Dietary Folate, Vitamin B6, and Vitamin B12 and Risk of Cardiovascular Diseases among Individuals with Type 2 Diabetes: A Case-Control Study

Background: Emerging evidence suggests that dietary one-carbon metabolism-related B-vitamins are associated with the reduced risk of cardiovascular disease (CVD) in the general population. However, only a few studies have assessed their associations in patients with type 2 diabetes (T2D). Objective: This study aimed to assess the associations between the intake of three one-carbon metabolism-related B-vitamins (folate, vitamin B6, and vitamin B12) and CVD risk in Chinese patients with T2D. Methods: A hospital-based case-control study of 419 patients with T2D and newly diagnosed CVD and 419 age- (±5 years) and sex-matched T2D-only controls was carried out in China. A validated 79-item semi-quantitative food-frequency questionnaire administered in face-to-face interviews was used to measure dietary B-vitamin intake. Conditional logistic regression was used to assess associations, which were tested by estimating odds ratios (ORs) with 95% confidence intervals (CIs). Results: Compared with the lowest quartile, the multivariable-adjusted ORs and 95% CIs for highest quartile were 0.32 (95% CI: 0.20, 0.52; p trend <0.001) for folate, 0.47 (95% CI: 0.30, 0.76; p trend = 0.002) for vitamin B6, and 1.02 (95% CI: 0.67, 1.55; p trend = 0.841) for vitamin B12. Consistent inverse associations were found for folate intake from eggs, vegetables, fruits, soy, and other foods but not for folate intake from grains. Conclusions: Findings suggest that the high consumption of folate and vitamin B6, but not that of vitamin B12, might be associated with the low risk of CVD in patients with T2D. This study suggests that dietary folate and vitamin B6 protect against CVD in patients with T2D.


Introduction
Cardiovascular diseases (CVDs) remain the leading cause of death worldwide, as well as in China, with approximately 3.97 million Chinese dying of the disease in 2016 [1,2]. Type 2 diabetes (T2D) is a common risk factor for CVD, and patients suffering from T2D are at a 2-to Wu 4-fold higher risk of CVD than the general population [3]. Thus, exploring approaches to reduce the CVD risk among patients with T2D is particularly important.
One-carbon metabolism-related B-vitamins, including folate, vitamin B 12 , and vitamin B 6 , have been proposed to play an important role in the development of CVD, mainly through their effects on homocysteine metabolism [4,5]. Homocysteine can be re-methylated into methionine through the one-carbon metabolism pathway, which requires a methyl group obtained from 5-methyltetrahydrofolate with vitamin B 12 acting as a co-enzyme, or can be further degraded into cysteine via vitamin B 6 -dependent reactions [6]. Elevated plasma homocysteine has been associated with arterial thrombosis and atherosclerosis, and underlying pathophysiological mechanisms, including increased oxidative stress, endothelial dysfunction, induced vascular inflammation, stimulated vascular smooth muscle cell proliferation, and increased platelet adhesiveness, potentially favour the progression of CVD [7]. Intervention studies on volunteers have also presented convincing evidence that taking supplements with folate, either alone or in combination with vitamin B 6 and/or B 12 , or consuming food rich in B-vitamins significantly lowers circulating homocysteine concentrations [8][9][10][11][12][13]. Higher homocysteine concentrations have been found in patients with diabetes than in generally non-diabetic populations and are considered to aggravate insulin resistance and vascular endothelial dysfunction in patients with T2D [14,15]. Compared with patients with diabetes without CVD complications, those with microor macrovascular diseases have higher prevalence rates of hyperhomocysteinaemia, and the interaction between high homocysteine levels and T2D with regard to CVD risk may be clinically relevant [16][17][18].
High B-vitamin intake has been associated with a significant reduction in the risk of CVD among generally non-diabetic populations [19][20][21]. The latest meta-analysis of 11 prospective cohort studies on the general population has demonstrated that an increase of 250 μg/day in folate or an increase of 0.5 mg/day in vitamin B 6 intake is associated with a 21% (95% CI: 11-31%) and 13% (95% CI: 4-22%) lower risk of coronary heart disease but did not find that vitamin B 12 has a protective effect [19]. Another large prospective follow-up study enrolled 43,732 healthy men and found that participants in the highest quintile of folate and vitamin B 12 intake have a 29% (95% CI: 4-48%) and 27% (95% CI: 1-47%) lower risk of ischaemic stroke than those in the lowest quintile but found no significant associations for vitamin B 6 [21]. However, no epidemiological study has assessed the effect of dietary B-vitamins on cardiovascular risk among patients with diabetes. In the present hospital-based study, we examined whether the dietary intake levels of folate, vitamin B 6 , and vitamin B 12 , which are nutrients involved in onecarbon metabolism, are associated with the risk of firstever CVD in a 1:1 matched case-control study with 419 pairs of Chinese individuals with T2D.

