点击显示 收起
1 From the Division of Human Nutrition and Epidemiology, Wageningen University, Wageningen, Netherlands; the Nutrition Research and Development Centre, Bogor, Indonesia; the Department of Medicinal Chemistry and Pharmacognosy, College of Pharmacy, University of Illinois at Chicago; the Leiden Institute of Chemistry, Leiden University, Leiden, Netherlands; and the Department of Gastroenterology, University Medical Center, St Radboud, Nijmegen, Netherlands.
2 Supported by the Netherlands Foundation for the Advancement of Tropical Research (WOTRO-NWO; grant WV 93-271) and the National Institutes of Health (grant RO1 70771 to RBVB). Hewlett-Packard provided the liquid chromatography–mass spectrometry instrument used in this investigation. 3 Reprints not available. Address correspondence to CE West, Division of Human Nutrition and Epidemiology, Wageningen University, PO Box 8129, 6700 EV Wageningen, Netherlands. E-mail: clive.west{at}staff.nutepi.wau.nl.
ABSTRACT
Background: More information on the bioefficacy of carotenoids in foods ingested by humans is needed.
Objective: We aimed to measure the time required for isotopic enrichment of ß-carotene and retinol in serum to reach a plateau, the extent of conversion of ß-carotene dissolved in oil with use of ß-carotene and retinol specifically labeled with 10 13C atoms, and the intraindividual variation in response.
Design: Indonesian children aged 8–11 y (n = 35) consumed 2 capsules/d, 7 d/wk, for 10 wk. Each capsule contained 80 µg [12,13,14,15,20,12',13',14',15',20'-13C10]ß-carotene and 80 µg [8,9,10,11,12,13,14,15,19,20-13C10]retinyl palmitate. Three blood samples were drawn per child over a period of 10 wk. HPLC coupled with atmospheric pressure chemical ionization liquid chromatography–mass spectrometry was used to measure the isotopic enrichment in serum of retinol with [13C5]retinol and [13C10]retinol and of ß-carotene with [13C10]ß-carotene. The ß-carotene in the capsules used had a cis-trans ratio of 3:1.
Results: Plateau isotopic enrichment was reached by day 21. The amount of ß-carotene in oil required to form 1 µg retinol was 2.4 µg (95% CI: 2.1, 2.7). The amount of all-trans-ß-carotene required to form 1 µg retinol may be lower.
Conclusions: The efficiency of conversion of this ß-carotene in oil was 27% better than that estimated previously (1.0 µg retinol from 3.3 µg ß-carotene with an unknown cis-trans ratio). The method described can be extended to measure the bioefficacy of carotenoids in foods with high precision, requiring fewer subjects than other methods.
Key Words: Bioavailability • ß-carotene • conversion • retinol • vitamin A • children • Indonesia • stable isotope • carbon isotopes • 13C • extrinsic labeling • LC-MS • liquid chromatography–mass spectrometry
See corresponding editorial on page 849.
INTRODUCTION
Vitamin A deficiency is a serious problem in many developing countries (1). Provitamin A carotenoids, such as ß-carotene, are the major source of vitamin A in the diet of a large proportion of the world's population. Unfortunately, the bioavailability of carotenoids in fruit and vegetables and the conversion of carotenoids to retinol are lower (2, 3) than assumed previously (4).
Bioavailability is the fraction of an ingested nutrient that is available for utilization in normal physiologic functions and for storage (5). We define bioconversion as the fraction of a bioavailable nutrient (here, absorbed provitamin A carotenoids) that is converted to the active form of a nutrient (retinol). We define bioefficacy as the efficiency with which ingested nutrients (here, dietary provitamin A carotenoids) are absorbed and converted to the active form of the nutrient (retinol). Because 1 µmol ß-carotene theoretically could form 2 µmol retinol, 100% bioefficacy would mean that 1 µmol dietary ß-carotene (0.537 µg) is 100% absorbed and converted 100% to retinol, yielding 2 µmol retinol (0.572 µg). Thus, the amount of ß-carotene required to form 1 µg retinol would be 0.537/0.572 = 0.94 µg.
Various factors, ordered in the acronym SLAMENGHI, influence the bioefficacy of carotenoids. The factors are S) the species of carotenoid, L) molecular linkage, A) the amount of carotenoids consumed in a meal, M) the matrix in which the carotenoid is incorporated, E) effectors of absorption, N) the nutrient status of the host, G) genetic factors, H) host-related factors, and I) interactions (6, 7).
