A human model to determine folate bioavailability from food: a pilot study for evaluation

Background: Knowledge about folate bioavailability from food is essential for the estimation of dietary requirements. Yet, there is a lack of data obtained from validated human studies performed with physiological folate doses. Objective: In this pilot study, a new model for the determination of folate absorption is developed and validated. Design: Under strictly standardized procedures, two healthy ileostomy volunteers consumed single portions of test foods or an oral dose of a pharmaceutical folate preparation of the natural folate diastereomer (6S)-5methyltetrahydrofolate. Relative folate absorption from oral doses versus an intramuscular injection of the same pharmaceutical preparation was determined using postdose plasma folate concentration curves. Nonabsorbed folate was estimated by postdose folate excretion into stomal effluent. Results: Estimated by plasma areas under the curve, relative folate absorption ranged from 47 to 67% for oral doses from the test foods strawberries and broccoli and the pharmaceutical (6S)-5-methyltetrahydrofolate preparation. During 10 h postdose, 19 /44% of the dietary folate was excreted with the stomal effluent. Varying gut passage times were observed for different food matrices by determining ileostomal folate excretion in 2 h intervals. Around 90% of the folate from the oral doses was recovered in the collected body fluids, plasma and stomal effluent, by 10 h postdose, independent of the size of the administered folate doses of 200 or 400 mg (0.4 or 0.9 mmol). Conclusion: The results imply that this model provides a suitable tool to estimate folate bioavailability from foods.


Introduction
An optimal folate status is linked to a diminished risk for neural tube defects and spontaneous abor-tions (1), low serum homocysteine levels (elevated concentrations are associated with an increased risk of occlusive vascular diseases), some forms of cancer (2) and improved cognitive or mental functions (3). This new concept of health-protective effects is reflected in recent recommendations of the US Food and Nutrition Board (4) and latest editions of several European and the Nordic Nutritional Recommendations (5).
However, information about the extent to which certain foods contribute to the increased folate intake is still incomplete. This emphasizes the need for a critical evaluation of dietary folate sources and of the impact of food-processing techniques on folate content and human folate bioavailability.
Several human models have been developed including long-and short-term protocols to determine either folate absorption or bioavailability. Long-term protocols usually investigate the impact of repeated folate ingestion on folate status, e.g. by feeding fortified foods, natural food folates or folic acid supplements (6,7). Subsequently, the folate status is often determined by plasma or erythrocyte concentrations; the latter reflects the folate status over the past 3 months. More recent studies also include serum homocysteine (8,9). However, longterm studies are intensive in respect to both cost and time, and reflect the total diet rather than effects of a single food source.
Short-term protocols often compare a single oral folate dose from a pharmaceutical folate preparation with fortified or natural food. Often, postdose plasma or urinary folate concentrations are determined (10Á/13). The area under the plasma response curve (AUC) or urinary folate excretion allows the determination of relative folate bioavailability from food versus pharmaceutical preparations. A drawback is that the application of a single folate dose of physiological magnitude can result in a minimal plasma or urine response. Therefore, pharmacological doses are often used that do not reflect physiological conditions (7,10,14,15).
Stable isotope protocols to study folate bioavailability were introduced by Gregory and co-workers (14,16), who studied bioavailability from isotopically labelled folates in fortified foods or as pharmaceutical preparations using high-performance liquid chromatography (HPLC) and subsequent gas chromatographicÁ/mass spectrometric (GC-MS) techniques. This dual-label oral/intravenous approach has been applied by European researchers to study relative folic acid bioavailability from some cereal grain foods (17). An advantage of the use of stable isotope protocols is that isotopically labelled folate from the dose can be differentiated from endogenous body folate. However, procedures are intensive and costly regarding sample preparation techniques, provision of labelled folates and MS equipment, and therefore not commonly applicable.
A new approach to determine folate bioavailability was used in a study by Konings et al. (18), who monitored apparent folate absorption from spinach in healthy volunteers with ileostomy by the simple balance of consumed and excreted dietary folate. Konings' study was the first one to compare direct folate absorption from stomal effluent data with relative folate absorption from plasma AUC.
