Daily consumption of Indian spinach ( alba) or sweet potatoes has a positive effect on total-body A stores in Bangladeshi men1–3

Marjorie J Haskell, Kazi M Jamil, Ferdaus Hassan, Janet M Peerson, M Iqbal Hossain, George J Fuchs, and Kenneth H Brown Downloaded from https://academic.oup.com/ajcn/article/80/3/705/4690550 by guest on 27 September 2021

ABSTRACT Ȃ44% (3). The World Health Organization recommends sup- Background: Recent evidence suggests that the equiv- plementing infants and children 5 y of age with high doses of alency of ␤-carotene from sources is lower than previously vitamin A to improve their vitamin A status (4). However, be- estimated. cause vitamin A is potentially teratogenic, high-dose vitamin A Objective: We assessed the effect of 60 d of daily supplementation supplements can be given safely to women of childbearing age ␮ with 750 g retinol equivalents (RE) of either cooked, puréed sweet only within the first 6 wk postpartum, when the likelihood of potatoes; cooked, puréed Indian spinach (Basella alba); or synthetic becoming pregnant is very low. ␤ sources of vitamin A or -carotene on total-body vitamin A stores in The provision of small daily doses of vitamin A from food may Bangladeshi men. be an alternative strategy for improving vitamin A status in pop- Design: Total-body vitamin A stores in Bangladeshi men (n ҃ ulations at risk of deficiency. Appropriate foods provide safe 14/group) were estimated by using the deuterated-retinol-dilution amounts of vitamin A and can be given to all population groups technique before and after 60 d of supplementation with either 0 ␮g at risk of deficiency, including women of childbearing age. An- RE/d (white vegetables) or 750 ␮g RE/d as sweet potatoes, Indian spinach, retinyl palmitate, or ␤-carotene (RE ҃ 1 ␮g retinol or 6 ␮g imal source foods, such as dairy foods, eggs, and liver, contain ␤-carotene) in addition to a low–vitamin A diet providing Ȃ200 ␮g preformed retinol, which is readily absorbed in the human intes- RE/d. Mean changes in vitamin A stores in the vegetable and tine; however, these foods are generally not affordable for pop- ␤-carotene groups were compared with the mean change in the ulations at risk of deficiency (5). In less-industrialized countries, retinyl palmitate group to estimate the relative equivalency of these 65–85% of vitamin A in the diet is estimated to be supplied by vitamin A sources. provitamin A carotenoids in vegetables and fruit (5). However, Results: Overall geometric mean (ȀSD) initial vitamin A stores recent evidence indicates that the bioavailability of provitamin A were 0.108 Ȁ 0.067 mmol. Relative to the low–vitamin A control carotenoids from plant sources is lower than previously assumed group, the estimated mean changes in vitamin A stores were 0.029 and suggests that plant sources of vitamin A may not be ef- mmol for sweet potato (P ҃ 0.21), 0.041 mmol for Indian spinach ficacious for improving vitamin A status (6, 7). In the most recent (P ҃ 0.033), 0.065 mmol for retinyl palmitate (P  0.001), and 0.062 edition of the Dietary Reference Intakes (8), the vitamin A mmol for ␤-carotene (P  0.002). Vitamin A equivalency factors equivalency factors for provitamin A carotenoids from foods (␤-carotene:retinol, wt:wt) were estimated as Ȃ13:1 for sweet po- were increased from 6:1 to 12:1 for ␤-carotene [12 ␮g tato, Ȃ10:1 for Indian spinach, and Ȃ6:1 for synthetic ␤-carotene. ␤-carotene ҃ 1 ␮g retinol ҃ 1 retinol acitivity equivalent (RAE)] Conclusion: Daily consumption of cooked, puréed green leafy veg- and from 12:1 to 24:1 for ␣-carotene, ␥-carotene, and etables or sweet potatoes has a positive effect on vitamin A stores in ␤-cryptoxanthin (24 ␮g other provitamin A carotenoids ҃ 1 ␮g populations at risk of vitamin A deficiency. Am J Clin Nutr 2004;80:705–14. 1 From the Program in International Nutrition, Department of Nutrition, University of California, Davis (MJH, JMP, and KHB), and the International KEY WORDS Deuterated retinol dilution, stable isotope, bio- Centre for Diarrhoeal Disease Research, Bangladesh (ICDDR,B): Centre for availability, vitamin A status, ␤-carotene, vitamin A stores, green Health and Population Research, Dhaka, Bangladesh (KMJ, FH, MIH, and leafy vegetables, sweet potatoes, Bangladesh GJF). 2 Supported by the US Department of Agriculture (grant no. 98-35200- 6099) and the Micronutrient Initiative (grant no. 5600-0001-04-300. Retinyl INTRODUCTION palmitate and ␤-carotene were donated by Roche (Parsippany, NJ), Vitamin A deficiency is a serious public health problem in and sweet potatoes were donated by the Allen Canning Company (Siloam low-income populations in less-industrialized populations (1). Springs, AR). 3 Reprints not available. Address correspondence to MJ Haskell, Univer- Young children and women of childbearing age are considered to sity of California, Program in International Nutrition, 3217A Meyer Hall, be at greatest risk of deficiency. In these populations, supple- One Shields Avenue, Davis, CA 95616. E-mail: [email protected]. mentation with vitamin A has been shown to reduce childhood Received September 12, 2003. mortality by Ȃ23% (2) and to reduce maternal mortality by Accepted for publication March 25, 2004.

