Sabine
Mönch†
a,
Michael
Netzel‡
b,
Gabriele
Netzel§
b,
Undine
Ott
c,
Thomas
Frank
d and
Michael
Rychlik
*ef
aLehrstuhl für Lebensmittelchemie, Technische Universität München, Lise-Meitner-Str. 34, D-85350 Freising, Germany
bDepartment of Human Nutrition, Institute of Nutrition, Friedrich Schiller Universität Jena, Dornburgerstr. 29, D-07743 Jena, Germany
cKuratorium für Dialyse und Nierentransplantation e.V., Zur Lämmerlaide 1, D-07751 Jena, Germany
dPrivate Consultant, D-65812 Bad Soden, Germany
eChair of Analytical Food Chemistry, Technische Universität München, Alte Akademie 10, D-85350 Freising, Germany. E-mail: michael.rychlik@tum.de; Fax: +49-8161-71 4216; Tel: +49-8161-71 3153
fBIOANALYTIK Weihenstephan, Research Center for Nutrition and Food Sciences, Technische Universität München, Alte Akademie 10, D-85350 Freising, Germany
First published on 3rd November 2014
Different sources of folate may have different bioavailability and hence may impact the standard definition of folate equivalents. In order to examine this, a short term human study was undertaken to evaluate the relative native folate bioavailabilities from spinach, Camembert cheese and wheat germs compared to pteroylmonoglutamic acid as the reference dose. The study had a single-centre, randomised, four-treatment, four-period, four-sequence, cross-over design, i.e. the four (food) items to be tested (referred to as treatments) were administered in sequences according to the Latin square, so that each experimental treatment occurred only once within each sequence and once within each study period. Each of the 24 subjects received the four experimental items separated by a 14-day equilibrium phase and received a pteroylmonoglutamic acid supplement for 14 days before the first testing and between the testings for saturation of body pools. Folates in test foods, plasma and urine samples were determined by stable isotope dilution assays, and in urine and plasma, the concentrations of 5-methyltetrahydrofolate were evaluated. Standard non-compartmental methods were applied to determine the biokinetic parameters Cmax, tmax and AUC from baseline corrected 5-methyltetrahydrofolate concentrations within the interval from 0 to 12 hours. The variability of AUC and Cmax was moderate for spinach and oral solution of pteroylmonoglutamic acid but high for Camembert cheese and very high for wheat germs. The median tmax was lowest for spinach, though tmax showed a high variability among all treatments. When comparing the ratio estimates of AUC and Cmax for the different test foods, highest bioavailability was found for spinach followed by that for wheat germs and Camembert cheese. The results underline the dependence of folate bioavailability on the type of food ingested. Therefore, the general assumption of 50% bioavailability as the rationale behind the definition of folate equivalents has to be questioned and requires further investigation.
The current dietary recommendations are based on the studies of Sauberlich et al.,13 who determined in a long-term study a 50% bioavailability of food folates relative to pteroylmonoglutamic acid. However, this generalization has been questioned because of recent human studies such as the short-term study performed by Prinz-Langenohl et al.,14 who determined a folate bioavailability of spinach ranging between 89–113% relative to pteroylmonoglutamic acid. Moreover, in a long-term study Brouwer et al.15 found a 98% folate bioavailability for citrus fruits and vegetables relative to pteroylmonoglutamic acid. This finding may also be due to enhanced stability of 5-methyltetrahydrofolic acid (5-CH3-H4folate) in concurrent presence of ascorbic acid.16 Even when 5-CH3-H4folate was used as the reference dose, bioavailabilities ranging between 99–120% for broccoli and strawberries were found by Witthöft et al.17
In preparation of this investigation, we performed a pilot study on folate bioavailability by using stable isotope dilution assays for analysis of plasma folates and an area under the curve (AUC) approach.18 However, the plasma monitoring time of 6 hours after the intake of a pteroylmonoglutamic acid supplement as the reference dose was found to be too short, as the plasma level did not return to the baseline. Additionally, the analytical tools for this model have recently been improved and were extended to the analysis of folates in urine and erythrocytes.19 5-Methyltetrahydrofolate (5-CH3-H4folate) is the folate derivative normally found in the circulation, and in addition, is the predominant type of folate present in food. However, in case of fortification or when supplements of pteroylmonoglutamic acid are used, non-metabolised pteroylmonoglutamic acid may be found in blood circulation.20
Based on these considerations and preliminary studies, the aim of the present study was to assess in a short-term human study the relative bioavailability of folates in several foods rich in folates by recording 5-CH3-H4folate levels in plasma post-dose for 12 hours. As the test foods, spinach, wheat germs, and a low-fat Camembert cheese were chosen. Moreover, the suitability of analysing folate levels in urine for assessing bioavailability was evaluated.