The present pilot study was carried out to develop and evaluate a new human ileostomy model for the determination of folate bioavailability from single test food doses (broccoli and strawberries) versus pharmaceutical folate doses. Using several parameters, the absorbed amount of folate was determined directly and indirectly. Relative folate absorption was estimated using short-term postabsorption plasma kinetics after single oral doses of test food compared with a reference dose of the naturally occurring diastereomer (6S)-5-methyltetrahydrofolate [(6S)-5-CH 3 -H 4 folate] by intramuscular injection. Folate excretion into urine was evaluated as a parameter for relative folate absorption. Folate excretion into the ileostomal effluent reflects non-absorbed folate, and by collecting individual fractions every 2 h postdose, not only total folate absorption but also a time pattern of intestinal passage can be determined.
Furthermore, the volunteers' compliance to the strictly standardized study protocol with respect to sampling times, dose ingestion and fasting periods was tested before the application of the model in a main trial.

Subjects
Three subjects were recruited, of whom two completed the study. These two were male ileostomists aged 55 and 77 years, and were apparently healthy based on routine haematological and biochemical measurements and a physical examination. They had a body mass index of 26.7 and 27.3 kg m (2 , were non-smokers and did not use any medication or vitamin supplements. Subjects were screened for fasting serum folate ( /3.4 nmol l (1 ), serum cobalamin ( /96 pmol l (1 ) and erythrocyte folate ( /215 nmol l (1 ) concentrations to ensure normal folate and vitamin B 12 status. They had undergone proctocolectomy 23 and 25 years earlier as a result of ulcerative colitis; at most, 10Á/15 cm of the distal ileum had been resected. Neither subject had undergone further surgical procedures since the initial surgery, with establishment of conventional ileostomy. Both had well-established and functioning ileostomies with no signs of inflammation since Food folate bioavailability the operation. The protocol was approved by the ethics committee of the Umeå University Hospital and was fully explained to the participants before acquiring written informed consent.

Study design
Volunteers participated in four strictly standardized independent study days (each 2Á/4 weeks apart) in a random order. On three occasions they received either a single oral poortion of each test food (strawberries, day S; or broccoli, day B) or a pharmaceutical preparation of (6S)-5-CH 3 -H 4 folate (day O) after an overnight fast. On one occasion, volunteers received an intramuscular injection of (6S)-5-CH 3 -H 4 folate (day I) as a reference dose. One volunteer, attended a fifth study day, on which no folate (day N) was given to account for baseline data regarding folate excretion into stomal effluent and the diurnal folate profile in plasma throughout the study day. Each study day was strictly standardized with respect to sampling procedures and times.
As the trial lasted for several months, the volunteers' folate status was standardized to avoid influences from the liver first-pass effect during absorption. Volunteers' folate body stores were saturated by a daily dose of 0.96 mg folic acid from day 9 to day 2 before each study day, similar to procedures described and used by Gregory and coworkers (14,16,19).

Preparation of test foods and diets
Test foods were given as single portions; either 400 g strawberries or 300 g steamed broccoli. Strawberries of the cultivar Camorosa were immediately frozen after collection and cleaning. For consumption, they were thawed covered overnight in a refrigerator. The broccoli was a frozen vegetable product (Signum † ) available on the Swedish market. A portion of 300 g was microwaved at 590 W for 7.5 min after the addition of 22.5 g water and 0.9 g NaCl. Strawberries and broccoli contained 50.79/ 9.4 mg 100 g (1 (n 0/3) and 64.49/5.9 mg 100 g (1 (n0/3) 5-CH 3 -H 4 folate, respectively.
Preparation of pharmaceutical doses (6S)-Ca-5-CH 3 -H 4 folate (Knoll-BioResearch/BASF Pharma, S Antonio, Switzerland) was produced according to GLP/pharmaceutical standards and tested by the manufacturer for chemical and microbiological purity (according to Ph. Eur. FDA and USP 23; endotoxin concentration B/0.3 EE mg (1 ). Aliquots of 1.5 mg (6S)-Ca-5-CH 3 -H 4 folate (corresponding to 2.55 mmol after correction for Ca and water content) were weighed into amber glass ampoules under sterile conditions and stored at (/208C until use. The intramuscular solutions were prepared 20 min before injection by dissolving the ampoule content with 3 ml sterile isotonic NaCl solution for injection (Fresenius Kabi, Uppsala, Sweden), and 1 ml (corresponding to 0.87 mmol or 400 mg Ca-5-CH 3 -H 4 folic acid) was injected intramuscularly through a sterile filter (Millex-GS, non-pyrogen, 0.22 mm; Millipore, France). The injected volume was determined by weighing the syringe before and after injection (four decimals). Aliquots of the remaining solution were stored at (/208C for 5-CH 3 -H 4 folate quantitation by HPLC, containing 392.3 mg ml (1 (853.8 nmol, 0.9978 g injected) and 378.6 mg ml (1 (823.9 nmol, 0.9826 g injected) for volunteer 1 and 2, respectively. The oral dose was prepared similarly 10 min before consumption by pipetting 1 ml of the (6S)-Ca-5-CH 3 -H 4 folate injection solutions [of 396.5 mg ml (1 (862.9 nmol) and 369.7 mg ml (1 (804.6 nmol)] into a glass of tap water (100 ml).