Am J Clin Nutr 2004;80:705–14. Printed in USA. © 2004 American Society for Clinical Nutrition 705 706 HASKELL ET AL retinol ҃ 1 RAE). The equivalency factors were increased be- cause recent evidence indicates that the efficiency of absorption of ␤-carotene from foods is lower than previously estimated (Ȃ16% compared with Ȃ33%) (8–10). The effect of supplementation with plant sources of vitamin A on vitamin A status in humans has been assessed in placebo- controlled trials by examining changes in plasma retinol concen- tration in response to supplementation with plant or synthetic sources of vitamin A (6, 7). However, because of homeostatic regulation, plasma retinol concentrations are not likely to change in response to supplementation, unless subjects are moderately FIGURE 1. Study protocol. or severely vitamin A deficient at the onset of the intervention. Even when plasma retinol concentrations respond to supplemen- greater risk of vitamin A deficiency than are men, enrollment of tation in depleted persons, the magnitude of increase may not be Downloaded from https://academic.oup.com/ajcn/article/80/3/705/4690550 by guest on 27 September 2021 directly proportional to the vitamin A bioavailability from a women into the study was not possible because it is not culturally particular food source. In contrast, the deuterated-retinol- acceptable in Bangladesh for women to spend the time required dilution (DRD) technique is an indirect method for quantitatively to complete the study protocol (Ȃ12 h/d for 113 d) at the study estimating total body stores of vitamin A in humans (11), and the facility. Preschool-aged children were not enrolled because at the 2 technique has been validated in 2 sets of surgical patients with time the study was conducted the plasma kinetics of [ H4]retinol adequate to large (11) or small to adequate hepatic vitamin A had not been described in young children, and it was not known reserves (12). Moreover, the paired-DRD technique (estimation whether the isotope dilution equation that is used for estimating of vitamin A pool size before and after supplementation) pro- total-body vitamin A stores in adults would be appropriate for use vides expected quantitative estimates of change in total-body in that age group. Moreover, children in Bangladesh are sched- vitamin A stores in response to supplementation with different uled to receive periodic large-dose supplementation as part of a amounts of vitamin A (13). The purpose of the present study was national intervention program. to assess quantitative changes in total body stores of vitamin A by The subjects were ranked according to their initial plasma using the paired-DRD technique before and after 60 d of supple- retinol concentration and randomly assigned in blocks of 5 to 1 mentation with an orange tuber (sweet potatoes), a green leafy of 5 treatment groups. The subjects in each treatment group vegetable [Indian spinach (Basella alba); local name: pui sak], or received a low–vitamin A diet that was supplemented twice per an equivalent amount of synthetic vitamin A, which was pro- day, at the noon and evening meals, with either 1) low–vitamin vided as either retinyl palmitate or ␤-carotene in oil, to determine A vegetables (white potato, cauliflower) and a corn oil capsule [0 the relative efficacy of plant sources of vitamin A for improving ␮g retinol equivalents (RE)/d], 2) sweet potato (80 g to provide vitamin A status. 2.25 mg ␤-carotene, or 375 ␮g RE/meal) and a corn oil capsule, 3) Indian spinach (75 g to provide 2.25 mg ␤-carotene, or 375 ␮g RE/meal) and a corn oil capsule, 4) low–vitamin A vegetables SUBJECTS AND METHODS and a vitamin A capsule (375 ␮g RE/meal as 685 ␮g retinyl palmitate in corn oil), or 5) low–vitamin A vegetables and a Subjects ␤-carotene capsule (375 ␮g RE/meal as 2.25 mg ␤-carotene in The study was conducted at the outpatient facility of the In- corn oil). Thus, a total of 750 ␮g RE/d was provided in the ternational Centre for Diarrhoeal Disease Research in Dhaka, vitamin A–supplemented groups. The subjects who were as- Bangladesh. Subjects attended the facility daily from Ȃ0730 to signed to the low–vitamin A control group, hereafter referred to 1930 and consumed all of their meals and snacks under super- as the control group, received a capsule containing 60 mg vitamin vision. The study was conducted during 3 separate cycles be- A on completion of the study protocol. cause of space limitations at the study facility. Before each cycle, A vitamin A equivalency factor of 6:1 was used to calculate Ȃ75 men (18–35 y of age) were screened for plasma retinol the portion sizes of the vegetables (6 ␮g ␤-carotene ҃ 1 ␮g concentration. From this group, we selected those who had the retinol ҃ 1 RE) because this was the recommended vitamin A lowest plasma retinol concentrations; no clinical evidence of equivalency factor (14) at the time the study was conducted. vitamin A deficiency, intestinal malabsorption, or other condi- Using the new recommended vitamin A equivalency factor of tions that might interfere with vitamin A absorption or metabo- 12:1 (12 ␮g ␤-carotene ҃ 1 ␮g retinol ҃ 1 RAE) (8), the esti- lism; a serum albumin concentration Œ 35 g/L; and a serum mated vitamin A equivalency of the ␤-carotene–containing food C-reactive protein concentration  10 mg/L. Written informed or capsule supplements was 188 ␮g RAE/meal, for a total of 375 consent was obtained from each of the participants. The study ␮g RAE/d. protocol was approved by the Institutional Review Board of the Universtiy of California, Davis, and the Ethical Review Com- Study design: paired-DRD technique mittee of the International Centre for Diarrhoeal Disease Re- One week before beginning the study procedures, the subjects search, Bangladesh. received 800 mg albendezole (Smith Kline Beecham Pharma- We chose to study adult males in Bangladesh because we ceuticals, Philadephia) for treatment of intestinal helminths. know from a previous study (13) that it was possible to identify Three days before beginning the study procedures, the subjects persons in this population with low to adequate total body stores began receiving the basal low–vitamin A diet (described below) of vitamin A on the basis of their plasma retinol concentrations. (Figure 1). On study day 1, the subjects received an oral dose of 2 Although women and preschool-aged children are likely to be at 10 mg [ H4]retinyl acetate, which was followed by a high-, BIOAVAILABILITY OF ␤-CAROTENE FROM FOODS 707 low–vitamin A breakfast (fried curried potato pastries, tea). palmitate or all-trans-␤-carotene (Roche Vitamins, Parsippany, Twenty days later, a blood sample was drawn for measurement NJ). It was not possible to mask the sweet potatoes and Indian of plasma retinol and carotenoid concentrations (referred to as spinach, but the placebo, vitamin A, and ␤-carotene capsules “initial” concentrations) and of the plasma isotopic ratio of were identical in appearance. 