Materials and Methods
Study Design A 1:1 individual age-(±5 years) and sex-matched case-control study was conducted between March 2013 and September 2015. Cases of CVD complicated with T2D and T2D-only controls who were hospitalized at The First Affiliated Hospital of Sun Yat-sen University in Guangzhou, Guangdong Province, China, were recruited. The design of the study was described in a previous work [22]. The procedures and protocols of the present study were approved by the Ethics Committee of The First Affiliated Hospital of Sun Yat-sen University (No. [2017]019). Written informed consent was obtained from all participants.

Study Subjects
Inclusion Criteria Patients were included if they were aged between 30 and 85 years, natives of Guangdong Province or had lived in Guangdong for at least 5 years, and had a history of at least 2 years of T2D. The case and control subjects of the study were hospitalized patients with T2D, which was defined on the basis of American Diabetes Association criteria (fasting plasma glucose >7.0 mmol L −1 or 2-h plasma glucose >11.1 mmol L −1 or both) or medication treatment [23]. The CVD diagnoses of case subjects were confirmed within 2 weeks before the interview. CVD was defined as non-fatal acute myocardial infarction [24], hospitalized unstable angina [25], and non-fatal stroke [26]. Control subjects were patients with T2D who had never had CVD incidents, had no symptoms of cardiac involvement, had normal electrocardiogram levels, and had negative exercise tests.

Exclusion Criteria
Patients were excluded if they had confirmed type 1 diabetes or gestational diabetes mellitus; they had a previous history of cancer, hepatic disease, renal disease, autoimmune disorders, diabetic retinopathy, and congenital heart disease; they had physical disabilities and mental disturbances; their dietary habits or routine activities had significantly changed over the previous year; they had incomplete dietary assessment (≥10% missing values) or had an implausible total daily energy intake (<700 or >4,200 kcal per day for males, <500 or >3,500 kcal per day for females); or they refused to participate in the study.

Data Collection
Individual information was collected by using a structured questionnaire that comprised sociodemographic characteristics (e.g., age, sex, educational level, household income, long-term occupation, work status in the past year, and marital status), lifestyle habits (e.g., tobacco smoking, alcohol consumption, and tea drink- ing), history of chronic diseases and medications (e.g., hypertension, dyslipidaemia, insulin use, and oral hypoglycaemic use), physical activity, and dietary intake. Face-to-face interviews were carried out by well-trained dietitians to obtain relevant information. Each interviewer completed the same proportion of interviews with cases and controls.
In accordance with labour intensity, the occupations of the patients were categorized into light work (e.g., teachers, clerks, administrative staff, health care staff, engineering technicians, financial, and legal staff), moderate work (e.g., culture and arts personnel, businessmen and service personnel, housewives, and self-employed persons), or heavy work (e.g., workers, farmers, fishermen, and uniformed security personnel). Participants who had smoked at least one cigarette per day or drank alcohol once a week continuously for at least 6 months currently were defined as current smokers or current alcohol drinkers. Former smokers were once smokers but had quit smoking at least 1 year prior. Former alcohol drinkers were once drinkers but had quit drinking at least 1 year prior. Tea drinkers were defined as individuals who drank tea at least twice a week. Physical activity was calculated with a 19-item questionnaire by calculating the product of the time spent on a variety of activities (e.g., work, transportation, housework, leisure sedentary activity, and physical exercise) with the mean metabolic equivalent (MET) for that activity [27].
Anthropometric data, including weight (kg) and height (m), were collected by using standard procedures and measuring equipment by trained personnel at enrolment. Body mass index (BMI, kg/m 2 ) was then calculated. Participants whose mean systolic blood pressure was ≥140 mm Hg and/or whose diastolic blood pressure was ≥90 mm Hg and/or who used antihypertensive drugs were defined as hypertensive [28]. Individuals with cholesterol level ≥200 mg/dL, low-density lipoprotein cholesterol level ≥130 mg/ dL, high-density lipoprotein cholesterol level <40 mg/dL, or triglyceride level ≥160 mg/dL or who were taking lipid-lowering medications were defined as dyslipidaemic in accordance with the China Atherosclerosis Society guidelines [29].