Until now, the bioefficacy of carotenoids was estimated in humans with use of oral-fecal balance techniques, measurement of plasma or chylomicron responses after single or multiple doses of carotenoids, and tracer methods. In the 1960s, 2 tracer studies measured the recovery of [14C]ß-carotene in lymph (8, 9). To date, only single doses of specifically extrinsically labeled [2H8]ß-carotene (10–12), intrinsically labeled ß-carotene biosynthesized in green algae grown with 13C as the sole carbon source (13), and intrinsically labeled ß-carotene from spinach grown on 30% 2H2O (14) have been administered for studying ß-carotene metabolism in a limited number of human subjects (n = 1–5).
We developed a method that can quantify the bioefficacy of carotenoids with use of multiple low doses of ß-carotene and retinol, each specifically labeled with 10 13C atoms. This enables the isotopic enrichment of both retinol and ß-carotene in serum to reach a plateau. From the degree of labeling of retinol with 5 and 10 13C atoms and the doses of [13C10]-labeled ß-carotene and retinol administered, the bioefficacy of ß-carotene can be calculated. The aim of this study in children in Indonesia was to measure the extent of conversion of ß-carotene dissolved in oil. We also measured the time required for the isotopic enrichment of ß-carotene and retinol in serum to reach a plateau and the intraindividual variation in response.
SUBJECTS AND METHODS
Subjects
The study was conducted from December 1997 to May 1998. Schoolchildren aged 8–11 y (grades 3 and 4) from a rural village in the Bogor District, West Java, Indonesia, were asked to participate in the screening for this study. The purpose and procedures of the screening and the intervention were explained to the parents or guardians of these children. Almost all parents allowed their children to participate and signed an informed consent form. During the screening, a physician examined 140 children and each child's age, sex, weight, and height were recorded. Venous blood samples (4 mL) were drawn from apparently healthy, nonfasted children to measure hemoglobin concentrations and to prepare serum in which the concentrations of retinol and total carotenoids were measured. In addition, children were asked to provide a single stool sample that was examined for protozoan cysts and worm eggs. Children were also asked to record their food intake in a diary for 7 consecutive days and were interviewed once or twice during this week to check the quality of the data in the diaries (24-h recall).
Thirty-six children were then invited to participate in the intervention on the basis of their serum retinol concentrations: < 0.70 µmol/L (low) or > 1.05 µmol/L (high). Together with their parents, these children were informed about the intervention in more detail and their parents signed a second informed consent form. After stratification for weight, height, sex, and serum retinol concentration, the 36 children were randomly allocated to 3 groups of equal size that were studied for 4 wk (group 1), 7 wk (group 2), or 10 wk (group 3). This enabled us to assess the time required for the isotopic enrichment of retinol and ß-carotene in serum to reach a plateau. To assess the effect of vitamin A status on bioefficacy, children with either low or high serum retinol concentrations were distributed throughout the 3 groups. It was not possible to carry out power calculations because this is a new method. Thus, we decided on a sample size of 6 children with low and 6 children with high serum retinol concentrations per group. From this number of children, it should be possible to measure the variation from which power calculations could be made for subsequent studies, even if there was limited dropout. The study was approved by the Medical Ethics Committee of the Ministry of Health, Indonesia; the Indonesian Institute of Science; and the Medical Ethical Committee of Wageningen University, Netherlands.
Study design
Each child received 2 capsules/d, 7 d/wk. Each capsule contained 80 µg [12,13,14,15,20,12',13',14',15',20'-13C10]ß-carotene (analyzed value) and 80 µg [8,9,10,11,12,13,14,15,19,20-13C10]retinyl palmitate (44 retinol equivalents, or RE; analyzed value) (Figure 1). The oily mixture for the capsules was prepared by dissolving the labeled ß-carotene and retinyl palmitate in highly unsaturated sunflower oil (>82% oleic acid and >10% linoleic acid; Hozol RBDW; Contined BV, Bennekom, Netherlands). all-rac--Tocopheryl acetate (Roche Nederland BV, Mijdrecht, Netherlands) was added to the oil as an antioxidant. These actions were carried out under subdued light. The capsules used in this study were made from bovine gelatin (Capsugel, Bornem, Belgium) and were filled with the oily mixture by multipipette. Each capsule contained 0.36 g oil and 150 µg vitamin E. The 13C10-labeled retinol and ß-carotene were synthesized at the Leiden Institute of Chemistry (15). These compounds were food grade, on the basis of criteria established by the US Pharmacopoeia (16), the Joint FAO/WHO Expert Committee on Food Additives (17), and the European Pharmacopoeia (18).