Collection and treatment of samples Venous blood samples were collected via an intravenous catheter, inserted according to routine procedures into the arm vein. Blood was collected into 5 ml K 3 -EDTA Vacutainers at defined timepoints 10 min predose and at 20, 40, 60, 90, 120 min and 3, 4, 6, 8 and 10 h postdose. The samples were stored on ice, excluded from light, and centrifuged at 1500 g for 10 min under cooling within 30 min. The plasma was transferred into 1.5 ml vials and  Subjects collected urine for 10Á/12 h after the oral dose on spontaneous bladder emptying into individual 500 ml plastic bottles containing 2 g ascorbic acid. The time of collection was noted down. Volume was determined by weighing. After thorough mixing, two aliquots of approximately 50 ml (if available) were stored frozen at (/208C for folate analysis in 125 ml bottles. Overnight collection of urine (12Á/24 h past intervention) was carried out repeatedly into a single 1 litre bottle containing 10 g ascorbic acid, and samples were treated as described above.
Ileostomy bags were changed every 2 h postdose over a period of 10Á/12 h. Exact time-points of bag change were noted. Locked bags were immediately frozen on dry ice. The content was quantified by weighing. During postdose overnight collection bags were changed at the volunteers' convenience, preferably as often as possible. Bags were stored frozen at (/208C until analysis. . Hog kidney deconjugase was prepared from 10 g hog kidney acetone powder (Sigma) and dialysed overnight (21). The dialysate was stored in portions at (/208C until use. Bovine folate binding protein (cat. no. F0524) was purchased from Scripps Laboratories (San Diego, CA, USA), and agarose affinity gel 10 and poly-prep chromatography columns from BioRad Laboratories (Richmond, USA). Deionized water was of Milli-Q grade.

Sample pretreatment
Plasma samples were thawed in the dark and centrifuged for 5 min at 10 000 rpm (Eppendorf 4517C). Samples were purified by solid-phase extraction (SPE) using strong anion exchange (SAX) columns (LiChrolut 200 mg, Merck) and a vacuum manifold. Columns were prepared by flushing with 2.5 ml of methanol and 2.5 ml of water; 0.5 ml of plasma was applied. The column was washed with 2 ml deionized water and eluted with 2.5 ml 0.1 M sodium acetate containing 10% (w/v) sodium chloride, 1% (w/v) ascorbic acid and 0.1% (v/v) mercaptoethanol. The eluted volume was quantified by weighing. Samples were transferred into amber autosampler vials, overlaid with nitrogen, loaded onto the cooled autosampler (88C) and analysed within 12 h.