[2H4]-5-Methyltetrahydrofolic acid, [2H4]-5-formyltetrahydrofolic acid, [2H4]-tetrahydrofolic acid, [2H4]-10-formylfolic acid and [2H4]-pteroylmonoglutamic acid were synthesised as reported recently.21
Ammonium formate buffer consisted of ammonium formate (10 g L−1) and ascorbic acid (1 g L−1) adjusted to pH 3.2.
Eluting solution for SPE was a mixture of aqueous sodium chloride (5%) and aqueous sodium acetate (100 mmol L−1) containing ascorbic acid (1%).
Before inclusion, the subjects underwent a screening evaluation regarding their medical history. Participants adhered to their usual diet, but they received a vitamin supplement with 800 μg of pteroylmonoglutamic acid for 14 days before the first testing and between the testings, which was discontinued two days prior to the start of the study. This “saturation” was done to improve uniformity among subjects and subsequently the precision of bioavailability estimates.24,25
The study had a single-centre, randomised, four-treatment, four-period, cross-over design. There were four treatment sequences in accordance with the Latin square, so that each experimental treatment occurred only once within each sequence and once within each period. Each subject had the following four experimental treatments separated by a 14-days equilibrium phase: 1294 nmol sum of folates via Camembert cheese (200 g), 534 nmol sum of folates via wheat germs (50 g), 1185 nmol sum of folates via heated spinach (500 g), and 852 nmol pteroylmonoglutamic acid via orally administered pteroylmonoglutamic acid solution (95 mL) serving as reference treatment. The order in which the treatments were given was randomised.
Between 8:00 and 9:00 a.m., after an overnight fast, volunteers took the test meals or drank the test solution, respectively, together with one slice of toast bread. During the experimental treatment periods (24 hours), the consumption of water was allowed ad libitum, and two further standardised and virtually folate-free meals consisting of wheat bread (9 slices per 500 g), butter (100 g), honey (250 g), apple sauce (355 g), and apricot jam (225 g) were offered for lunch and dinner. All food items were common brands and purchased at local supermarkets in the City of Jena, Germany.
Between the test periods, i.e., during the equilibrium phase, the participants were instructed to take the folate supplementation as mentioned above while keeping their normal dietary habits unchanged. For the determination of the biokinetic profile of 5-CH3-H4folate in plasma, venous blood samples were drawn predose, as well as 1, 2, 3, 4, 5, 6, 8, 10, 12 and 24 hours after the administration of the dose. Each blood sample (9 mL) was collected in an EDTA-coated tube (Sarstedt, Nuernbrecht, Germany). Plasma and red blood cells were obtained by centrifuging the blood for 10 min at 2000g and 4 °C. In addition, the volunteers collected the complete postdose urine for 24 h into 2 L opaque brown urine containers which were stored refrigerated during the collection periods. Plasma, red blood cells and urine samples were stored frozen at −24 °C until further preparation and analysis.