FIGURE 1. . Molecular structure of synthesized [12,13,14,15,20,12',13', 14',15',20'-13C10]ß-carotene and [8,9,10,11,12,13,14,15,19,20-13C10]retinyl palmitate. The asterisks indicate the positions of the 13C labels.
The capsules were consumed after a low-retinol, low-carotenoid meal Monday through Friday and after a chocolate wafer on the weekend. The foods provided contained some fat to promote the absorption of the retinol and ß-carotene from the capsules. Five different menus were used, each of which contained a fixed combination of a lunch served at 1100 and an afternoon meal served at 1500. The 5 menus were randomly allocated over the days of each week. On weekdays, a dietitian supervised and recorded compliance with consumption of the capsules and attendance at meals. On the weekend and on holidays, a teacher supervised and recorded compliance with consumption of the capsules and whether each child received a wafer. Four village health volunteers prepared all meals during the intervention. For each ingredient, the recipes listed the amount to be purchased and the cleaned amount to be cooked. The amount of food to be distributed was listed for each dish. Children recorded their food consumption daily in a diary. For the foods provided, the children recorded the amount not eaten; for the foods not provided, they recorded which foods and drinks they had consumed and in which quantity. To check the quality of the data in the diaries, children were interviewed once per week about which foods and drinks they had consumed and in which quantity on the previous day (24-h recall).
Nonfasting blood samples (8 mL) were collected between 0930 and 1100 from an antecubital vein on days 0, 21, and 28 for group 1; on days 21, 42, and 49 for group 2; and on days 21, 63, and 70 for group 3 (Figure 2). The children were examined by a physician on the day of blood collection and their weights and heights were measured. While the children wore a school uniform but no shoes, weight was measured to the nearest 0.1 kg with a digital electronic scale (770 alpha; Seca, Hamburg, Germany) and height was measured to the nearest 0.1 cm with a microtoise.
FIGURE 2. . Study design of the intervention period. During the period shown by the shaded box, children consumed 2 capsules/d, 7 d/wk, and were provided a low-retinol, low-carotenoid diet 5 d/wk. Each capsule contained 80 µg [13C10]ß-carotene and 80 µg [13C10]retinyl palmitate. Three blood samples were drawn per child as indicated by the arrows.
Methods
Estimation of energy and nutrient intakes
Concentrations of retinol and ß-carotene in the capsules were analyzed by HPLC (19). To measure the nutrient contents of the meals, a duplicate sample of a menu was collected at random once per week. The random allocation was done in such a way that each menu was sampled once during the first 5 wk of the intervention and once during the last 5 wk. Duplicate samples of all dishes of both meals of a collection day were weighed before and after removal of the inedible portion (bones, peels, etc). The edible portions were then pooled, blended, and portioned into 3 plastic bags. All bags were stored at –20°C. Within 3 wk of collection, one bag of each duplicate sample was used to measure fat content by the Soxhlet method, protein content by the micro-Kjeldahl method, water content by evaporating the sample, and ash content by dry ashing the sample (20). Carbohydrate content was calculated by difference. Within 5 wk of collection, another bag of each duplicate sample was used to measure iron content (20). All analyses were carried out in duplicate at the laboratory of the Nutrition Research and Development Centre (NRDC) in Bogor. The last bag of each duplicate sample was stored for 2 mo before being packed on dry ice and transferred to the Division of Human Nutrition and Epidemiology in Wageningen. There the samples were stored for 2 mo at –80°C until analyzed for retinol and carotenoids by HPLC (19).
The energy content of the meals in kJ was calculated by multiplying the weight (in g) of fat, protein, and carbohydrates by 37, 17, and 17, respectively. Results of both duplicate samples per menu were averaged. Time and conditions of storage were adequate to obtain reliable results for the analyses carried out in food and serum (21–23). Energy and nutrient intakes during the screening and during the intervention were calculated by using the diaries, 24-h recalls, records of attendance at meals, and a computer program suite (KOMEET, version 2.0c, and VBS-EDIT, version 1.0; B-Ware Nutrition Software, Wageningen, Netherlands) with a nutrient database (Bg95K98) based on that developed by de Pee et al (24) to which the energy and nutrient contents of the foods provided (on the basis of the duplicate analyses) were added.