2 [ H4]retinol to retinol for estimation of initial vitamin A pool size. For the next 60 d, the subjects received their assigned dietary Estimation of vitamin A pool size treatment (described below). At the end of the 60-d supplemen- Total body stores of vitamin A were estimated before and after tation period, a blood sample was drawn for measurement of the 60-d supplementation period by using plasma isotopic ratios postsupplementation plasma concentrations of retinol and caro- 2 that were measured 20 d after oral administration of [ H4]retinyl tenoids (referred to as “final” concentrations). Immediately acetate and the isotope dilution equation described by Furr et al thereafter, the subjects received the basal low–vitamin A diet for 2 (11). Because [ H4]retinyl acetate was used for the pool size a period of 10 d to allow the vitamin A that was consumed during estimates both before and after the period of dietary supplemen- the 60-d supplementation period to equilibrate with endogenous tation, the plasma isotopic ratio measured on study day 113 (to vitamin A stores. After the 10-d stabilization period (study day estimate final pool size) had to be adjusted for the contribution of Downloaded from https://academic.oup.com/ajcn/article/80/3/705/4690550 by guest on 27 September 2021 91), a blood sample was drawn for measurement of the plasma 2 2 isotope remaining from the first dose of [ H4]retinyl acetate that isotopic ratio of [ H4]retinol to retinol, and a second oral dose of 2 was administered for estimation of the initial pool size. This was 10 mg [ H4]retinyl acetate was administered to the subjects. accomplished by subtracting the plasma isotopic ratio of Blood was drawn 20 d later (study day 113) for measurement of 2 2 [ H4]retinol to retinol measured on day 92 from the isotopic ratio the plasma isotopic ratio of [ H4]retinol to retinol for estimation measured on day 113. The subjects were provided the basal of the final vitamin A pool size. low–vitamin A diet for 10 d before measurement of the plasma Diets isotopic ratio on day 92 to allow the supplemental nonlabeled vitamin A that was provided during the 60-d supplementation The basal low–vitamin A diet consisted of rice, wheat flat period to equilibrate with vitamin A stores. Thus, it is very un- bread, lentils, small amounts of curried chicken or mutton, and 2 likely that the plasma ratio of [ H4]retinol to retinol on day 92 was pale vegetables and fruit, such as cauliflower, cabbage, white affected by any residual nonlabeled vitamin A that was con- potatoes, white squash, and banana, all of which are very low in sumed during the supplementation period. The subjects were vitamin A content. On the basis of food-composition tables (15), given the low–vitamin A diet throughout the rest of the study Ȃ ␮ Ȃ the basal diet provided 200 g RE/d ( 100 RAE/d). The period to minimize the effect of nonlabeled dietary vitamin A on subjects were allowed to consume selected low–vitamin A and the plasma isotopic ratios that were measured on study day 113 low-fat foods, such as wheat flat bread, lentils, and white fruit and to estimate the final vitamin A pool size. vegetables ad libitum at the breakfast meal and at the midafter- noon snack to allow for differences in individual caloric require- Evaluation of abbreviated method for assessing ments. The meals were standardized to contain the same amount bioavailability of ␤-carotene of fat (Ȃ16 g/meal) and similar amounts of fiber (Ȃ29 g/meal; On the first day of the dietary supplementation period (study references 15–17). 2 Canned sweet potatoes (from the same production lot) were day 22), the subjects received an oral dose of [ H8]retinyl acetate (0, 5, or 10 mg) to determine whether the plasma isotopic ratios donated by the Allen Canning Company (Siloam Springs, AR) 2 and shipped to Bangladesh. Canned sweet potatoes were used of [ H8]retinol to retinol on days 3, 5, or 7 during the supple- because suitable local orange tubers and fruit were seasonal and mentation phase differed between the dietary treatment groups were not available in local markets during the full study period. and whether these plasma isotopic ratios could be used to predict Indian spinach is a low-cost green leafy vegetable that is avail- changes in vitamin A pool size in response to supplementation. able year round in Bangladesh. The Indian spinach was obtained The results of this component of the study will be presented separately. Final vitamin A pool sizes were estimated by using daily from the same supplier. 2 the plasma isotopic ratio of [ H4]retinol to unlabeled retinol; The sweet potatoes and Indian spinach were prepared to op- 2 timize the bioavailability of ␤-carotene by following standard- thus, any [ H8]retinol that remained in the body was not included ized recipes. The sweet potatoes were puréed by using an electric in the final estimate of total body stores of vitamin A. The final food processor and sautéed in corn oil for 5 min with onion, salt, pool size estimates reflect only the amount of unlabeled vitamin cardamom, and ground chili seeds. The Indian spinach was A in the body. steamed for 10 min, puréed, and sautéed in corn oil for 5 min with garlic, salt, and ground chili seeds. Each portion of vegetables Laboratory methods contained Ȃ6.8 g corn oil. The carotenoid content of the cooked, Plasma concentrations of retinol, lutein, ␣-carotene, puréed vegetables was measured by using HPLC (18) to deter- ␤-carotene, and ␣-tocopherol were measured by using HPLC mine the portion sizes required to supply 2.25 mg (375 ␮g RE) as (18) on a Shimadzu Class VP (Shimadzu, Columbia, MD) all-trans-␤-carotene. equipped with a photo-diode array detector and autosampler. Portions of sweet potatoes or Indian spinach were weighed Pre- and postsupplementation plasma samples for each subject onto tared plates, along with the other foods, at the noon and were analyzed together during the same set of HPLC analyses. evening meals. Subjects were supervised during mealtimes and For quality control, a plasma pool was prepared and calibrated by were asked to consume all of the food provided. Groups receiving using control serum (fat-soluble vitamins) from the National the vitamin A–containing foods received a placebo capsule (corn Institute of Standards (Gaithersburg, MD). Three aliquots of the oil) with each meal. Subjects in the vitamin A and ␤-carotene plasma pool were analyzed with each set of study samples. The groups received capsules containing 375 ␮g RE as either retinyl within-day CV for the measurements of retinol, lutein, 708 HASKELL ET AL