Ileostomy samples were thawed overnight at 58C, sheltered from light, in the original ileostomy bags and homogenized for 120 s with an Ultra-Turrax T25. Aliquots of 4 g were added to 20 ml 0.1 M potassium phosphate buffer, pH 6.0, 2.0% (w/v) ascorbic acid and 0.1% (v/v) mercaptoethanol in screw-capped vials under nitrogen atmosphere. Strawberries were crunched in a bag while halffrozen, and 20 g mashed material was added 1:1 (w/ w) to the phosphate buffer and homogenized for 60 s. Then, 4 g of the homogenate was added to 20 ml phosphate buffer in screw-capped vials. Broccoli was thawed and roughly cut, and 15 g was homogenized with 30 g phosphate buffer for 120 s in an extraction vial. All samples were extracted under nitrogen atmosphere for 12 min in a boiling water bath, then cooled on ice in the dark to room temperature, and the pH was adjusted to 4.9 with 1.5 M phosphoric acid. Hog kidney deconjugase preparation was added (2 ml to ileostomy samples, 3 ml to food samples), and the samples were overlaid with nitrogen and incubated for 3 h at 378C in a shaking water bath. Samples were heated for 5 min in a boiling water bath to deactivate the enzyme, cooled on ice and centrifuged at 15 000 g and 58C. The supernatant was collected and the pellet resuspended in approximately 10 ml 0.1 M potassium phosphate buffer, pH 6.0, 0.1% (v/v) mercaptoethanol and centrifuged as above. Supernatants were combined, made up to a volume of 50 ml, mixed and stored as subsamples at (/208C until purification. Extracts from food samples were purified by SPE using SAX columns (Isolute 500 mg, International Sorbent Technology) and an elution buffer of 0.1 M sodium acetate containing 10% (w/v) sodium chloride, 1% (w/v) ascorbic acid and 0.1% (v/v) mercaptoethanol according to Jastrebova et al. (21). When purifying ileostomy samples the volumes were modified: 1.0 ml sample was applied to SAX, Food folate bioavailability a first portion of 0.5 ml was discarded after elution and a second portion of 3 ml was collected and quantified by weighing (three decimals). Samples were transferred into amber vials, overlaid with nitrogen, loaded onto the cooled autosampler (88C) and analysed within 12 h.
To 55 ml urine, 5.5 ml of 0.5 M potassium phosphate buffer, pH 7.0, was added; if necessary, the pH was adjusted to pH 7.0 with 1 M NaOH. The samples were left for 30 min at room temperature, sheltered from light, for salt precipitation and then filtered (Munktell V120H).
Affinity chromatography columns containing 0.5 mg folate binding protein (Scripps, USA) in 2 ml agarose gel per column were prepared according to Konings (22). For urine purification, the columns were equilibrated with 5Á/10 ml 0.1 M K 2 HPO 4 , pH 7. Desalted urine (20Á/40 ml) was loaded onto each affinity column. The columns were washed with 5 ml 0.05 M K 2 HPO 4 , pH 7. Samples were eluted with 10 ml 0.02 M trifluoroacetic acid/0.01 M dithiothreitol into a volumetric flask containing 60 ml 1M piperazine and 0.2% Na ascorbate (200 ml of a freshly prepared 10% (w/v) Na ascorbate solution) according to Karilouto et al. (23). Samples were transferred into amber vials, loaded onto the cooled autosampler (88C) and analysed within 12 h by HPLC. Affinity columns were washed with 15 ml 0.1 N HCl and re-equilibrated with 15Á/20 ml 0.1 M K 2 HPO 4 , pH 7.

Folate quantitation
Test foods, pharmaceutical folate preparations and human samples were quantified by HPLC as described by Jastrebova et al. (21) on a HP 1100 series system equipped with a gradient quarternary pump, a cooled autosampler (88C), a thermostatically controlled column compartment (238C), a diode array and a fluorescence detector. The system was controlled by Chemstations software (Rev A.07.01 [682] Hewlett-Packard 1990Á/1995). The analytical column was Zorbax SB C8, 150)/4.6, 5 mm (Agilent Technologies). The mobile phase consisted of 0.03 M phosphate buffer, pH 2.3, using a linear gradient with acetonitrile starting at 6% with a lag time of 5 min, rising to 25% within 20 min. The flow rate was 0.4 ml min (1 with a total run time of 33 min. The fluorescence detector (FL) was set to ex/em 0/290/356 nm and the diode-array detector to 290 nm. External calibration was carried out using the standards H 4 folate, 5-CH 3  Spiking of urine samples with 0.5Á/1 ng ml (1 5-CH 3 -H 4 folate (34Á/67% of initial folate concentration) before desalting and affinity purification resulted in recoveries of 92Á/111%. A 5-CH 3 -H 4 folate solution [1 ng ml (1 , in 1.0% Na ascorbate (w/ v), pH 7.0] was included in every batch of samples for affinity chromatography, and applied systematically onto each column in turn, resulting in a CV of 4.5% (n0/18). 5-CH 3 -H 4 folate concentrations in a urine sample, desalted, purified and analysed in triplicate, resulted in a CV of 6.8%. Four urine samples, prepared with one sample each on 2 days, varied from 1.7 to 10.2%.