Analysis of blood and stool samples
Immediately after blood was collected, a portion of whole blood was removed to count white blood cells (25) and measure hemoglobin concentrations (26). These analyses were done on the day of blood collection at the NRDC. The remaining blood was placed on ice, protected from light, and centrifuged within a few hours (750 x g for 10 min at room temperature) at the NRDC. In serum samples collected during the screening, concentrations of retinol and total carotenoids were measured by HPLC at the NRDC (27). In serum samples collected during the intervention, concentrations of retinol and various carotenoids were measured by HPLC in Wageningen (28). For these latter analyses, serum was stored in a series of containers and frozen at –20°C for 2 mo before being packed on dry ice and transferred to Wageningen. All containers were then stored at –80°C. One container was stored for 2 mo until analyzed for both retinol and carotenoids by HPLC. Two other containers were stored for 4 mo before being packed on dry ice and transferred to the Department of Medicinal Chemistry and Pharmacognosy in Chicago. There the samples were stored for 2 mo at –80°C until analyzed for the degree of isotopic enrichment of retinol (29) and ß-carotene (30) by HPLC coupled with atmospheric pressure chemical ionization liquid chromatography–mass spectrometry (APCI LC-MS).
For the LC-MS method, retinol and ß-carotene were extracted from 0.2- and 1.0-mL serum samples, respectively, and analyzed by reversed-phase HPLC with a C30 column interfaced to a mass spectrometer equipped with positive ion APCI. To assess the ratio of labeled to unlabeled ß-carotene, selected ion monitoring was carried out at mass-to-charge ratios (m/z) of 537 and 547. These ions corresponded to circulating ß-carotene and orally administered [13C10]ß-carotene, respectively. To assess the ratio of labeled to unlabeled retinol, selected ion monitoring was carried out at m/z 269, 274, and 279. These abundant fragment ions corresponded to the loss of water from the protonated molecule of circulating retinol, [13C5]retinol (metabolically formed from orally administered [13C10]ß-carotene), and [13C10]retinol (formed by hydrolysis of orally administered [13C10]retinyl palmitate), respectively. Although the LC-MS method focused on the quantification of all-trans-ß-carotene, the C30 HPLC method also resolves cis isomers from the all-trans peak. Therefore, this LC-MS method may be applied to the quantification of cis as well as all-trans-ß-carotene.
The presence of protozoan cysts and worm eggs was diagnosed in stools by the Ridley method. The load of worm eggs was quantified by the Kato Katz method (31).
Calculation of the bioefficacy of ß-carotene converted to retinol
A mathematical model was developed that uses the isotopic enrichment in serum of both ß-carotene with [13C10]ß-carotene and retinol with [13C5]retinol and [13C10]retinol at plateau isotopic enrichment. This model, which we refer to as CarRet PIE, was used to estimate carotenoid bioavailability and bioconversion and, thus, bioefficacy. This stable-isotope method is based on the isotopic enrichment of retinol and ß-carotene in serum reaching a plateau during multiple dosing with [13C10]retinol and [13C10] ß-carotene. The intake of retinol and ß-carotene from other sources is kept constant and as low as possible. It is assumed that such retinol and ß-carotene mixes completely with labeled retinol and labeled ß-carotene. The bioefficacy of dietary provitamin A (in this study, ß-carotene in oil) compared with that of dietary retinol was calculated as shown in the first 3 columns of Table 1. To assess the time required for isotopic enrichment of ß-carotene and retinol in serum to reach a plateau, the regression coefficients of vitamin A activity of ß-carotene in oil compared with that of retinol in oil were calculated for each group, pooling the data for the second and third time points in groups 2 and 3.
View this table:
TABLE 1. . Description of the ß-carotene and retinol at plateau isotopic enrichment (CarRet PIE) mathematical model
Statistical methods
Data are shown as means and 95% CIs or SDs (in the case of descriptive measures). Variables that were not normally distributed are expressed as medians and 25th to 75th percentiles. Kruskal-Wallis tests were carried out to compare energy and nutrient intakes among groups during the screening and during the intervention and to compare characteristics among groups. Wilcoxon's tests were carried out to compare energy and nutrient intakes between the screening and the intervention and to compare energy and nutrient intakes between the data obtained from the diaries and those obtained from the 24-h recalls. Friedman's tests were carried out to compare serum retinol and ß-carotene concentrations on various blood collection days within each group (eg, for group 2 on days 21, 42, and 49). Kruskal-Wallis tests were carried out to compare the following variables among groups on day 21: serum retinol concentration, serum ß-carotene concentration, isotopic enrichment of retinol with [13C5]retinol and with [13C10]retinol, isotopic enrichment of ß-carotene with [13C10]ß-carotene, and vitamin A activity of ß-carotene in oil compared with that of retinol. Mann-Whitney U tests were carried out to compare the same variables within each group on each blood collection day between those with initial low and high serum retinol concentrations. Spearman correlation coefficients were calculated between serum retinol concentration at screening and vitamin A activity of ß-carotene in oil compared with that of retinol on days 21, 28, 42, 49, 63, and 70. Friedman's tests were used to compare the cis-trans ratio in serum ß-carotene on days 0, 21, and 28 within group 1. To assess the time required for the isotopic enrichment of ß-carotene and retinol in serum to reach a plateau, a Kruskal-Wallis test was used to compare the regression coefficients of the vitamin A activity of ß-carotene in oil compared with that of retinol among groups. A one-sample t test was used to test whether the regression coefficient differed from 0 when the 3 groups were pooled.