TABLE 1 Initial characteristics of study participants by treatment group1 Treatment group

Control Sweet potato Indian spinach Vitamin A ␤-Carotene Age (y) 23.6 Ȁ 3.2 21.6 Ȁ 1.9 22.4 Ȁ 3.7 21.2 Ȁ 1.8 22.6 Ȁ 3.1 Height (cm) 165 Ȁ 7 162 Ȁ 6 164 Ȁ 5 162 Ȁ 7 164 Ȁ 6 Weight (kg) 50.9 Ȁ 4.1 48.4 Ȁ 5.1 49.4 Ȁ 4.8 49.7 Ȁ 5.9 51.5 Ȁ 6.1 BMI (kg/m2) 18.6 Ȁ 1.3 19.0 Ȁ 2.9 18.4 Ȁ 1.3 18.8 Ȁ 1.6 19.1 Ȁ 1.8 Retinol (␮mol/L) 0.90 Ȁ 0.17 0.96 Ȁ 0.23 0.94 Ȁ 0.20 0.96 Ȁ 0.22 0.95 Ȁ 0.21 1 All values are x៮ Ȁ SD; n ҃ 14/group. There were no significant differences between the treatment groups, P Œ 0.21 (ANOVA).

␣-carotene, ␤-carotene, and ␣-tocopherol concentrations in the retinol concentration. Treatment group assignments were un- Downloaded from https://academic.oup.com/ajcn/article/80/3/705/4690550 by guest on 27 September 2021 plasma pool samples were ͨ5%, ͨ10%, ͨ10%, ͨ7%, and ͨ masked only after all of the statistical analyses were completed. 6%, respectively. To assess accuracy, control plasma from the All statistical analyses were performed by using SAS software National Institute of Standards was analyzed. The measured con- (release 6; SAS Institute Inc, Cary, NC). centrations of retinol, lutein, ␣-carotene, ␤-carotene, and ␣-tocopherol were within 3.1%, 3.4%, 9.1%, 4.3%, and 2.6% of the certified values for the control plasma. The all-trans-␤- RESULTS carotene content of the cooked, puréed foods (Indian spinach and Subjects sweet potato) was determined by using HPLC (Class VP; Shi- madzu) (18). Three ␤-carotene standards (Fluka Chemical Co, A total of 70 subjects who had plasma retinol concentrations Buchs, Switzerland) were saponified, extracted, and analyzed ranging from 0.52 to 1.25 ␮mol/L at the time of the screening with each set of food samples according to the same procedures procedures were enrolled in the study (14 subjects/group). The (18). The interday CV for the ␤-carotene concentration of the mean age, mean values for anthropometric characteristics, and standards was 8%, and the interday CV for the ␤-carotene mean plasma retinol concentrations of the study participants by concentration of the food samples was 12%. The plasma iso- treatment group at the time of screening are shown in Table 1. 2 2 There were no significant differences in mean age, height, topic ratios of [ H4]retinol to retinol and [ H8]retinol to retinol were determined by using gas chromatography–mass spectrom- weight, body mass index, or plasma retinol concentration be- etry as previously described (19). Briefly, retinol was isolated tween the treatment groups. from plasma by using HPLC, and the tert-butyldimethylsilyl ␤ derivative of retinol was formed. Isotopic ratios were measured -Carotene concentration in Indian spinach and sweet by using gas chromatography–mass spectrometry on a Shimadzu potatoes Ҁ QP 5000 quadrupole mass spectrometer with 1.12 ҂ 10 17 (70 On the basis of the HPLC analyses, the mean (ȀSD) all-trans- eV) electron ionization. Standards with known weight ratios of ␤-carotene concentration in cooked Indian spinach was 30.5 Ȁ 2 2 [ H4]retinol to retinol and [ H8]retinol to retinol were analyzed 2.1 ␮g/g, and the mean concentration of cis isomers of with each set of study samples. The interday CV for the ␤-carotene was Ȃ8.1 Ȁ 0.4 ␮g/g (Ȃ21% of total ␤-carotene). 2 2  [ H4]retinol:retinol and [ H8]retinol:retinol standards was 6%. The mean lutein content of the Indian spinach was 70.7 Ȁ 5.0 ␮g/g. The mean all-trans-␤-carotene concentration in cooked Statistical analysis sweet potatoes was 28.3 Ȁ 1.3 ␮g/g, and the mean concentration ␤ Ȃ Ȁ ␮ Ȃ Descriptive statistics were calculated for each variable. Vari- of cis isomers of -carotene was 6.0 0.3 g/g ( 17% of total ␤ ables that were not normally distributed were transformed to -carotene); lutein was not detected in sweet potatoes. ␣ ␤ natural logarithms for the statistical analyses. Analysis of covari- -Carotene and -cryptoxanthin were not detected in Indian ance was used to compare mean changes in outcome variables spinach or sweet potatoes. Portion sizes of cooked Indian spinach ␮ (total-body vitamin A stores, plasma concentrations of retinol (75 g/meal; 375 g RE) and cooked sweet potatoes (80 g/meal; ␮ and carotenoids) between treatment groups, with treatment 375 g RE) were determined by using the estimated all-trans- ␤ group as the main effect and the initial value of the outcome -carotene concentration in each vegetable. The vitamin A ac- ␤ variable as a covariate. For the comparison of mean changes in tivity of cis isomers of -carotene was not included in the deter- plasma retinol concentration between treatment groups, initial mination of portion sizes because of uncertainty about the ␤ ␤ values and the dose of [2H ]retinyl acetate administered to sub- bioavailability of cis- -carotene relative to all-trans- -carotene. 8 ␤ jects were used as covariates. In addition, a main effect for study If the vitamin A activity of cis isomers of -carotene is estimated ␤ cycle and an interaction between treatment group and study cycle to be one-half that of -carotene, the total vitamin A activity Ȃ ␮ Ȃ were included in each analysis to test whether pooling the study would be 426 gRE( 213 RAE) in a serving of Indian Ȃ ␮ Ȃ cycles was statistically justified. Linear regression was used to spinach and 415 gRE( 208 RAE) in a serving of sweet examine the relation between estimated change in vitamin A pool potatoes. size and initial vitamin A pool size, the relation between esti- mated change in vitamin A pool size and estimated change in Plasma retinol concentrations plasma ␤-carotene concentration, and the relation between esti- The overall initial mean (ȀSE) plasma retinol concentration mated change in plasma retinol concentration and initial plasma was 1.27 Ȁ 0.15 ␮mol/L. Initially, 13 (18.6%) of the subjects had BIOAVAILABILITY OF ␤-CAROTENE FROM FOODS 709

