| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
ORIGINAL RESEARCH COMMUNICATION |
1 From the Wageningen Centre for Food Sciences, Wageningen, Netherlands (RMW, IAB, MBK, and PV), and the Division of Human Nutrition, Wageningen University, Wageningen, Netherlands (RMW, IAB, ES, MBK, and PV)
2 Supported by the Wageningen Centre for Food Sciences [an alliance of major Dutch food industries, the Nederlands Instituut voor Zuivelonderzoek (NIZO), the Nederlandes Organisatie voor toegepast-natuurwetenschappelijk onderzoek (TNO), Maastricht University, Wageningen University, and the Dutch Government].
3 Reprints not available. Address correspondence to R Winkels, Wageningen University, Division of Human Nutrition, Bomenweg 2, 6703 HD Wageningen, Netherlands. E-mail: renate.winkels{at}wur.nl.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Objective: In a 4-wk dietary intervention study, we determined the aggregate bioavailability of food folates from fruit, vegetables, and liver relative to that of folic acid.
Design: Seventy-two healthy adults were randomly divided into 4 treatment groups. Group A (n = 29) received a high-folate diet with 369 µg food folate/d and a placebo capsule; groups B, C, and D (n = 14 or 15) received a low-folate diet with 73 µg food folate/d and folic acid capsules. These capsules contained 92 µg folic acid/d for group B, 191 µg for group C, and 289 µg for group D. In addition, all 72 subjects daily ingested a capsule with 58 µg [13C11]-labeled folic acid. We measured the percentage of [13C11]-labeled folate in plasma folate at the end of the intervention and ascertained the changes in serum folate concentrations over the 4 wk of the intervention.
Results: Bioavailability of food folate relative to that of folic acid was 78% (95% CI: 48%, 108%) according to [13C11]-labeled folate and 85% (52%, 118%) according to changes in serum folate concentrations.
Conclusions: The aggregate bioavailability of folates from fruit, vegetables, and liver is 
80% of that of folic acid. The consumption of a diet rich in food folate can improve the folate status of a population more efficiently than is generally assumed.
Key Words: Food folate • bioavailability • folic acid • stable isotopes
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
An inadequate folate intake will lead to a decrease in serum and erythrocyte folate, to increased blood concentrations of homocysteine, and ultimately to macrocytic anemia (5). It may also increase the risk of neurocognitive disease, cancer, and cardiovascular disease (6-9). High intake of folic acid by women of childbearing age reduces the risk of neural tube defects in the infants of these women (10).
Folate bioavailability is defined as the proportion of an ingested amount of folate that is absorbed in the gut and that becomes available for metabolic processes. In human intervention studies, relative bioavailability is usually assessed by comparison with a reference dose of folic acid. The bioavailability of food folate is generally lower than that of folic acid, but the extent of the difference is unclear (11, 12).
Recommended daily allowances for food folate take into account its lower bioavailability. The basis for adapting the US recommendations for folate to the lower bioavailability of food folate (5) was derived from a study by Sauberlich et all (13), whic stated that the bioavailability of food folate was no more than 50% of that of folic acid. Unfortunately, they did not indicate how that finding was obtained.
Other long-term dietary intervention studies found bioavailability estimates between 30% and 98% of those of folic acid (14, 15). Thus, no clear figure exists for the bioavailability of natural folates as a proportion of that of folic acid. We therefore performed a large, long-term dietary intervention study in which we compared the bioavailability of food folate with that of different doses of folic acid; we simultaneously assessed bioavailability from changes in serum folate concentrations and from stable isotope dilution.
|
SUBJECTS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
T genotype (17). We excluded 18 subjects on the basis of the following criteria: use of B vitamins within the period 3 mo before the study; body mass index (BMI; in kg/m2) <18 or >30; use of drugs that interfere with folate metabolism; serum vitamin B-12 concentrations <119 pmol/L; serum creatinine concentrations >125 µmol/L; plasma total homocysteine concentrations >26 µmol/L; and the presence of cardiovascular disease, cancer, rheumatoid arthritis, epilepsy, or gastrointestinal disorders.
The remaining 75 subjects were stratified by serum folate, MTHFR 677CT genotype, and sex. Within each stratum, randomized blocks assigned the subjects to 1 of the 4 intervention groups.
All subjects gave written informed consent. The Medical Ethical Committee of Wageningen University approved the study.