The intraindividual CV was assessed by one-way analysis of variance with use of the vitamin A activity of ß-carotene in oil compared with that of retinol as the dependent variable and subject number as a factor. The root mean square (residual SD) is a measure of reproducibility. The intraindividual CV was derived by dividing this SD by the mean vitamin A activity of ß-carotene in oil compared with that of retinol for all data.
All tests were two-sided and P values <0.05 were considered significant. The computer package SPSS (version 7.5.3; SPSS Inc, Chicago) was used for all statistical calculations.
RESULTS
On the first day of the intervention, we made minor changes in the distribution of the children throughout the 3 groups because some children were absent. As a result, group 1 comprised 7 children with an initial serum retinol concentration <0.70 µmol/L (low) and 5 children with an initial serum retinol concentration >1.05 µmol/L (high), group 2 comprised 4 children with initial low and 8 children with initial high serum retinol concentrations, and group 3 comprised 6 children with initial low and 6 children with initial high serum retinol concentrations. Serum retinol concentrations of samples obtained during the intervention, which started 2.5 mo after the screening, showed that classification of children according to vitamin A status could not be maintained. Therefore, the effect of vitamin A status on the bioefficacy of ß-carotene could not be assessed.
Data on the isotopic enrichment of retinol with [13C10]retinol were not available for 2 children from group 1 on day 28 and for 1 child from group 3 on day 21. For this latter child, no data were available on the isotopic enrichment of retinol with [13C5]retinol on day 21. Therefore, for these days, the amount of ß-carotene required to form 1 µg retinol could not be calculated for these children. Data are presented for 35 of the 36 children enrolled because one child from group 3 (with an initial high serum retinol concentration) withdrew from the study.
The descriptive characteristics of the children on day 21 are given in Table 2; on this day, data were available for all children. No significant differences in any characteristic were found among the 3 groups.
View this table:
TABLE 2. . Characteristics of the population on day 211
The drawing of the third blood sample from 3 children (one from group 1 and 2 from group 2) was postponed for 1 wk because the children had been absent for 4 d during the week before the planned blood drawing day. During the additional week, these children continued on the experimental regimen. Blood was drawn only on days on which a child did not show signs of infection as judged by a physician. White blood cell counts in all blood samples taken were <1 x 1010/L, indicating lack of apparent infection on the day of blood collection (25).
Children in groups 1, 2, and 3 consumed 2 capsules/d for 27, 41, or 69 d, respectively. On day 22, which was a religious Muslim holiday, no capsules were provided. Records of attendance showed that 96% of the capsules were consumed (25th and 75th percentiles: 91%, 100%). Data from the 3 children for whom blood sampling was postponed for 1 wk were not included in these capsule compliance calculations, but were included in all other calculations. The median compliance of groups 1, 2, and 3 was 100%, 98%, and 92%, respectively. Correction of data for the lower compliance of group 3 did not change the conclusions of this study.
Energy and nutrient intakes
Energy and nutrients provided and consumed during the screening and the intervention are shown in Table 3. No food or capsules were provided during the screening period. Energy and nutrient intakes during the screening and the intervention did not differ significantly among the 3 groups studied, except for retinol intake during the intervention (20, 16, and 8 µg retinol/d for groups 1, 2, and 3, respectively). Because these differences were not biologically relevant, only intake data from the total group are shown. Energy, fat, protein, carbohydrate, iron, and retinol intakes were significantly higher during the intervention than during the screening (P < 0.05). The additional retinol was derived from the capsules (87 RE/d). During the intervention, retinol intake from foods not provided by us was low (15 RE/d). ß-Carotene intake was significantly lower during the intervention than during the screening. ß-Carotene intake from the food and capsules provided was 223 µg ß-carotene/d, of which 160 µg/d was derived from the capsules and 63 µg/d from the foods. During the intervention, ß-carotene intake from foods not provided by us was 155 µg/d.