TABLE 2 Initial and final plasma concentrations of retinol, carotenoids, and ␣-tocopherol by treatment group1 Treatment group

Control Sweet Potato Indian spinach Vitamin A ␤-Carotene Retinol ␮mol/L Initial 1.35 Ȁ 0.08 1.24 Ȁ 0.13 1.32 Ȁ 0.15 1.15 Ȁ 0.19 1.29 Ȁ 0.19 Final 1.12 Ȁ 0.05a 1.24 Ȁ 0.06b 1.43 Ȁ 0.06b,c 1.24 Ȁ 0.07b 1.49 Ȁ 0.07c ␤-Carotene Initial 0.04 Ȁ 0.01 0.07 Ȁ 0.01 0.07 Ȁ 0.01 0.06 Ȁ 0.01 0.05 Ȁ 0.01 Final 0.04 Ȁ 0.01a 0.28 Ȁ 0.05b 0.34 Ȁ 0.05b 0.06 Ȁ 0.01a 0.58 Ȁ 0.05b Lutein Initial 0.08 Ȁ 0.04 0.09 Ȁ 0.03 0.10 Ȁ 0.02 0.07 Ȁ 0.01 0.08 Ȁ 0.02 Final 0.07 Ȁ 0.02a 0.07 Ȁ 0.02a 0.28 Ȁ 0.13b 0.07 Ȁ 0.02a 0.07 Ȁ 0.02a ␣-Carotene

Initial 0.03 Ȁ 0.01 0.04 Ȁ 0.02 0.04 Ȁ 0.01 0.04 Ȁ 0.00 0.03 Ȁ 0.01 Downloaded from https://academic.oup.com/ajcn/article/80/3/705/4690550 by guest on 27 September 2021 Final 0.02 Ȁ 0.01a 0.01 Ȁ 0.01a 0.09 Ȁ 0.01b 0.04 Ȁ 0.01a 0.04 Ȁ 0.01a ␣-Tocopherol Initial 13.5 Ȁ 2.2 13.8 Ȁ 1.3 14.0 Ȁ 1.2 12.5 Ȁ 1.5 13.2 Ȁ 1.3 Final 12.6 Ȁ 0.8a 13.3 Ȁ 0.7a 16.6 Ȁ 0.6b 12.6 Ȁ 0.7a 13.2 Ȁ 0.7a 1 ៮ Ȁ ҃ 2 All values are geometric x SE; n 14/group. For the comparison of plasma retinol concentrations, initial values and the dose of [ H8]retinyl acetate were covariates. Final values with different superscript letters are significantly different, P ͨ 0.03 (analysis of covariance with treatment group as the main effect and initial value as the covariate). plasma concentrations  1.05 ␮mol/L. One subject (1.4%) had final mean lutein concentration in the Indian spinach group was a plasma concentration  0.70 ␮mol/L. After supplementation significantly higher than the final mean concentrations in the the mean plasma retinol concentration decreased significantly in other treatment groups (P  0.0001). The overall initial mean the control group (P  0.0001) and increased significantly within ␣-carotene concentration was 0.04 Ȁ 0.01 ␮mol/L. The final the synthetic ␤-carotene group (P  0.0001). There were no mean ␣-carotene concentration in the Indian spinach group was significant changes in mean plasma retinol concentration in the significantly higher than the final mean concentrations in the other treatment groups (P ͧ 0.12). The final mean plasma retinol other treatment groups (P  0.0001). The overall initial mean concentrations in the sweet potato, Indian spinach, vitamin A, plasma ␣-tocopherol concentration was 13.4 Ȁ 1.5 ␮mol/L. The and ␤-carotene groups were significantly higher than the final final mean ␣-tocopherol concentration in the Indian spinach mean concentration in the control group (P  0.004) (Table 2). group was significantly higher than the final mean concentra- The final mean plasma retinol concentration in the ␤-carotene tions in the other treatment groups (P  0.0001). group was significantly higher than the final mean concentra- tions in the control, sweet potato, and vitamin A groups (P ͨ Mean change in estimated vitamin A pool size 0.03) but was not significantly different from the final mean Ȁ concentration in the Indian spinach group (P ҃ 0.17). (Note that The overall initial mean vitamin A pool size was 0.108 the subject with an initial plasma retinol concentration  0.70 0.067 mmol. The raw initial and final geometric mean vitamin A ␮mol/L was in the vitamin A group. His plasma retinol concen- pool sizes and mean changes in pool size are shown by treatment tration increased from 0.64 to 0.87 ␮mol/L, and his estimated group in Table 3. The adjusted mean change in vitamin A pool vitamin A pool size increased from 0.028 to 0.070 mmol.) For the size was estimated for each of the treatment groups by using comparison of final mean plasma retinol concentrations between analysis of covariance with control for initial values as described treatment groups, the initial plasma retinol concentrations and previously. The adjusted mean changes in vitamin A pool size in 2 the Indian spinach (0.022 mmol; P ҃ 0.034), vitamin A (0.046 the dose of [ H8]retinyl acetate that was administered on day 1 of  ␤  the supplementation period were used as covariates for adjust- mmol; P 0.001), and -carotene groups (0.043 mmol; P ment of any potential effect of these variables. (Any uncertainty 0.002) were significantly larger than the adjusted mean change in 2 the control group (Ҁ0.019 mmol). The adjusted mean change in regarding the effect of [ H8]retinyl acetate on final mean plasma retinol concentrations is not critical because vitamin A equiva- pool size in the sweet potato group (0.010 mmol) was larger than that in the control group, but the difference was not significant lency factors were estimated on the basis of relative changes in ҃ vitamin A pool size, as described below. Plasma retinol concen- (P 0.21) (Table 3). trations were not used to estimate vitamin A pool size or vitamin A equivalency factors.) Estimation of vitamin A equivalency factors Vitamin A equivalency factors for ␤-carotene from the vege- ␣ Plasma carotenoid and -tocopherol concentrations table or synthetic sources were estimated by comparing the mean The overall initial mean plasma ␤-carotene concentration was change in vitamin A pool size in each of these groups with that in 0.07 Ȁ 0.01 ␮mol/L, and initial concentrations did not differ the group who received vitamin A as retinyl palmitate (Table 3). significantly between treatment groups. The final mean As shown in Table 3 and as described above, the mean vitamin A ␤-carotene concentrations in the sweet potato, Indian spinach, pool size in the control group decreased in response to consump- and ␤-carotene groups were significantly higher than those in the tion of the low–vitamin A diet during the supplementation phase control and vitamin A groups (P  0.002) (Table 2). The overall of the study. Thus, to estimate the net mean change in pool size initial mean lutein concentration was 0.08 Ȁ 0.02 ␮mol/L. The in the groups who received a source of supplemental vitamin A 710 HASKELL ET AL