Design
The trial included 4 parallel intervention groups, as shown in Figure 1
. Group A (the food folate group) consumed a diet high in food folate and a placebo capsule. Groups B, C, and D (the folic acid groups) all received a diet low in food folate plus a daily capsule, which contained 92, 191, and 289 µg folic acid, respectively. In addition, all 75 subjects received a capsule with 58 µg [13C11]-labeled folic acid/d. The folic acid groups (B through D) served to calibrate our outcome measurements—namely. the percentage of [13C11]-labeled folate in the total pool of plasma folate and the change in serum folate in terms of folic acid intake. We constructed calibration lines of the percentage of [13C11]-labeled folate in the total pool of plasma folate and of the change in serum folate as a function of folic acid intake in groups B through D. We then projected the mean percentage of labeled folate in plasma folate in group A and the mean change in serum folate in group A on the corresponding calibration line to assess the dose of folic acid to which the food folate in group A was equivalent. The study lasted 31 d; the folate intervention started on day 3 and ended on day 31. Two fasting blood samples were drawn at baseline (days 1 and 3) and after the 4-wk intervention (on days 29 and 31).
|
90% of their total daily energy requirement during the intervention. Subjects daily had a limited free choice from a list of products that provided the remaining 10% of energy. These free-choice items were mainly nonalcoholic drinks, alcoholic drinks (
1 beer/d), candies, and sweet sandwich fillings. The items contained a low amount of folate and fat. Subjects kept a diary in which they recorded which free-choice items they consumed, whether they took their capsules, illnesses, use of medication, and any deviations from the diet.
The diets we supplied consisted of constant food items and varying food items. The constant part of the diet consisted of whole-wheat bread; margarine; sweet sandwich fillings; cheese, ham, or both; cookies; milk; boiled potatoes, rice, or pasta; and meat or a meat replacement. These constant food items provided 9 MJ/d (2100 kcal/d) for a typical participant and were similar for all groups. The varying food items provided 2 MJ/d (480 kcal/d) and differed between group A and groups B through D (Table 1).
|
100 µg folate/9 MJ constant foods—as calculated from food tables (19).
The varying food items for group A were products rich in food folate—350 µg per 2 MJ of varying foods—as calculated from food tables; for groups B through D these products were replaced by similar products low in food folate—30 µg per 2 MJ—as calculated from food tables (Table 1
). Vegetables, fruit juices, liver paste, and fruit contributed most to the folate intake in the high-folate group (Figure 2). The varying part of the diet was nearly the same for all 29 subjects in group A, independent of the subject’s energy requirement; only the amounts of dressing, sauce, and dessert were adjusted to individual energy requirements (Table 1). Likewise, the varying part of the diet for groups B through D was nearly the same for all 43 subjects in those groups.
|
Meals were prepared from conventional foods and drinks. Folic acid fortification is not mandatory in the Netherlands, and the foods and beverages that we used did not contain added folic acid.
Measurement of nutrients in food samples
We collected a total of 4 duplicate diets on each day of the trial: 2 duplicate diets representative of group A and 2 representative of groups B through D; the energy content of these duplicates was 11 MJ (2600 kcal) according to food tables. On weekdays, we collected the duplicates every day at lunchtime. We also collected a duplicate of a package with foods and beverages for the weekend. This was prepared for consumption during the actual weekend and then worked up as described below. We collected, pooled, and analyzed the items of the constant part of the diet separately from the varying food items. For the food folate analyses, hot items were cooled to 10 °C immediately after collection in a blast chiller before they were added to the cold items. We added 250 mL of chilled CHES/HEPES buffer (pH 7.85) with 2% ascorbic acid and 0.2 mol 2-mercaptoethanol/L per kg of food, homogenized it in a blender under a flow of nitrogen, and stored the samples at –80 °C until they were analyzed.
Food folate was analyzed in a selection of the duplicate diets: we selected one random Monday duplicate out of the 4 Mondays of the study of group A and analyzed food folate in both the constant and the varying food items. The same step was taken for all other weekdays. We repeated this selection for diets from groups B through D. In this way, we avoided the thawing and homogenization that would have been necessary if we had pooled the diets of all 4 Mondays of the study. We decided not to analyze folate in all samples, because food folate analysis is time- and labor-intensive. Food folate was analyzed with an HPLC method (20). In short, duplicate diets were thawed, extracted with CHES/HEPES buffer in a boiling water bath for 10 min, and cooled in a water bath at 0 °C. Samples were subjected to trienzyme treatment, purified on an affinity column (Folate Binding Protein; Scripps, San Diego, CA), and analyzed by HPLC with fluorescence and ultraviolet detection. All analyses were performed in duplicate by splitting the sample after it had been thawed.