View this table:
TABLE 3. . Energy and nutrients provided and consumed daily during the screening and the intervention1
Estimates of energy, fat, protein, carbohydrate, iron, and retinol intakes based on the diaries were higher than those based on the 24-h recalls. Except for retinol, these differences were significant (data not shown). The estimate of ß-carotene intake based on the diaries was lower than that based on the 24-h-recalls, but this difference was not significant.
Bioefficacy of ß-carotene converted to retinol
Shown in Table 4 are the concentrations of retinol and ß-carotene in serum during the intervention. There were no significant differences in concentrations of either retinol or ß-carotene in serum within groups (eg, for group 2 on days 21, 42, and 49) or among groups (eg, on day 21 for groups 1, 2, and 3). Although labeled ß-carotene can be synthesized as >95% in the all-trans configuration, the ß-carotene in the capsules used in this study had a cis-trans ratio of 3:1. Because ß-carotene intake from the capsules was low (160 µg/d), the cis-trans ratio of serum ß-carotene was also low (ranging from 0.01 to 0.18) and did not change significantly during the intervention (day 0 compared with days 21 and 28 in group 1; data not shown).
View this table:
TABLE 4. . Serum concentrations of retinol and ß-carotene and degree of isotopic enrichment of retinol with [13C5]retinol and [13C10]retinol and of all-trans-ß-carotene with [13C10]ß-carotene1
Also shown in Table 4 is the isotopic enrichment in serum of retinol with both [13C5]retinol and [13C10]retinol and of ß-carotene with [13C10]ß-carotene. There were no significant differences in the isotopic enrichment of retinol and ß-carotene or in the vitamin A activity of ß-carotene in oil compared with that of retinol among groups on day 21. Serum concentrations of retinol and ß-carotene, the degree of isotopic enrichment of retinol and ß-carotene, and the vitamin A activity of ß-carotene in oil compared with that of retinol also did not differ significantly within any group on any blood collection day between those with initial low and high serum retinol concentrations, except for the serum retinol concentration on day 49 in group 2 (data not shown). There was no significant correlation between the serum retinol concentration at screening and the vitamin A activity of ß-carotene in oil compared with that of retinol on days 21, 28, 42, 49, 63, and 70 (correlation coefficients ranged from –0.142 to 0.289).
To calculate the vitamin A activity of ß-carotene in oil compared with that of retinol, we used the formulas described in Table 1. Data from a 9-y-old boy on day 21 are shown in Table 1 as an example. Shown in Figure 3 are values for the mean vitamin A activity of ß-carotene in oil compared with that of retinol per group per day. The coefficient of the regression of the vitamin A activity of ß-carotene in oil compared with that of retinol was not significantly different among groups. When the 3 groups were pooled, the regression coefficient was not significantly different from 0 (P = 0.137). Thus, in these children, isotopic enrichment of retinol and ß-carotene in serum reached a plateau by day 21. On the basis of data from days 21 to 70, the mean vitamin A activity of ß-carotene in oil compared with that of retinol was 0.4149 (95% CI: 0.3637, 0.4661). This corresponds to 2.4 µg (95% CI: 2.1, 2.7) ß-carotene required to form 1 µg retinol in the body. If the data are recalculated assuming that all-trans-ß-carotene and cis-ß-carotene are absorbed to the same extent and that the vitamin A potency of cis-ß-carotene is 50% of that of all-trans-ß-carotene, the vitamin A activity of ß-carotene in oil compared with that of retinol is higher: 0.6638 (95% CI: 0.5819, 0.7458), corresponding to 1.5 µg (95% CI: 1.3, 1.7) ß-carotene required to form 1 µg retinol.
FIGURE 3. . Vitamin A activity on a weight basis of ß-carotene (with a cis-trans ratio of 3:1) dissolved in oil compared with that of retinol. Data are means of values obtained from subjects in each group on various days: for group 1 () on days 21 (n = 12) and 28 (n = 10); for group 2 (; n = 12) on days 21, 42, and 49; and for group 3 (•) on days 21 (n = 10), 63 (n = 11), and 70 (n = 11). For calculation of the regression coefficient, the mean values on days 42 and 49 in group 2 and on days 63 and 70 in group 3 were calculated. The regression coefficients for group 1 (solid line), group 2 (dashed line), and group 3 (dotted line) were 0.0049, 0.0016, and 0.0015, respectively.