Kinetic calculations
Non-absorbed folate from the oral dose was estimated by quantification of total folate excretion into stomal effluent over time. To express the net increase in plasma folate above baseline concentrations, the baseline (predose) folate concentration (C 0 ) was subtracted from each postdose folate concentration (C 20 (/C 600 ). The (positive) AUC was calculated according to: where C T (/C T(1 ]/0. When folate concentrations at the last blood sampling (C 600 ) were above baseline concentrations (C 0 ), the theoretical AUC (AUC 6000 ) was estimated by extrapolation: where the elimination constant was estimated as (24): The relative amount of absorbed folate from single oral doses compared with (6S)-5-CH 3 -H 4 folic acid administered by intramuscular injection was calculated individually for each volunteer using the (positive) AUC by the equation: Calculations were done using Microsoft Excel 97.SR.  Fig. 1, indicating a longer intestinal passage for complex food matrices than for pharmaceutical preparations, shown by a delayed t max .
For technical reasons, blood collection had to be finalized after 600 min. For both volunteers, plasma folate levels decreased mostly to predose concentrations 8Á/10 h after ingestion of test food. After administration of pharmaceutical preparations (both orally and by intramuscular injection), which were twice as high as food samples, final plasma folate concentrations were still increased by 0.9Á/3.0 ng ml (1 (2.0Á/6.5 nmol l (1 ) above predose concentrations. In these cases, the theoretical AUC 00 was estimated by extrapolation, comprising 7Á/11% of the total AUC (Table 1). Relative bioavailability for test foods and pharmaceutical preparations, calculated by extrapolated AUCs, ranged from 47 to 64% and from 53 to 67% for volunteer 1 and 2, respectively.

Ileal folate excretion
Folate forms detected in stomal effluent were 5-CH 3 -H 4 folate and H 4 folate. In Table 2, 5-CH 3 -H 4 folate excretion into stomal effluent is shown for 10 h, 12 h and 24 h postdose for each volunteer and test day. On the injection day I and day N, no 5-CH 3 -H 4 folate was administered via the intestine. Only 2Á/5 mg (4Á/11 nmol) 5-CH 3 -H 4 folate was found in stomal effluent over 10Á/12 h postdose during the strictly standardized study day. 5-CH 3 -H 4 folate excretion on these occasions was similar for both volunteers. After application of oral folate doses, however, stomal 5-CH 3 -H 4 folate excretion increased, ranging from 40Á/150 mg (87Á/325 nmol) 10 h (1 to 55Á/160 mg (120Á/350 nmol l (1 ) 24 h (1 postdose. 5-CH 3 -H 4 folate excreted with stomal effluent 10 h postdose corresponded to about 20Á/ 45% of the oral doses. Fig. 2   Relative and absolute folate absorption Agreement between findings on relative folate absorption based on plasma AUC and non-absorbed folate into ileostomal effluent can be assessed by determination of the mean folate recovery of the human model, which amounts to approximately 90% over 10 h postdose without consideration of the minimal urinary folate excretion. Recovered folate from administered doses amounted to 85, 87 and 95% for volunteer 1 and 86, 73 and 106% for volunteer 2 on test days O, B and S, based on data from extrapolated plasma AUCs (Table 1) and folate excretion into stomal effluent 10 h postdose (Table 2).

Discussion
This new human model was developed and evaluated for the determination of folate bioavailability from food. Relative folate absorption was estimated by plasma AUCs after ingestion of a test food versus a reference dose. In contrast to other studies, in which an oral dose of a folic acid pharmaceutical was given as a reference dose (11,18), the biological form (6S)-5-CH 3 -H 4 folate, which is given by intramuscular injection, was chosen. Volunteers' body stores were standardized by saturation before each test day to minimize the hepatic first-pass uptake affecting the plasma response. Folate excretion into stomal effluent was used as parameter to assess  absolute folate absorption from given doses. By introducing one study day without folate application, basic ileal folate excretion from bile was accounted for.
Estimation of the AUC, which is appropriate for pharmacokinetic studies involving drugs, is not perfectly suited to studying the bioavailability of nutrients, which are present in the circulating blood and stored in the body. Nevertheless, this model is used in many studies on folate bioavailability (7, 10, 11). Scott et al. (15) pointed out that the hepatic first-pass effect and enterohepatic circulation are Random order of individual study days for each subject at intervals of 2 Á/4 weeks; oral application of (6S)-5-CH 3 -H 4 folate with 100 ml tap water, intramuscular injection of (6S)-5-CH 3 -H 4 folate; ingestion of 300 g steamed broccoli, ingestion of 400 g thawed strawberries; 5-CH 3 -H 4 folate concentrations are given as C(/C 0 (after subtraction of basal predose 5-CH 3 -H 4 folate concentration).