Dietary Assessment
Dietary information was obtained from all participants who were assessed through an in-person interview by means of a validated 79-item semi-quantitative, paper-based, food-frequency questionnaire (FFQ) [30]. The reference period for the dietary survey was the year prior to diagnosis for patients with CVD or prior to the interview for control subjects. For each food item, five possible frequencies (never, per year, per month, per week, and per day) and the predefined amounts of food consumption were recorded in accordance with the choice of the participants. The food items were grouped under cereals, soy, vegetables, fruits, meats, eggs, fishery products, dairy products, beans and nuts, and drinks. The participants also reported their use of nutritional supplements. Food photographs with standard portion sizes were provided to help the participants quantify the amount of food they usually consumed. The selected choice for each food was then converted into a daily intake value (grams per day). The daily dietary intakes of nutrients (including folate, vitamin B 6 , and vitamin B 12 ) and total energy were calculated by multiplying the daily intake of each food item by the nutrient content based on the Chinese Food Composition Table, 2009 [31]. The reproducibility and validity of the FFQ have been previously reported in detail [30]. The correlation coefficients were 0.35, 0.26, and 0.50 for dietary folate, vitamin B 6 , and vitamin B 12 , respectively, when comparing the second FFQ and 18-day dietary records and were 0.60, 0.57, and 0.60 for folate, vitamin B 6 , and vitamin B 12 , respectively, when comparing the two FFQ [32,33].

Statistical Analysis
All analyses, except sex-stratified analyses, were performed for men and women combined. After log transformation, the average daily intakes of dietary folate, vitamin B 6 , and vitamin B 12 were separately adjusted for total energy intakes for men and women by using the residual model (energy-adjusted intake = a + b, where a = residual for subject from a regression model with nutrient intake as the dependent variable and total energy intake as the independent variable and b = the expected nutrient intake for a person with mean energy intake) [34]. The energy-adjusted intakes of folate, vitamin B 6 , and vitamin B 12 were grouped into quartiles (Q1-Q4) based on control subjects by sex, and the sex-specific cut-offs were then applied to the cases.
Group comparisons between cases and controls were performed through the t test, χ 2 test, and Wilcoxon rank-sum test as appropriate. The conditional logistic regression model was used to estimate the odds ratio (OR) and 95% confidence interval (CI) of each quartile, and the lowest quartile group (Q1) was defined as the reference. Univariate and multivariable models were used to calculate ORs with their corresponding 95% CIs. In the multivariable model, the adjusted variables included age (year), BMI (kg/ m 2 ), T2D duration (year), physical activity (MET-h/day), educational level (primary education degree or below, junior high school, senior high school or technical secondary school, college degree or above), household income (≤2,000, 2,001-4,000, 4,001-6,000, or ≥6,001 Yuan/month/person), occupation (light, moderate, or heavy work), work status (full-time/part-time or retired/ non-permanent), marital status (married or unmarried), smoking status (never, former, or current), alcohol consumption (never, former, or current), tea drinker (yes or no), hypertension status (yes or no), dyslipidaemia status (yes or no), antidiabetic medication use status (yes or no), antihypertensive drug use status (yes or no), lipid-lowering drug use status (yes or no), multivitamin supplement use status (yes or no), and log-transformed daily energy intake (kcal/day). The forward stepwise method was used to select variables for inclusion in the multivariable analyses, and only those that were statistically significant (p < 0.10) were included in the final models. Tests for trends were conducted by entering the categorical variables as continuous variables. All analyses were also repeated by using unconditional logistic regression modelling because undertaking a strict matched analysis would lead to the loss of relevant information [35].
Stratified analyses according to sex, BMI, smoking status, alcohol consumption, tea drinking status, hypertensive status, dyslipidaemia status, and T2D duration were also conducted. Stratification for alcohol consumption and smoking status was done only for the "never" and "current" groups because the "former" group was rather small. Given that these factors (except for sex) were not matched between case and control subjects, unconditional logistic regression adjusted for covariates was used rather than a conditional approach. Interactions were evaluated by using multiplicative models by including the product term in multivariable logistic regression. All analyses were conducted using SPSS 23.0 (IBM Corp., Armonk, NY, USA), and two-sided p < 0.05 was considered statistically significant.