TABLE 3 Initial and final vitamin A pool sizes, changes in vitamin A pool size, and estimated vitamin A equivalency factors by treatment group1 Adjusted change in Estimated relative Initial Final Change in Adjusted change pool size relative to vitamin A Group pool size2 pool size2 pool size3 in pool size4 that in control group equivalency factor5 mmol mmol mmol mmol mmol Control 0.109 Ȁ 0.061 0.091 Ȁ 0.056 Ҁ0.018 Ҁ0.019a —— Sweet potato 0.095 Ȁ 0.044 0.106 Ȁ 0.056 0.011 0.010a,b 0.029 13.4 Indian spinach 0.115 Ȁ 0.087 0.138 Ȁ 0.107 0.023 0.022b 0.041 9.5 Vitamin A 0.096 Ȁ 0.068 0.139 Ȁ 0.061 0.043 0.046b 0.065 1.0 ␤-Carotene 0.123 Ȁ 0.075 0.168 Ȁ 0.079 0.045 0.043b 0.062 6.3 1 n ҃ 14/group. 2 Geometric x៮ Ȁ SD. Downloaded from https://academic.oup.com/ajcn/article/80/3/705/4690550 by guest on 27 September 2021 3 Geometric x៮. 4 Values with different superscript letters are significantly different, P  0.034 (analysis of covariance with initial value as the covariate). 5 Estimated on the basis of the following formula: (4500 ␮g ␤-carotene consumed per day/750 ␮g retinol equivalents retinyl palmitate consumed per day) ҂ (adjusted mean change in pool size in the vitamin A group relative to that in the control group/adjusted mean change in pool size in the vegetable or synthetic ␤-carotene group relative to that in the control group).

relative to the change in the control group, the adjusted mean size on the basis of these theoretical assumptions is 0.065 mmol, change in pool size in the control group (Ҁ0.019 mmol) was which is equal to the estimated change (0.065 mmol) on the basis subtracted from the adjusted mean change in pool size in each of of the paired-DRD technique. the supplemented groups (Table 3). Relative to the control group, the adjusted mean changes in pool size were 0.029 mmol for the Change in vitamin A pool size by initial vitamin A pool sweet potato group, 0.041 mmol for the Indian spinach group, size 0.065 mmol for the retinyl palmitate group, and 0.062 mmol for the ␤-carotene group. These estimates of the net mean change in To determine whether the observed changes in vitamin A pool pool size in the vegetable and ␤-carotene groups were compared size were related to initial pool size, linear regression analysis with the net mean change in pool size in the retinyl palmitate was used to compare the 2 variables (Figure 2). In the groups ␤ group to estimate the relative vitamin A equivalency of the veg- who received either synthetic vitamin A or -carotene, there was ҃Ҁ etables and synthetic ␤-carotene according to the following a significant negative relation (slope 0.48 (vitamin A), Ҁ ␤ ҃ equation: 0.45 ( -carotene); P 0.01 for each comparison) between initial pool size and percentage change in pool size, which indi- Relative vitamin A equivalency cates that the percentage change in pool size was greater when initial vitamin A pool size was small. This relation was not ϭ ␮ ␤ (4500 g -carotene consumed per day/ observed for the vegetable groups, in which the slopes were not 750 ␮g RE retinyl palmitate consumed per day) significantly different from zero, but the slopes for the vegetable groups were significantly different from those for the vitamin A ϫ (adjusted mean change in vitamin A pool size and ␤-carotene groups (P ҃ 0.004). in the retinyl palmitate group relative to that in the control group/adjusted mean change in vitamin A pool size in the vegetable or synthetic ␤-carotene group relative to that in the control group) (1)

Theoretical compared with observed change in vitamin A pool size in the retinyl palmitate group To determine whether the estimated change in vitamin A pool size in the group who received retinyl palmitate was reasonable, we calculated the expected change in pool size on the basis of the amount of vitamin A that the subjects received during the sup- FIGURE 2. Percentage change in vitamin A pool size by initial vitamin A pool size and treatment group. n ҃ 68. The slopes of the retinol and plementation period, with the assumptions that 50% of the daily ␤-carotene groups were significantly different from zero and from the slopes dose of vitamin A was retained and that the fractional catabolic of the placebo and spinach groups, P  0.05 (analysis of covariance with rate for vitamin A is 0.5%/d (20). The expected change in pool interactions for treatment group and initial vitamin A pool size). BIOAVAILABILITY OF ␤-CAROTENE FROM FOODS 711 equivalency factors, the relative change in plasma ␤-carotene in the vegetable groups was compared with that in the synthetic ␤-carotene group, as shown in Table 4, by using the following equation: Relative vitamin A equivalency ϭ (4500 ␮g ␤-carotene consumed per day/ 750 ␮g RE retinyl palmitate consumed per day) ϫ (mean adjusted change in plasma ␤-carotene in the ␤-carotene group/ FIGURE 3. Change in vitamin A pool size versus change in plasma