Food folate bioavailability
affected by the folate status of body stores, which could result in variable plasma responses. In the present study, volunteers' folate status was standardized and hepatic first-pass folate uptake was minimized by saturating the body stores with a daily dose of a commercial folic acid supplement according to procedures described by Gregory's group (14,19). Folic acid presaturation was completed 2 days before each study day to avoid interference from bile containing folate derived from the saturation dose.
The enterohepatic circulation results in overestimation of absorbed folate using plasma AUC when biliary folate is completely reabsorbed, or results in overestimation of non-absorbed folate in stomal effluent when reabsorption is incomplete. Information on biliary folate excretion is inconsistent. This study standardized for possible effects from biliary folate excretion by strictly standardizing ingestion of test foods, collection of samples and consumption of a low-fat/low-folate snack at defined times. Volunteers were not allowed to consume coffee or tea until 10 h postdose. Data from study days N and I, when no oral folate dose was administered, could be carefully interpreted as baseline data regarding effects from the enterohepatic circulation. The minimal AUC on day N after no folate administration ( B/5% of the AUC from day I, volunteer 1) indicates that neither bile folate nor the low-folate/ low-fat snack at lunch influenced postdose plasma folate concentrations remarkably. It is recom-mended that an N day is included for each volunteer to account for possible interindividual variation in bile excretion.
The effects of the size of the test dose on plasma response were assessed, considering 200 and 400 mg (435 and 870 nmol) folate as moderate to large physiological doses. Both doses resulted in increased folate concentrations in plasma, urine and ileostomal effluent (Tables 1 and 2). Postdose plasma AUCs increased 6-fold to more than 20-fold compared with the small AUC after no folate administration (day N, volunteer 1).
The chosen number and time-points of blood collection allowed kinetic interpretation of plasma curves, indicating a similar pattern for both volunteers regarding gut passage and folate absorption. Gut passage time increased for pharmaceutical 5-CH 3 -H 4 folateB/strawberries B/broccoli, as interpreted from increasing t max (Fig. 1). Pharmaceutical preparations of 5-CH 3 -H 4 folate administered by intramuscular injection resulted in greater AUCs for both volunteers than all oral doses, which could be accounted for by incomplete intestinal absorption and/or possibly folate destruction during gut passage. When estimating relative folate absorption from test meals by AUC, it is important to refer to the AUC of a controlled (known) amount of ingested pharmaceutical. This is guaranteed by the application of a reference dose by injection, where no confounding influences can reduce the resulting AUC size. Konings et al. (18) and others (11) used Total excreted as% of intake Á/ an oral pharmaceutical dose of folic acid as a reference dose. That might result in overestimation of relative folate absorption from food samples, as it is not possible to account for underestimation of the reference AUC caused by destruction of the pharmaceutical folate dose during gut passage.
In dual-label stable isotope trials, usually a bolus intravenous injection of folic acid is usually pre- Food folate bioavailability ferred as reference dose (7,14,17). In this study the reference dose was administered by intramuscular injection, because with a given number and timepoints of blood sampling the assessment of the plasma concentration curve is more accurate after an intramuscular than an intravenous injection owing to a slower folate invasion into the plasma compartment. Furthermore, this study used a new approach, using as a reference substance the biological folate form (6S)-5-CH 3 -H 4 folate, which is dominantly present in food, to ensure similar handling by the body.
Volunteers' absolute and relative plasma AUCs were different in size; this may be carefully interpreted as interindividual variation. High interindividual variation in AUCs was observed by Konings et al. (18) (variation of individual AUCs on 3 test days resulted in CVs between 75 and 90%, n0/11) and Prinz-Langenohl et al. (11) (variation in individual AUCs on 3 test days resulted in CVs between 34 and 47%, n 0/20). Because only two volunteers completed the current study, it is not possible to analyse the results statistically.
Usually, especially after administration of the smaller folate dose (200 mg, 435 nmol), plasma folate concentrations returned to baseline levels 8Á/ 10 h postdose, indicating the completeness of plasma collection. However, plasma folate curves can be extrapolated to estimate the total AUC if initial concentrations are not quite reached. It is desirable to keep the period of postdose blood collection and fasting as short as possible, to increase compliance by the volunteers.