Results
The characteristics of the case-control pairs are shown in Table 1. In total, 419 (235 male and 184 female) eligible cases of T2D complicated with CVD and 419 T2D-only controls were included in the current study. The mean age of the cases was 62.1 (SD 9.7) years and that of the controls was 62.1 (SD 9.5) years. Compared with the control subjects, the cases had a significantly higher BMI and proportions of heavy workers; patients with hypertension status; antihypertensive drug use and lipid-lowering drug use; lower T2D duration; physical activity; education level; and proportions of tea drinkers and antidiabetic medication users (all p < 0.05). No significant differences between cases and controls were observed in terms of age, household income, work status in the past year, marital status, smoking status, alcohol consumption, dyslipidaemia, and multivitamin supplement use.  1.07, 1.88) μg/day, respectively. The cases had a lower intake of folate and vitamin B 6 than the controls (all p < 0.05). No significant differences between the cases and controls were observed for vitamin B 12 intake. The details are shown in Table 2.
The main dietary sources of folate, vitamin B 6 , and vitamin B 12 in this study are listed in Table 3. Overall, eggs and vegetables, including pakchoi, rape, lettuce, spinach, amaranth, and macaroni, contributed >35.8% of the total folate intake; meat and vegetables were the major sources of total vitamin B 6 ; and meat and fish were the major sources of total vitamin B 12 . The five main sources of these nutrients accounted for 49.7% of folate intake, 50.8% of vitamin B 6 , and 88.2% of vitamin B 12 intake.
Except for vitamin B 12 and folate among cases, the intakes of folate, vitamin B 6 , and vitamin B 12 by the cases and controls were all significantly correlated with each other with Spearman's correlation coefficients of −0.032 High intakes of folate and vitamin B 6 were each related to a low risk of CVD among T2D patients ( Table 4). The associations were consistently inverse in univariate and multivariable analyses. After adjustment for various potential confounders, the adjusted ORs for the highest quartile of dietary folate and vitamin B 6 were found to be 0.32 (95% CI: 0.20, 0.52; p trend <0.001) and 0.47 (95% CI: 0.30, 0.76; p trend = 0.002), respectively, compared with those for the lowest quartile. The association for vitamin B 12 was not statistically significant, and the adjusted OR (95% CI) comparing the extreme quartiles was 1.02 (95% CI: 0.67, 1.55; p trend = 0.841).
The associations between folate intake and CVD according to the sources of folate are shown in Table 5. A consistent inverse association was found between folate intake from eggs, vegetables, fruits, soy, other foods and CVD risk (all p trends <0.05) but not among folate intake from grains (p trend = 0.432).
Stratified analyses were further conducted on the basis of sociodemographic factors, lifestyle, and comorbidity status (Fig. 1). No significant interactions were observed, and the associations of folate with CVD risk were similar for all strata (p interaction: 0.070-0.944). These findings remained unchanged when patients using multivitamin supplements, antidiabetic medication, non-lipid-lowering drugs, and non-anti-hypertensive drugs were excluded from sensitivity analyses.