The intraindividual variation in response comprises both the biological variation within subjects and the variation in the chemical analyses. The latter was arbitrarily set at 0 because it is thought to be much lower than the biological variation within subjects. The SD of the vitamin A activity of ß-carotene in oil compared with that of retinol was 0.0895, which, divided by the mean based on all data (0.4149), resulted in an intraindividual CV of 22%.
Parasitic infestation and anthropometry
The prevalence and intensity of intestinal parasites were low in the children studied. Therefore, we expect that the effect of intestinal parasites on the findings of this study was negligible.
On days 0, 21, and 28 there were no significant differences in weight and height among the 3 groups. Within groups 1 and 2, height was constant and body weight increased by 1 kg during the intervention (28 and 49 d, respectively). Within group 3, height increased by 1 cm and body weight by 1 kg during the intervention (70 d). Because these changes were not expected to affect the outcome of the study, no adjustments in the statistical analyses were made for them.
DISCUSSION
In this study of Indonesian schoolchildren, 1.0 µg retinol was derived from 2.4 µg ß-carotene dissolved in oil. The intraindividual CV was low (22%). Furthermore, the isotopic enrichment of retinol and ß-carotene in serum reached a plateau by day 21.
The method used in this study has several advantages. First, the doses of labeled ß-carotene and retinol used were very low and thus did not perturb the metabolism of unlabeled retinol and ß-carotene. Second, the stable isotopes used entail no health risk and thus can be applied to study humans. Third, the multiple doses enabled the isotopic enrichment of both retinol and ß-carotene in serum to reach a plateau. This plateau makes analysis of data and subsequent calculation and interpretation of results easier than in single-dose studies in which data are analyzed with use of complicated mathematical models. Single-dose studies can be useful for qualitatively studying the kinetics of the bioavailability and bioconversion of ß-carotene. Fourth, administering specifically labeled [13C10]ß-carotene plus [13C10]retinol enables the latter to be distinguished from circulating retinol (mainly [12C]retinol) and from retinol formed in the body from the [13C10]ß-carotene administered ([13C5]retinol). This is impossible after the administration of uniformly labeled retinol and ß-carotene. Without administering retinol, although it would be possible to distinguish between circulating retinol and retinol formed in the body from labeled ß-carotene, it would be impossible to establish retinol absorption and thus to quantify retinol and ß-carotene bioavailability.
Tang et al (12) developed a stable-isotope method that uses single doses of [2H8]ß-carotene and [2H8]retinyl acetate given on separate occasions up to 2 y apart. It would have been possible to administer [2H8]ß-carotene and [2H8]retinyl acetate simultaneously with subsequent measurement of [2H8]retinol and [2H4]retinol, thus eliminating differences related to giving the labeled compounds at different times. However, the low isotopic purity of the 2H-labeled compounds used in Tang et al's method would decrease the signal-to-noise ratio, especially when enrichment approaches baseline. Thus, the use of not only 2H-labeled compounds but also single doses would possibly contribute to a lower intrinsic accuracy of Tang's method compared with our method.
A fifth advantage of our method is that the APCI LC-MS method developed for this study does not require derivatization or saponification of serum samples before analysis, whereas the use of gas chromatography–mass spectrometry does require these labor-intensive processes (32). Intrinsic labeling yields molecules with a variable number of 13C or 2H atoms. This unnecessarily complicates the use of MS for the detection and subsequent quantification of bioavailability. Finally, for this study, we synthesized extrinsically and specifically labeled retinol and ß-carotene. These compounds have several advantages, such as the high incorporation of 13C atoms (99%) and the absence of scrambling (ie, leakage of 13C atoms to positions in the molecule other than those in which they were initially incorporated). Scrambling occurs more often when deuterium is used for labeling, thus requiring measurement at different masses in the MS method. Summation of the signals leads to higher analytic variation. In addition, the labeled compounds can be synthesized as 95% in the all-trans configuration, which is the configuration in which they occur in nature.
The low intraindividual CV in the vitamin A activity of ß-carotene in oil compared with that of retinol was derived in part from the experimental design. The high compliance with capsule consumption (96%) also contributed to the low variation observed. Unfortunately, as noted above, only 25% of the labeled ß-carotene was present as all-trans-ß-carotene. Although care was taken to prevent isomerization of synthesized all-trans-ß-carotene, cis-isomerization of labeled ß-carotene probably occurred during capsule preparation (33). Isomerization did not continue after capsule preparation because the ratio of cis-ß-carotene to all-trans-ß-carotene in the capsules was measured by HPLC after 9, 22, 29, and 50 d of storage at 4 and 30°C and remained stable. More knowledge on the metabolism of cis-ß-carotene is required to decide whether the conversion factor given here may be even lower.