Duplicate diets for the analyses of energy in the diets were stored at –20 °C. After the study, we thawed these duplicates, pooled the constant parts of each diet per week, pooled the varying food items of each diet per week, and homogenized the pooled items. Samples were stored at –20 °C until they were analyzed. Total fat, protein, dry matter, ash, and fiber were analyzed in these duplicates. From these analyses, we calculated the amount of fat, protein, carbohydrates, fiber, and energy in the diets actually eaten by the subjects.
Capsules
The capsules with folic acid (Merck Eprova, Schaffhausen, Switzerland) were specially produced for this study. We ordered capsules containing 100, 200, and 300 µg. To analyze the actual amount of folic acid, we dissolved capsules in CHES/HEPES buffer pH 7.85 with 2% ascorbic acid and 0.2 mol 2-mercaptoethanol/L in a boiling water bath and measured folic acid by using HPLC with ultraviolet detection (20). The amount of folic acid in a random sample of 20 capsules of each dose was 95% of the expected dose.
[13C11]-Folic acid was synthesized for us by ARC (Apeldoorn, Netherlands). Folic acid was labeled with 6 13C atoms in the benzoic acid structure and with 5 13C atoms in the glutamate part (21). These capsules contained a mean (±SD) of 58 ± 11 µg folic acid (n = 20) according to HPLC analysis (20), which was 85% of the expected dose.
An independent research assistant from Wageningen University coded the placebo and folic acid capsules: subjects and investigators were blinded to the folic acid treatment. The assistant did not reveal the code until all data had been gathered and double-checked. Subjects ingested their capsules just before the hot lunch.
Measurements in blood
For the measurement of serum folate, blood samples were drawn into coagulation tubes, stored in the dark for 30–120 min at room temperature, and centrifuged at 2600 x g for 10 min at 4 °C. Serum was pipetted off and stored at –80 °C. We measured the serum folate concentration with a chemiluminescence immunoassay (Access Immunoassay; Beckman Coulter) in the samples for days 1, 3, 29, and 31. Baseline values represent the mean of days 1 and 3, and week 4 values represent the mean of days 29 and 31. Intraassay and interassay CVs of this immunoassay were <15%.
For the measurement of labeled folate, blood samples were drawn into EDTA-containing tubes that were immediately placed on ice and centrifuged within 30 min at 2600 x g for 10 min at 4 °C. Plasma was pipetted off and stored at –80 °C. We measured the percentage of labeled folate in the total pool of folate in plasma with liquid chromatography–tandem mass spectrometry [(LC-MS/MS) 21], as in the following equation:
 | (1) |
Labeled folate was analyzed only in the samples for days 1 and 29. We chose to analyze labeled folate in only one baseline sample and one follow-up sample because we expected that the estimate for bioavailability derived from labeled folate data would be more precise than the estimate derived from serum folate data, even if analyzed only in one sample. The labeled compound does not occur naturally and was not detected in baseline samples. We restricted our measurements to the 5-methyltetrahydrofolate fraction of the plasma, because this is the most abundant folate vitamer in plasma (22).
Calculation of bioavailability
We plotted the individual percentages of labeled folate in the plasma folate of subjects in groups B, C, and D against the intake of supplemental folic acid. We fitted a linear regression line through these points to construct a calibration line of percentage of labeled folate against the intake of supplemental folic acid. The mean percentage of labeled folate in group A was then projected on this calibration line to assess by interpolation the dose of folic acid to which the additional amount of food folate in group A (folateadditional) was equivalent (Figure 3); this dose was called Xa. Relative bioavailability of food folate was derived from Xa and folateadditional according to the following equation:
 | (2) |
 | (3) |
 | (4) |
 | (5) |
= the mean intake of folic acid in groups B, C, and D in µg; and Xi = the individual intake of folic acid of subjects in groups B, C and D in µg.