␤ Downloaded from https://academic.oup.com/ajcn/article/80/3/705/4690550 by guest on 27 September 2021 ␤-carotene concentration in the sweet potato, Indian spinach, and ␤-carotene mean adjusted change in plasma -carotene groups. n ҃ 38. The slopes were significantly different from zero (P ҃ 0.015) but did not differ significantly from one another (P ҃ 0.54) (analysis of in the vegetable groups) (2) covariance with interaction after control for initial vitamin A pool size). Using this procedure, the estimated vitamin A equivalency fac- tors for sweet potato and Indian spinach were 15.5:1 and 11.9:1, Change in plasma retinol concentration by initial plasma respectively (Table 4). These factors are similar to those that retinol concentration were obtained by comparing relative changes in vitamin A pool Overall, there was a significant negative relation between the size between the retinyl palmitate group and the vegetable groups change in plasma retinol concentration and initial plasma retinol [sweet potato: 15.5:1 (based on change in plasma ␤-carotene) concentrations (r ҃Ҁ0.33, P  0.001; after control for the dose compared with 13.4:1 (based on change in vitamin A pool size) 2 ҃ of [ H8]retinyl acetate and initial plasma retinol concentration), (P 0.96); Indian spinach: 11.9:1 (based on change in plasma but the slopes for this relation did not differ significantly between ␤-carotene) compared with 9.5:1 (based on change in vitamin A treatment groups (P ҃ 0.37). pool size) (P ҃ 0.62)]. However, the sample size was sufficient to detect only large differences in these ratios. Change in estimated vitamin A pool size compared with change in plasma ␤-carotene concentration There was a significant linear relation between change in DISCUSSION plasma ␤-carotene concentration and change in vitamin A pool The paired-DRD technique was used to assess quantitative size in the vegetable and ␤-carotene groups (r ҃ 0.49, P ҃ 0.015; changes in vitamin A pool size in response to supplementation Figure 3); as expected, no relation between these variables was with sweet potato, Indian spinach, retinyl palmitate, or observed for the retinyl palmitate and control groups, who re- ␤-carotene to determine the relative efficacy of plant sources of ceived very little, if any, ␤-carotene in their diets. Among the vitamin A for improving vitamin A status. The mean changes in vegetable and ␤-carotene groups, the slopes and intercepts of the vitamin A pool size in the Indian spinach, vitamin A, and regression lines did not differ significantly between treatment ␤-carotene groups were significantly larger than the mean groups (P ҃ 0.55); however, the sample size was sufficient to change in the control group. The mean change in vitamin A pool detect only large differences in these variables (a difference size in the sweet potato group was larger than that in the control of ͧ1.75 in slopes and a difference of ͧ0.058 mmol in vitamin group, but the difference was not significant. For the retinyl A pool size). As an alternative method for estimating vitamin A palmitate group, the expected theoretical change in vitamin A

TABLE 4 Initial and final plasma ␤-carotene concentrations, changes in plasma ␤-carotene concentration, and estimated vitamin A equivalency factors by treatment group1 Adjusted change in Estimated relative Initial Final Change in Adjusted change ␤-carotene relative to vitamin A Group ␤-carotene2 ␤-carotene2 ␤-carotene3 in ␤-carotene4 that in control group equivalency factor5 ␮mol/L ␮mol/L ␮mol/L ␮mol/L ␮mol/L Control 0.04 Ȁ 0.01 0.04 Ȁ 0.01 0.00 0.00a 0 — Sweet potato 0.07 Ȁ 0.01 0.28 Ȁ 0.05 0.21 0.21b 0.21 15.4 Indian spinach 0.07 Ȁ 0.01 0.34 Ȁ 0.05 0.27 0.27b 0.27 11.9 Vitamin A 0.06 Ȁ 0.01 0.06 Ȁ 0.01 0.00 0.00a 0 — ␤-carotene 0.05 Ȁ 0.01 0.58 Ȁ 0.05 0.53 0.54b 0.54 6.0 1 n ҃ 14/group. 2 x៮ Ȁ SE. 3 x៮. 4 Values with different superscript letters are significantly different, P  0.002 (analysis of covariance with initial value as the covariate). 5 Estimated on the basis of the following formula: (4500 ␮g ␤-carotene consumed per day/750 ␮g retinol equivalents retinyl palmitate consumed per day) ҂ (mean adjusted change in plasma ␤-carotene concentration in ␤-carotene group/mean adjusted change in plasma ␤-carotene in vegetable groups). 712 HASKELL ET AL an optimal indicator for assessing change in vitamin A status in response to supplementation. This is illustrated in the present study by the greater magnitude of response to supplementation for vitamin A pool size than for plasma retinol concentration. In the present study, Indian spinach was prepared to optimize the bioavailability of ␤-carotene by steaming, puréeing, and then sautéeing the spinach in oil (6.8 g oil/serving). This food- preparation method probably enhanced the bioavailability of ␤-carotene by lessening food matrix effects, which are known to reduce the bioavailability of carotenoids from green leafy veg- etables (21, 22). In the study in Indonesian schoolchildren (7), the food-preparation techniques were not described, and the fat con- tent of the ␤-carotene–containing food supplements was not