In this study, 2.4 µg ß-carotene was required to form 1 µg retinol. If the efficiency with which dietary ß-carotene is absorbed and converted to retinol in the body is 100%, then 0.94 µg ß-carotene would have been required. Thus, in this study, bioefficacy was only 39%. At this stage of development of our model, it is not possible to quantify bioavailability and bioconversion separately. Note that in this study none of the values for the vitamin A activity of ß-carotene in oil compared with that of retinol were 1.06, which corresponds to a bioefficacy of ß-carotene of 100%. Vitamin E was added to the capsules as an antioxidant for the oil, but we do not know to what extent the vitamin E affected the bioefficacy of the ß-carotene in oil.
The isotopic enrichment of retinol and ß-carotene in serum had reached a plateau when first measured on day 21. Therefore, we now plan to examine whether the plateau of isotopic enrichment is reached earlier.
Estimates of intake of several nutrients derived from the 24-h recalls were lower than the estimates from the dietary records, as reported earlier (34, 35). The main purpose of the 24-h recall method was to make the children aware of their food intake. Neither of the methods used provides reliable measurements of food intake at the individual level, but both were appropriate for assessing whether differences in food intake existed among the 3 groups during the intervention and between the screening and the intervention.
Dietary intake of ß-carotene was higher during the screening than during the intervention because we provided a low-retinol, low-carotenoid diet during the intervention, but there were no significant differences in serum ß-carotene concentrations between days 0 and 21. This can largely be explained by the low bioavailability of ß-carotene from food. For energy and all other nutrients, intakes were higher during the intervention than during the screening, indicating that in this age group dietary records may overestimate food intake.
This advanced stable-isotope technique can be extended to address the important question of the vitamin A potency of ß-carotene in fruit and vegetables. Although this method has not yet been tested for its ability to measure the bioefficacy of ß-carotene in fruit and vegetables, it is thought to be capable of doing so if the following design is used. Subjects would be studied for 2 consecutive periods of a maximum of 3 wk each. During both periods, the subjects would consume 2 or 3 capsules/d, 7 d/wk, and each capsule would contain small doses (maximum: 50 µg) of [13C10]ß-carotene and [13C10]retinyl palmitate. The subjects would be provided a low-retinol, low-carotenoid diet during both periods; during the second period, this diet would be supplemented with 2 or 3 portions of fruit or vegetables per day. Blood samples, preferably 2 fasting samples taken 2 d apart, would be drawn at baseline and at the end of each period. The degree of isotopic enrichment in serum retinol and ß-carotene would be measured as in this study. Then, the change in isotopic enrichment during the second period (compared with the first period) would be calculated. The extent of dilution in isotopic enrichment during the second period indicates the bioefficacy of ß-carotene in fruit or vegetables.
Because of the precision of this method, fewer subjects would be required to measure the bioefficacy of dietary ß-carotene than were required with the methods used previously. To illustrate this, we performed sample size calculations for a hypothetical study in which the vitamin A activity of ß-carotene in oil is assessed in 2 groups, one with high and one with low serum retinol concentrations. To determine which difference in bioefficacy can be expected, we can use the findings of Villard and Bates (36) that vitamin A–deficient rats had roughly a 27% higher ß-carotene 15,15'-dioxygenase activity than did control rats, indicating that vitamin A deficiency increases bioconversion. We can then calculate the minimum sample size required to detect a 27% difference between 2 groups in the vitamin A activity of ß-carotene in oil compared with that of retinol, with a power of 0.90 and an value of 0.05, with the following formula (37):
ACKNOWLEDGMENTS
We thank the Ministry of Health in Indonesia; the district health office in Bogor; the district and subdistrict offices of education and culture; Hendra Sutanto and the staff of the Health Centre in Situ Udik; the teachers and children of the primary school in Situ Udik; the village health volunteers in Situ Udik; Susilowati Herman, Susie Suwarti, Reviana Christiani, Yuniar Rosmalina, Yetty Yuniar, Pak Sulasono, and coworkers for assistance and laboratory analyses at the Nutrition Research and Development Centre in Bogor; Paul Hulshof, Peter van de Bovenkamp, and Truus Kosmeijer for laboratory analyses; Jan Harryvan for preparing the capsules; Jan Burema for assistance with the statistical analyses at the Division of Human Nutrition and Epidemiology at Wageningen University; and Khairul Amanah Triutami, Wiwit Wulandari, and Ondine van de Rest for assistance in the field.
REFERENCES