Standard noncompartmental methods were applied to determine the biokinetic parameters:26Cmax (observed maximum concentration), tmax (time of Cmax), AUC from baseline corrected 5-CH3-H4folate concentrations (limited within the interval from 0 to 12 h). The range of biokinetic evaluation was limited to 12 hours postdose, because it became obvious during data review that 5-CH3-H4folate concentrations increased from 12 to 24 h post-dose (see also Fig. 1). The AUC was calculated according to the linear trapezoidal rule. The amount of 5-CH3-H4folate excreted into urine from time zero up to 24 h (Ae0–24) was determined by multiplying the 5-CH3-H4folate concentration with the volume of the 24 h urine sample. The fraction of orally administered folate excreted into urine (‘%Excretion’) was calculated by dividing Ae0–24 through the respective dose administered. Concentrations below the limit of quantification (LOQ) were set to zero.
The primary biokinetic parameters for inferential statistics were Cmax and AUC after logarithmic data transformation. Prior to logarithmic transformation, the Cmax and AUC values were normalised to dose (i.e., assuming dose-proportionality) since no equimolar doses were administered. The data were analyzed with a linear mixed effects model with fixed terms for treatment, period, sequence and sex, and random term for subject within sequence-by-sex:
Log(Parameter) = Sequence + Subject(Sequence × Sex) + Period + Treatment + Sex + Error, |
For Cmax and AUC, estimate and 90% confidence interval (CI) for the ratio of treatment means (test/reference) were obtained by computing estimate and 90% CI for the contrast giving the difference between treatment means within the linear mixed effects model framework, and then converting to ratio of geometric means by the antilog transformation. Equivalence was concluded if the 90% CI for the ratio was entirely within the 0.80 to 1.25 equivalence reference interval.
The secondary PK parameter was ‘%Excretion’. It was subjected to the same linear mixed effects model analysis as the primary PK parameters.
The level of statistical significance was fixed at p < 0.05. No adjustment of the alpha-level was made for multiple analyses.
Food | Tetrahydrofolate (μg per 100 g) | 5-Methyl-tetrahydrofolate (μg per 100 g) | 5-Formyl-tetrahydrofolate (μg per 100 g) | 10-Formylfolate (μg per 100 g) | Pteroylmonoglutamic acid (μg per 100 g) | Sum of folates (nmol per 100 g) |
---|---|---|---|---|---|---|
a n.d. not detectable. | ||||||
Spinach | 10.3 | 72.2 | 6.6 | 15.6 | n.d. | 237 |
Camembert cheese | 144.7 | 46.2 | 54.5 | 40.2 | n.d. | 647 |
Wheat germs | 36.0 | 30.6 | 313.6 | 65.2 | 25.8 | 1068 |
After administration of the test foods, the mean plasma concentrations of 5-CH3-H4folate were determined as displayed in Fig. 1. It is worth noting that the widely varying, mean plasma concentrations partially rose again after 4–6 h post-dose. However, this increase is not attributable to the intermediate consumption of the low-folate lunch, which was provided after the 4 hour blood sample was drawn, as it contained less than 5% of the folate dosage of the treatments. The low-folate dinner was provided after the 10 hour blood sample has been drawn. Apart from 5-CH3-H4folate, further folate vitamers in plasma were not considered as they occurred only intermittent in traces and always below their LOQ.