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The diaries of the subjects did not show any deviations from the provided diets that may have affected the results. Capsule intake was verified by counting the returned capsules and by checking the subjects’ diaries: mean capsule intake was 99%, and the lowest intake was 90%. Characteristics measured at screening did not differ significantly between groups (Table 2). The mean intakes of energy protein, fat, carbohydrates, and fiber during the trial did not differ significantly between groups (Table 3). According to analysis of duplicate diets, the food folate intake in group A was 369 µg/d, and that in groups B through D was 73 µg/d (Figure 1); thus group A had 296 µg/d more food folate than did groups B through D. Homocysteine concentrations decreased slightly with increasing intakes of folic acid (Table 4).
|
|
|
). The calibration line was calculated with the use of the following equation:
 | (6) |
). This value was entered into the equation of the calibration line (Figure 3), and it showed that the extra 296 µg food folate ingested by group A subjects was equivalent to 232 µg folic acid. Therefore, the relative bioavailability of food folate calculated from labeled folate data were—ie, (232/296) x 100—was 78% (95% CI: 48%, 108%).
Serum folate concentrations increased linearly with increasing intakes of folic acid in groups B through D (Figure 4). The calibration line was calculated with the use of the following equation:
 | (7) |
). This value was entered into the equation of the calibration line (Figure 4), and it showed that the extra 296 µg food folate ingested by group A subjects was equivalent to 252 µg folic acid. Therefore, the relative bioavailability of food folate calculated from serum folate data—ie, (252/296) x 100—was 85% (95% CI: 52%, 118%).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
In the current trial, we carefully controlled the intakes of food folate and folic acid. A problem in studies that aim to compare the bioavailability of food folate and that of folic acid has been the fact that the actual intake of food folate may differ from the targeted dose, because the stability of food folates during cooking is poor and the compliance of subjects with the intervention diet can be low (12). These factors will lead to an underestimation of the bioavailability of folate from foods. In contrast to food folate, folic acid is very stable, and compliance with taking a pill is likely to be higher than compliance with a prescribed diet. We strictly controlled the intake of both food folate and folic acid in our subjects: the subjects came to our laboratory daily during weekdays to consume a hot lunch plus supplements, and thus a large part of the daily intake of folate and folic acid was supervised. Furthermore, we provided 90% of all the foods and beverages for consumption off-site and asked subjects to note in a diary whether they deviated from these supplies. In addition, we based our estimates of bioavailability on the analyzed amounts of folic acid and folate in the capsules and in the duplicates of the diets as actually eaten.
We measured 2 markers for folate bioavailability: the percentage of labeled folate in the plasma folate and changes in serum folate. We expected that the most precise estimate for bioavailability would result from the labeled folate data. Because the labeled compound does not occur naturally, the percentage of labeled folate in plasma had to be evaluated only after the intervention: this approach eliminated the extra error in subtracting 2 measurements from each other. However, the precision was similar for labeled folate and serum folate: the CIs surrounding both estimates for bioavailability had a similar width. For serum folate, the estimate for bioavailability and the width of the CI did not change when we based our calculations only on the measurements in samples from day 1 and day 29 (data not shown). Therefore, in this experimental set-up, the expensive stable-isotope method yielded no advantage over serum folate measurements. However, the fact that the 2 methods yielded similar numbers emphasizes the internal consistency of the study and strengthens the confidence in our outcome. We used folic acid labeled with 11 13C atoms in the current intervention, because we had it in stock from earlier experiments. Using cheaper forms of folic acid would not have made the intervention much cheaper 3because the LC-MS/MS analysis is much more expensive than is the use of the labeled compound itself.
The participants ingested the folic acid capsules just before their hot lunch, which may have affected the serum folate responses in groups B through D and, hence, the estimate for relative bioavailability. Pfeiffer et al (24)found that the serum folate response to a folic acid supplement taken together with a light meal was 15% lower than the folate response to a folic acid supplement taken on an empty stomach. The difference was not significant, but if it were valid, then the calibration line as plotted in Figure 4 would have been steeper and the bioavailability estimate for food folates would have been lower if subjects had ingested the capsules on an empty stomach. We plotted a linear calibration line in Figure 2. However, a lack of fit test showed that the linear model was not the best model with which to describe our data: the addition of a second-order term to the model improved the R2 and the model’s fit to the data (data not shown). Estimation of bioavailability from this second-order regression line produced a slightly higher value for bioavailability of food folate (89%) than did the linear model (85%). This finding reinforces our conclusion that the bioavailability of folate is higher than previously reported. We decided to use a linear regression model to fit our data, because other, larger studies (25, 26) found that the relation between the intake of folic acid and the change in serum folate is linear over the range of intakes that we used.
