD risk. The results showed that dietary folate intake was not associated with CVD events at low concentrations, while at moderate to high concentrations (>300 µg/d), the relevant amounts were associated with CVD events (Fig. 2). The associations of dietary intake of folate with stroke ( Fig. 2B) and myocardial infarction are presented (Fig. 2C).
Female participants had lower prevalence of CVD events, hypertension and CVD mortality than males. Male participants had higher age, BMI, and current drinking and smoking rates than female participants (Supplementary Table S6). Stratified by gender, we observed that dietary folate intakes were inversely associated with CVD mortality for females (Fig. 3). More details were described in Supplementary Table S7.

Sensitivity analyses
After excluding participants with CVD events within two years of follow-up, the results were unchanged (Supplementary Table S8). Additionally, after excluding the participants who had baseline diabetes and hypertension, the results remained unchanged. In addition, after further adjusting for the waist-to-hip ratio, intake of vitamin D, and vitamin supplements, the results remained unchanged. Furthermore, we calculated an E value (E-values = 1.29) to evaluate the impact on unmeasured confounding.

DISCUSSION
In this large prospective study, we found inverse associations between folate intake and the risk of incidence of CVD events  (stroke, MI) and CVD mortality. Moreover, dietary folate intakes were inversely associated with CVD mortality for females. Each increase in folate intake was associated with a 5% lower risk of total CVD events, a 5% lower risk of stroke, a 6% lower risk of MI and a 10% lower risk of CVD mortality. In addition, each tablespoon/day higher intake of cooked vegetables and intake of raw vegetables was associated with 0.04 and 0.02 higher dietary folate intake by one SD, respectively.

Comparison with other studies
Our results are consistent with a number of prospective cohort studies of Americans and Europeans [19,20]. In a Nurses' Health Study including 80,082 women with no history of CVD events, cancer, hypercholesterolemia, and intake of folate were associated with a lowered risk for coronary heart disease [12]. In a metaanalysis involving 13 randomized controlled trials, using folic acid alone or a combination of folic acid and a small amount of cyanocobalamin (≤0.05 mg/day) reduced the risk of future stroke by 25% in countries without folate food fortification [21]. Studies have shown that cereal products fortified with folate in the United States can reduce neural tube defects (NTDs) [22]. Taken together, our study supports the fortification of food with folic acid. In addition, our research also analyzed the relationship between vitamin B intake and dietary factors. Dietary intake of folate was related to consumption of fruit, vegetables, cereal and oily fish in our study. Our findings provide evidence to increase fruit, vegetable, cereal, and oily fish intake for the prevention of CVD events. Although participants without CVD at baseline had slightly lower folate, B6, and B12 than those with CVD, after stratification by confounders, dietary folate intake was higher among those without CVD than those with CVD in each of the sex, age, and other strata. Multivariate analysis adjusted for these confounding factors. At present, the relationship between dietary vitamin B6 and vitamin B12 intake and CVD events is inconclusive. A cohort study showed that high dietary intake of vitamin B6 but not vitamin B12 was associated with a lower risk of stroke and coronary heart disease among Japanese patients [7]. Our findings indicate that there is no significant relationship between the intake of vitamin B6 and vitamin B12 and the incidence of stroke or MI. This may be due to the differences in study design, ethnicities, follow-up time, etc. among different studies. Therefore, further prospective studies with larger sample sizes in different populations are needed to further estimate this association.
Although men had higher dietary intakes of folate and vitamin B6 than women at baseline, we only observed that dietary intakes of folate reduced CVD mortality in women which may be related to women's healthier lifestyles.

Biological plausibility
The association we observed may involve folic acid intake reducing homocysteine concentrations in the blood. Some studies have shown that homocysteine is considered an independent vascular risk factor for sulfur-containing amino acids [23,24]. Deficiency of folic acid, vitamin B6 and vitamin B12 will lead to the occurrence of hyperhomocysteinemia [25]. Folate intake is inversely correlated with blood homocysteine, high levels of which may cause vascular damage. The metabolic process of abnormally rising homocysteine concentrations will produce a large amount of oxygen free radicals and other substances, which will damage the morphology of endothelial cells [25]. Accompanied by dysfunction, it will eventually lead to abnormal function of the fibrinolytic system and the occurrence of coagulation [26]. In addition, it will also produce a large amount of fibrinogen, which greatly enhances platelet activity and aggregation, produces a large number of blood clots, and causes hypertension, cerebral infarction, etc [27,28]. Homocysteine requires folic acid and vitamins as specific enzymes and cofactors in the normal metabolic conversion process. Deficiency of folic acid and vitamins leads to abnormal increases in homocysteine levels in the body, which may cause vascular damage [29,30].

Strengths and limitations
This study has several strengths. First, our study is based on a large, prospective cohort of more than 500,000 participants, which provided a large number of outcome events and the adequacy of information on lifestyle habits, diet, and other covariates, which enabled us to conduct adequate analyses and dose-response analysis. Second, we could conduct sensitivity analyses that could help to minimize confounding factors. Third, we performed several sensitivity analyses to confirm the robustness of our study. In addition, through a Cox proportional hazard model with folate dietary intake modeled as a continuous exposure variable, we tested the threshold of potential benefits of vitamin B dietary intake to provide more clinically relevant information. Finally, we calculated the E-value to quantify the potential impact of unmeasured confounding factors. Of course, there are some potential limitations in our study. First, we cannot exclude the influence of the remaining other constituents of vitamin supplements. However, we considered that the participants taking vitamin supplements may have a potential impact on the results, but after excluding those taking vitamin supplements, our sensitivity analysis correlation results did not change. Second, participants might change their diets after they develop some intermediate diseases, which may affect our results. To reduce the bias from this source, we excluded participants with a history of cancer and CVD events at baseline. The third limitation is selection bias. We observed that the participants included in the study had lower BMI and age and higher physical activity and vitamin B intake (Supplemental Table  S9), although these differences were not significant. Therefore, we cannot rule out the possibility that the observed associations are influenced by healthy lifestyle factors in participants with high vitamin B intake, while we carefully adjusted for potential confounding factors in our analysis.

CONCLUSIONS
In summary, higher intakes of folate were inversely associated with a lower risk of total CVD events, stroke, MI, and CVD mortality. Our research suggests that higher cooked vegetable intake, fresh fruit intake, raw vegetable intake, cereal intake, and oily fish intake may help reduce the risk of CVD events. The public health significance of this result is that the consumption of even a moderate increase in folic acid-rich foods may play a role in the prevention of CVD [31]. Although we also demonstrated associations with high dietary folate intake and CVD, the mechanism of folate in the prevention of CVD, as well as the optimal dose and combination, need to be further studied.

DATA AVAILABILITY
Data may be obtained from a third party and are not publicly available. We used UK Biobank data to analyse and report the findings. Data access policy can be obtained from https://www.ukbiobank.ac.uk/. Fig. 3 Association of dietary intake of folate with CVD events stratified by sex. Results were adjusted for age, sex and ethnic (white, mixed, Asian or Asian British, black or black British, other), physical activity (<250 min/week, 250-550 min/week, >550 min/ week), smoking status (never, former, current), employment (no, yes), high density lipoprotein cholesterol (continuous), low density lipoprotein cholesterol (continuous) and total cholesterol (continuous) and body mass index. Hazard Ratios (HRs) at per 1-standard deviation (SD) of each vitamin, estimated from Cox regression models.