Using dual-label stable isotope protocols, urinary isotope excretion ratios are often chosen as criteria of absorption, requiring either large folate doses exceeding the threshold of renal absorption or saturation of body tissues (7,15) to improve uniformity among subjects. After application of labelled folic acid from fortified foods and intravenous injections, only 1Á/3% of the dose was found as intact folate in 24 h urine (14), while major excretion appears via metabolites. According to the present results (Table 2), intact 5-CH 3 -H 4 folate amounting to 1Á/3% of the administered doses was excreted into 24 h urine. Pfeiffer et al. (14) reported that after ingestion of folic acid 4Á/13% of the urinary folate was excreted in the form of folic acid and not as 5-CH 3 -H 4 folate, the only folate form present in predose urine. Folic acid was not given in the present pilot trial. The parameter urinary folate excretion was considered to be limited using this study design (without isotopic labelling), possibly providing qualitative rather than quantitative information.
The ileostomy model to study nutrient digestion and absorption was introduced by Sandberg et al. (25). It offers a method for direct and quantitative determination of small bowel excretion, providing that the influence of bacteria on the effluent can be minimized. In the current study, both volunteers had a very low intestinal bacterial activity according to an earlier study (data not shown). Since this study used a strictly standardized protocol for effluent collection, in which the bag was changed every 2 h with immediate freezing on dry ice, it may be assumed that no interference occurred from minor bacterial activity.
So far, only one folate ileostomy trial has been published (18), indicating a high folate absorption between 73 and 91% from two spinach meals and an oral dose of folic acid based on 24 h postdose stomal folate excretion. Volunteers showed a remarkably high interindividual variation, as expressed by CVs of 60Á/90% (n0/11) for individual ileostomal folate excretion on the same test days. Variable folate excretion was observed for both volunteers (Table 2), and folate absorption from the test foods estimated from 10 h postdose ileostomal effluent excretion was between 55 and 80%. In agreement with Konings et al. (18), the main folate form detected in stomal effluent samples was 5-CH 3 -H 4 folate after ingestion of test foods containing this vitamer as the dominant folate form.
Collection of individual 2 h effluent samples, a procedure newly introduced by this model, demonstrates that a 10Á/12 h effluent collection under standardized conditions may be sufficient to estimate non-absorbed folate (Fig. 2). Furthermore, matrix-dependent differences in gut passage time are indicated for pharmaceutical preparations and test foods, as seen by delayed maximal folate excretion.
Based on the results of this pilot study, a 10Á/12 h collection period of individual ileostomy samples is recommended, including for each volunteer a baseline day with no folate application (N day) to account for individual basal folate excretion. Effluent samples from the N and I days, both after no oral folate application, demonstrated that neither the intake of the low-folate snack nor periods of fasting remarkably increased stomal folate excretion during the standardized 10Á/12 h postdose period (Table 2), as supposed by Konings et al. (18).
Reports regarding age-related effects on folate absorption are inconsistent, partly owing to trials using inadequate test doses exceeding physiological test doses, as reviewed by Gregory (26). However, using an intestinal perfusion technique to study luminal disappearance and subsequent urinary recovery of radiolabelled folate compounds, Bailey et al. (27) showed that the absorption of monoglutamyl and polyglutamyl folates was equivalent in young and elderly volunteers. Owing to both a lack of recent studies showing altered absorption with age, and the good health status of the volunteers, there was no apparent reason to exclude one volunteer, aged 77 years, as no age range for exclusion could be defined.
In summary, the human model presented is a suitable tool to study folate absorption from foods, using stomal folate excretion and plasma AUC to determine absolute and relative folate absorption. The new study design allows correction for interindividual variation by inclusion of a day (N day) accounting for basic plasma and stomal effluent data. A new approach is the use of the biological form of (6S)-5-CH 3 -H 4 folate as a reference dose, given by intramuscular injection. Folate determination in individual 2 h effluent samples postdose gives information on the gut passage times of test foods. The volunteers complied well with the strictly standardized study design. The intervals chosen for taking samples, as well as the portion sizes of the test foods, aiming for a folate dose of approximately 200 mg (435 nmol), seemed appropriate. This model will be used to determine folate absorption from various processed foods and from pharmaceutical folate preparations using a greater number of volunteers to enable statistical and biokinetic interpretation of the data.