Table 2 summarises the biokinetic parameters of baseline corrected 5-CH3-H4folate. The variability of AUC and Cmax was moderate for spinach and oral solution of pteroylmonoglutamic acid (CV% 30–60%), but high for Camembert cheese (CV% 60–90%) and very high for wheat germ (CV% >90%). The time to attain the maximum concentrations (tmax) was highly variable among treatments. However, the median of tmax was lower by trend for spinach than for the pteroylmonoglutamic acid solution. This finding is in line with recent findings from a double-label ileostomy study which showed a lower tmax for labelled 5-CH3-H4folate than for labelled pteroylmonoglutamic acid.29
Parameterb | Camembert | Wheat germs | Spinach | Oral solution |
---|---|---|---|---|
a Tabulated values are arithmetic mean ± SD (CV%) [geometric mean] of n = 24 subjects except for tmax where values are median (min, max). b AUC, Cmax and tmax were determined/calculated within the 0 to 12 hour interval. c AUC is the positive AUC, i.e., concentrations falling below the individual predose values were discarded. | ||||
C max (nmol L−1) | 4.47 ± 3.93 (88) [3.52] | 7.94 ± 8.04 (101) [4.45] | 17.7 ± 9.78 (55) [14.9] | 15.1 ± 6.82 (45) [13.5] |
t max (h) | 5.00 (1.00, 12.03) | 10.00 (1.00, 12.02) | 3.00 (1.00, 12.00) | 4.52 (0.98, 12.15) |
AUC (nmol h L−1)c | 20.6 ± 16.3 (79) [12.5] | 43.8 ± 51.5 (118) [19.3] | 123 ± 66.3 (54) [96.3] | 113 ± 55.2 (49) [93.8] |
The results of the linear mixed effects model analysis on AUC and Cmax of baseline corrected 5-CH3-H4folate (ratio estimates [geometric mean] with 90% confidence limits) are summarised in Table 3. Camembert and wheat germs were not bioequivalent to pteroylmonoglutamic acid solution in terms of dose-normalised Cmax and AUC of 5-CH3-H4folate, since the treatment ratio estimate and 90% CI were outside the predefined 80–125% acceptance range (Table 3). Spinach was also not bioequivalent since the treatment ratio estimate and 90% CI of dose-normalised Cmax and AUC of 5-CH3-H4folate did not fall completely within the predefined 80–125% acceptance range (Table 3). The statistical tests on model effects revealed that the model treatment effect was proven as statistically significant for AUC (p < 0.001) and Cmax (p < 0.001), whereas sex, sequence and period effects were not. Regardless of the statistic evaluation it is worth comparing the ratio estimates given in Table 3 for the different test foods. From both AUC and Cmax the highest bioavailability is indicated for spinach, which appears higher than that from wheat germs and lowest bioavailability from Camembert cheese can be assumed.
Dependent | Reference | Test | Lower 90% CI | Ratio [%Ref] | Upper 90% CI |
---|---|---|---|---|---|
ln(AUC) | Oral solution | Camembert | 5.1 | 8.8 | 15.4 |
Oral solution | Wheat germs | 19.5 | 33.0 | 56.0 | |
Oral solution | Spinach | 43.4 | 73.0 | 122.9 | |
ln(Cmax) | Oral solution | Camembert | 11.9 | 17.4 | 25.5 |
Oral solution | Wheat germs | 36.3 | 52.1 | 74.8 | |
Oral solution | Spinach | 55.7 | 79.6 | 113.7 |
The percentage of 5-CH3-H4folate excreted into urine relative to the respective dosage (‘%Excretion’) was 1.7%, 2.7%, 6.8% and 21.8% after administration of Camembert, wheat germs, spinach and pteroylmonoglutamic acid solution, respectively, with variabilities (%CV) between 50 and 108%. The ‘%Excretion’ was subjected to the linear mixed effects model analysis as secondary analysis. As the point estimate (geometric mean treatment ratio [% of Reference] of ‘%Excretion’) and the 90% confidence intervals fell outside the prespecified 80–125% acceptance interval, average bioinequivalence of folate originating from Camembert, wheat germs and spinach vs. oral pteroylmonoglutamic acid solution was proven for the urine data (results no shown). The model treatment effect was proven as statistically significant. There was also a significant period effect of unknown cause, but no sequence and sex effect. Thus, the analysis of the urine data – except for spinach – point in the same direction as the corresponding plasma data.