Discussion
Our present hospital-based case-control study showed a statistically significant inverse relationship between the dietary intake of folate and vitamin B 6 and the risk of CVD among patients with T2D after adjusting for potential confounding factors. However, we found no independent associations for vitamin B 12 intake. The associations of folate with CVD risk were similar in all stratified analyses, and no significant interactions were observed. Several meta-analyses [19,20,36] relating the associations of dietary folate and vitamin B 6 with the risk of CVD in the general population have suggested that high intakes of folate and vitamin B 6 can reduce the risk of CVD with relative risk values for the highest compared with the lowest category ranging from 0.79 to 0.94 for folate and from 0.87 to 0.94 for vitamin B 6 . The higher homocysteine levels in patients with diabetes than in agematched control individuals suggested that homocysteine level is a potential cardiovascular risk marker in the setting of T2D subjects [37]. Thus, we hypothesized that sufficient B-vitamins could offer a distinct function in preventive approaches in individuals with T2D as well. However, the results of previous research are inconclusive [38][39][40]. A large prospective randomized trial, namely, the Heart Outcomes Prevention Evaluation 2 (HOPE 2) study, demonstrated that reducing homocysteine by supplementation with folic acid, vitamin B 6 , and vitamin B 12 for 5 years reduces the risk for overall stroke in patients with pre-existing CVD or with diabetes and additional risk factors by 25% [38]. Another randomized, double-blinded, placebo-controlled trial, the Women's Antioxidant and Folic Acid Cardiovascular Study (WA-FACS), did not find that supplementation with a combination of folic acid, vitamin B 6 , and vitamin B 12 has a protective effect against total cardiovascular events among high-risk women despite significantly lowering homocysteine levels [39]. Differences in study design and population, sample size, CVD outcomes, covariate adjustments, and adherence to treatment may be possible reasons for between-study heterogeneity. In our casecontrol study, although the patients with T2D in the highest quintile of folate and vitamin B 6 intake had approximately 60% and 50% lower risks of CVD than those in the lowest quartile, the intake of vitamin B 12 was not related to the risk of CVD. Our results were consistent Statistically significant results are bolded. Numbers, odds ratios, and 95% confidence intervals. a Quartile cut-off points were based on residual energy-adjusted intake among the control subjects. b Crude and adjusted ORs (95% CI) from the unconditional logistic model. Covariates include age, BMI, T2D duration, physical activity, educational level, household income, occupation, work status, marital status, smoking status, alcohol consumption, tea drinker, hypertension status, dyslipidaemia status, antidiabetic medication using status, antihypertensive drugs using status, lipid-lowering drugs using status, multivitamin supplement using status, and daily energy intake (log transformed). with the findings of the HOPE 2 study [38] but were different from the results of the WAFACS study [39], showing that high intakes of folate and vitamin B 6 can reduce the risk of CVD in patients with T2D. However, given the scarcity of available evidence, further research will still be necessary to detect the effects of B-vitamins on cardiovascular outcomes.
The beneficial effects of folate and vitamin B 6 intake on the risk of CVD remain speculative but biologically plausible because they are essential nutrients involved in the one-carbon metabolism pathway, which plays an important role in homocysteine metabolism [4,6,41]. Several previous interventional studies have provided convincing evidence that supplementation with B-vitamins significantly reduces the blood levels of homocysteine [8][9][10][11][12]. A meta-analysis of observational studies demonstrated that a 3 μmol/L reduction in serum homocysteine concentrations is associated with an 11% (OR 0.89; 95% CI: 0.83, 0.96) reduction in ischaemic heart disease risk and a 19% (OR 0.81; 95% CI: 0.69, 0.95) reduction in stroke risk [42]. Therefore, the protective role of folate and vitamin B 6 in CVD possibly contributes to the effects on homocysteine status [43]. In one aspect, homocysteine significantly induces the mRNA and protein expression of C-reactive protein in vascular smooth muscle cells, which initiates an inflammatory response [44]. In another aspect, the role of homocysteine in endothelial dysfunction is thought to be mediated by common and plausible mechanisms, including nuclear factor-κb activation, inflammation, oxidative stress, and the inhibition of endothelial nitric oxide synthase, that promote atherosclerosis and vascular disease [45]. Furthermore, increased homocysteine levels promote platelet adhesion to endothelial cells and have also been associated with high levels of prothrombotic factors, which lead to the augmentation of thrombus formation [46]. In addition to exerting homocysteine-lowering effects, folate and vitamin B 6 may lower the risk of CVD via other anti-atherosclerotic properties, such as improving insulin and lipid metabolism [47] and preventing oxygen radical generation and lipid peroxidation [48].