HPLC analysis of folate in food generally results in lower values than does microbiological analysis (1, 27), and therefore the bioavailability figures for food folate generally will be higher when based on HPLC analysis. Previous bioavailability studies (13-15) analyzed food folate with microbiological assays. Therefore, we also analyzed our food samples microbiologically (28), and that approach yielded a food folate intake in group A of 474 µg/d and 136 µg/d in groups B through D; the food folate in group A was therefore 338 µg/d greater than that in groups B through D. This finding resulted in a bioavailability of (232/338) x 100 = 68% (95% CI: 42%, 95%) according to labeled folate data or (252/338) x 100 = 75% (95% CI: 45%, 103%) according to changes in serum folate. Thus, the estimate for bioavailability was 10% lower with microbiologically derived figures than when based on HPLC analysis.
Our findings disagree with a statement by Sauberlich et al (13), who said that "when compared with synthetic PGA (pteroyl glutamic acid), dietary folates appeared to be no more than 50% available." Unfortunately, those authors did not indicate how this finding was obtained. In our opinion, the less-than-suitable design and small size of this study—with 3–4 subjects per treatment group—and the absence of a folic acid group make it impossible to compare the bioavailability of natural folates with that of folic acid from these data. Our results also differ from those of Hannon-Fletcher et al (15). In that study, subjects consumed a folate-depleted carrier meal to which food folates extracted from spinach or from yeast were added or they consumed the carrier meal together with folic acid in a tablet; meals were provided daily for 30 d. On the basis of changes in serum folate, the bioavailability of spinach folate and yeast folate was 36% and 62%, respectively, of that of folic acid. However, the addition of these 2 sources of folate to meals is not representative for folates from whole diets: in whole diets, folates originate from various sources and matrixes of foods. Furthermore food intake was not controlled. These factors limit the usefulness of the bioavailability findings in that study. The findings in the current study are, however, in excellent agreement with those of Brouwer et al (14), who conducted a 4-wk, highly controlled intervention study and found that bioavailability of food folate from fruit and vegetables was 78% of that of folic acid, according to changes in plasma folate.
We composed a diet high in folate by selecting fruit and vegetables rich in folate, by providing liver paste as sandwich filling, and by adding egg yolk to sauces and salad dressings. Fruit and vegetables were the main source, providing 73% of the additional folate in group A (Figure 2). Besides fruit and vegetables, unfortified bread and cereals are important sources of folate in the general population (1, 2). However, it was not feasible to include these foods in the varying food items of group A, because we could not replace them with similar products low in folate for groups B through D. Bioavailability of folates from food is influenced by a number of food-related factors, eg, the species of folate in the food, the number of glutamate residues attached to the folate molecule, and the food matrix (29). Because cereals and fruit and vegetables are all plant foods, similarities in factors that affect bioavailability for these food groups are likely to exist. Therefore, we speculate that our findings will also be applicable for diets in which cereals are an important folate source, but this possibility requires confirmation.
Our findings and those of Brouwer et al (14) indicate that bioavailability of food folates is higher than generally assumed—namely, for folate as measured by HPLC, it is 80% of that of folic acid from capsules taken with a meal. Subjects in group A consumed a total of 369 µg folate/d from foods, which is equivalent to 295 µg folic acid. Their diet did include 25 g liver paste/d and 400 mL orange juice/d, which may not be to everyone’s liking. Nevertheless, we think it is quite feasible for a healthy varied diet to provide the equivalent of 300 µg folic acid/d. The value of 50% bioavailability for folates from food, as used in the construction of recommended daily allowances (5), underestimates the bioavailability of food folates. The use of this low number could lead to skepticism about the importance of a healthy diet as a source of folate: our data suggest that such skepticism may be unwarranted. Our data show that consumption of a diet rich in folate can improve the folate status of a population more efficiently than is generally assumed.
|
ACKNOWLEDGMENTS |
|---|
All authors were involved in the design of the study; RMW wrote the manuscript, and IAB, ES, MK, and PV reviewed and edited the manuscript; ES designed the diets; and IAB, ES, and RMW contributed to the performance of the study. None of the authors had a personal or financial conflict of interest.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
, group B; 
, group C; x, group D; —, the linear regression line through the individual data points of groups B, C, and D. The mean percentage of labeled folate in group A was 9.7%. This was projected on the regression line and corresponded to an estimated intake of folic acid of 232 µg (broken lines). The R2 of the regression line was 0.23, and the slope was –0.01616 (95% CI: –0.02550, –0.00683).