In the context of this study the individual predose concentration of 5-CH3-H4folate was defined as baseline. To prove absence of a diurnal rhythm in plasma folate levels, which would have led possibly to a more complicated correction of the baseline, the folate levels of an additional subject were analyzed over 24 hours under a virtually folate-free diet. The provided food (for the virtually folate-free diet) as well as all other standardised conditions such as facility, medical personnel, blood and urine collection procedure were identical to that of the actual study days. Since there was little variation in the subject's folate levels, the concentrations varied between 13.8 and 17.3 nmol L−1 during 24 hours, the baseline correction procedure used in this study seemed to be justified.
In contrast to the original plans for the study protocol no equimolar doses of folate/pteroylmonoglutamic acid were administered to the subjects. This was compensated by dividing the plasma values by the dose. From a similar study on the bioavailability of spinach folates14 it can be deduced that in the chosen dose range no deviation from dose-proportionality should occur, so that this approach is certainly justified. Comparison of the dose-normalised AUC between test (food folate) and ‘reference’ pteroylmonoglutamic acid has been accepted as a valuable indicator of absorption, provided the post-dosing plasma measurement test period is long enough to capture ≥80% of the whole AUC (extrapolated to infinity). In the majority of cases, the determination of the whole AUC was not possible due to increasing 5-CH3-H4folate concentrations toward the end of the study (Fig. 1). Therefore, the terminal elimination phase could not be determined reliably in this study, which, however, would have been needed for a correct biokinetic evaluation. The range for AUC determination was limited to 12 hours post-dose to cover at least the initial absorption and metabolism of pteroylmonoglutamic acid. In consideration of this, it remains open whether the use of the urine data had some advantage over plasma data, since the absorption phase was satisfactorily covered. It was also shown that estimates of ‘%Excretion’ were subject to a high degree of variability, and cannot be taken as more reliable than those obtained from plasma concentration-time profiles.32 Thus, urinary excretion is not recommended as a substitute for blood concentration data; rather, these studies should be used in conjunction with blood level data for confirmatory purposes.33
Bioavailabilities of folates from the foods used in this study could not be calculated when using the model applied as the kinetics of plasma 5-methyltetrahydrofolic acid response to food folates is different to that from pteroylmonoglutamic acid as shown by Wright et al.34 and recently by a dual-label ileostomy model.29 However, relative bioavailabilities can be estimated from the model presented if the following four conditions are fulfilled: (1) that physiological doses of folates and pteroylmonoglutamic acid are initially reduced and then methylated in the intestine and the liver and that essentially only 5-methyltetrahydrofolic acid appears thereafter in circulation, as it is the case for absorbed physiological doses of all naturally-occurring reduced folates; (2) that plasma 5-methyltetrahydrofolic acid response derives entirely from (biotransformed) newly absorbed folate;35 (3) generally, dose-proportionality of absorption existed in the dose-range investigated; and (4) saturation of the subjects with pteroylmonoglutamic acid does not significantly alter folate absorption. Under these assumptions it can be stated that the relative bioavailability of folate from spinach was higher than that from wheat germs and Camembert cheese. It is further assumed that through the chosen study design (randomised cross-over) any systematic errors are reduced as far as possible. Nevertheless, due to the limitations outlined above the results of this study should be interpreted with due caution.
Footnotes |
† Current address: BIOANALYTIK Weihenstephan, Research Center for Nutrition and Food Sciences, Technische Universität München, Alte Akademie 10, D-85350 Freising, Germany. |
‡ Current address: Centre for Nutrition and Food Sciences, Queensland Alliance for Agriculture and Food Innovation (QAAFI), The University of Queensland, 39 Kessels Road, Coopers Plains, QLD 4108, Australia. |
§ Current address: ARC Centre of Excellence in Plant Cell Walls, Centre for Nutrition and Food Sciences, Queensland Alliance for Agriculture and Food Innovation (QAAFI), The University of Queensland, St. Lucia, QLD 4072, Australia. |
This journal is © The Royal Society of Chemistry 2015 |