In our study, the main food sources of folate were eggs (12.5%) and pakchoi, rape, and lettuce (12.4%), and vita- min B 6 was mainly derived from lean pork (20.1%) and Chinese cabbage, mustard greens, and broccoli (10.8%). Pakchoi, rape, lettuce, Chinese cabbage, mustard greens, and broccoli are all dark green leafy vegetables, and increasing the intake of dark green vegetables is associated with a reduced risk of CVD as observed in similar studies [49,50]. However, the meta-analyses of prospective studies on lean pork consumption and CVD risk did not provide consistent results [51,52] likely because the advantageous effects of folate and vitamin B 6 derived from lean pork are offset by the significantly high cholesterol and saturated fatty acid contents, which have been shown to increase CVD, of pork [53].
Considering that cigarettes and alcohol are important risk factors for the aetiology of CVD, we adjusted for smoking status (never intake, former intake, or current intake) and alcohol consumption (never intake, former intake, or current intake) in the multivariable analysis. In alcohol and cigarette consumption-stratified analyses, the inverse association between dietary folate and cardiovascular events remained significant in non-drinkers and non-smokers but not in current drinkers and current smokers, and the interaction was not statistically significant. However, our results were different from the protective effect reported by the latest meta-analysis [19]. Considering the number of participants who currently drank alcohol or smoked, this result may be attributed to the low statistical power of these subgroups.
Vitamin B 12 plays an important role as a cofactor in onecarbon metabolism. However, consistent with the findings of some studies, a null association was observed between vitamin B 12 intake and CVD risk in our study [19,54]. One possible explanation is that vitamin B 12 has a relatively moderate effect on homocysteine levels. Previous studies have shown that supplementation with folic acid reduces homocysteine by 13-25%, and the addition of vitamin B 12 produced only a further reduction of 7% [55]. Another possible interpretation is that the lower intake of vitamin B 12 (1.40 μg/day) than that of other nutrients (e.g., 160.19 μg/ day folate and 0.74 mg/day vitamin B 6 ) results in a large random measurement error and greater variation for vitamin B 12 and may attenuate the associations to null [56].
The strength of our study is its good matching by age and sex between the patients with CVD and the control subjects, thus increasing its statistical power. Additionally, the use of a validated and reproducible FFQ, the assessment of portion size by means of visual aids, the use of trained interviewers. and the equal number of case and control interviews conducted by the individual investigators can reduce information bias.
Several limitations of this study should be noted. Firstly, our study cannot rule out the possibility that disease affected the participants' dietary habits, and reverse causality exists inevitably in the case-control design. We recruited all cases that were interviewed within 2 weeks of diagnosis and excluded all participants from our study who had substantially changed their dietary habits during the prior year to minimize this possibility. Secondly, although we did adjust for multiple potential confounding factors, the residual confounding effects of unmeasured or unknown variables may have affected the results. Thirdly, we could not eliminate potential selection bias due to the hospital-based case-control study design. The dietary habits and sociodemographic characteristics of cases and controls recruited from the studied hospitals might differ from those of patients admitted to other hospitals or the general population. Fourthly, given that we could not obtain data on plasma homocysteine levels, the interactions among one-carbon metabolism-related Bvitamin intake, homocysteine levels, and CVD could not be evaluated. Fifthly, avoiding recall bias in dietary assessment by using a FFQ in a case-control study is difficult. Finally, given that the participants of this study were all residents of the Guangdong area in South China, we caution against generalizing the results to other populations.

Conclusion
This case-control study demonstrates that high intakes of folate and vitamin B 6 , but not of vitamin B 12 , are associated with a low risk of CVD among T2D patients. These results provide preliminary evidence for the feasibility of a dietary programme for the prevention of CVD in patients with T2D. However, such evidence should be confirmed by further studies.