FIGURE 4. Mean (Ȁ SEM) percentage changes in plasma retinol con- specified. Downloaded from https://academic.oup.com/ajcn/article/80/3/705/4690550 by guest on 27 September 2021 centration (ᮀ) and estimated vitamin A pool size (o) in response to supple- Intestinal helminths (Ascaris lumbricoides) can have an ad- mentation by treatment group. n ҃ 68. The percentage change in plasma verse effect on ␤-carotene absorption when the intensity of in- retinol concentration and the percentage change in vitamin A pool size were Œ significantly different, P ҃ 0.0002 (nonparametric signed-rank test). The fection is high [ 3200 eggs/g (epg) feces] (23). In the present percentage changes in plasma retinol in the groups who received a vitamin A study, the subjects were treated with an antihelminthic drug be- source were significantly different from the percentage change in the control fore beginning the study procedures. In the Indonesian study, the group (P  0.004), and the percentage change in plasma retinol in the incidence of ascaris infection was Ȃ60%, the median intensity of ␤ -carotene group was significantly different from the percentage changes in infection was 4720 epg (25th–75th percentiles: 610–16 990 the vitamin A, sweet potato, and control groups (P ͨ 0.03) but was not significantly different from the percentage change in the Indian spinach epg), and the children were not treated with an antihelminthic. ҃ 2 group (P 0.17) (analysis of covariance with initial value and [ H8]retinyl There was a significant negative correlation between serum acetate dose as covariates). The percentage changes in vitamin A pool size in ␤-carotene concentrations and intensity of ascaris infection in the Indian spinach, vitamin A, and ␤-carotene groups were significantly children who received ␤-carotene from fruit but not in children different from the percentage change in the control group, P  0.04 (analysis ␤ of covariance with initial value as the covariate). who received -carotene from vegetables (7). Nevertheless, it is conceivable that the presence of roundworms reduced the overall absorption of ␤-carotene from dark green leafy vegetables. pool size was identical to the observed change in vitamin A pool In a related study, lactating Indonesian women were supple- size, which indicates that the paired-DRD technique provides mented for 12 wk with 3.5 mg ␤-carotene/d (0.583 RE/d, 0.292 reasonably good estimates of the change in pool size in response RAE/d) as a mixture of green leafy vegetables and carrots or as to supplementation, as was seen in a previous study (13). a ␤-carotene–enriched wafer in a placebo-controlled trial. The Final mean plasma retinol concentrations in the supplemented vegetable supplement contained 7.8 g fat/serving (6); however, groups were also significantly higher than the final mean con- there was almost no change in the serum ␤-carotene concentra- centration in the control group. Although plasma retinol concen- tion (0.03 ␮mol/L; 15% increase above initial value) in the trations responded to supplementation, the magnitude of change women who received the vegetables. By contrast, serum in plasma retinol concentration was less than was observed for ␤-carotene concentrations increased significantly (0.73 ␮mol/L; mean changes in vitamin A pool size (Figure 4). Thus, in this 384%) in women who received the ␤-carotene–enriched wafer study population, change in vitamin A pool size was a more containing the same amount of ␤-carotene and 4.4 g fat. This sensitive indicator for detecting a change in vitamin A status than suggests that ␤-carotene absorption was lower in the vegetable was change in plasma retinol concentration. group because of a food matrix effect. However, the incidence of Vitamin A equivalency factors were estimated as 13.4:1 for intestinal helminths in the Indonesian women was Ȃ80% [me- sweet potato and 9.5:1 for Indian spinach. In Indonesian school- dian intensity: 13 020 epg (25th–75th percentiles: 1580–40 540 children, a vitamin A equivalency factor of 12:1 was reported for epg)]. The adverse effect of intestinal helminths on ␤-carotene orange and yellow fruit and vegetables, and a factor of Ȃ26:1 was absorption may be greater when ␤-carotene is in a complex food reported for green leafy vegetables (7). The value of 13:1 for matrix than when it is in a simpler matrix. In the present study, sweet potato in the present study is similar to the value of 12:1 for plasma ␤-carotene concentrations increased in response to sup- orange fruit and vegetables; however, the value of 10:1 for Indian plementation with sweet potato (0.21 ␮mol/L; 300%) and Indian spinach is much less than the factor of 26:1 for dark green leafy spinach (0.27 ␮mol/L; 385%), which suggests that the food- vegetables in the Indonesian study. There are several factors that processing techniques used or the treatment for intestinal hel- may account for the different estimates of vitamin A equivalency minths enhanced ␤-carotene absorption in the study subjects. factors for dark green leafy vegetables. In particular, the studies In the present study, a vitamin A equivalency factor of 6.3:1 differed in 1) the methods that were used for assessing change in was estimated for synthetic ␤-carotene in corn oil, which is vitamin A status, 2) food preparation techniques, and 3) the higher than previous estimates of Ȃ2:1 to 4:1 (24–27). The treatment of intestinal helminths. earlier estimates (2:1 and 3.3:1) were derived from comparisons In the present study, vitamin A equivalency factors were es- of the amount of ␤-carotene or vitamin A that was required to timated on the basis of relative quantitative changes in vitamin A reverse abnormal dark adaptation in a small number of vitamin pool size, whereas in the Indonesian study (7), these factors were A–depleted subjects (24, 25). More recent estimates (3.8:1 and estimated on the basis of relative changes in serum retinol con- 2.4:1) are based on stable-isotope methods for estimating ab- centration in response to supplementation with plant or synthetic sorption and bioconversion of ␤-carotene to retinol (26, 27). One sources of vitamin A. As mentioned earlier, serum retinol is not possible explanation for the higher estimate of 6:1 in the present BIOAVAILABILITY OF ␤-CAROTENE FROM FOODS 713 study is that the daily doses of ␤-carotene were given with a meal present study. The relative mean changes in plasma ␤-carotene that contained Ȃ29 g dietary fiber, which may have reduced concentration in response to consumption of equivalent amounts ␤-carotene absorption. of synthetic ␤-carotene or ␤-carotene from food sources may be an alternative method for estimating vitamin A equivalency fac- Relation between percentage change in vitamin A pool tors in populations with low to adequate initial vitamin A pool size and initial vitamin A pool size sizes; however, this requires further investigation. In summary, the results of the present study indicate that daily In the present study, the percentage change in vitamin A pool consumption of green leafy vegetables, prepared as described, size was negatively related to initial vitamin A pool size in the has a positive effect on vitamin A status in Bangladeshi men. groups who received synthetic vitamin A or ␤-carotene, which Further research is needed to assess the effects of food- suggests that subjects with low vitamin A status may be more preparation techniques, intestinal parasites, and initial vitamin A responsive to treatment with vitamin A than are subjects with status on the efficacy of plant sources of vitamin A for improving higher vitamin A status. It has been suggested previously that vitamin A status. vitamin A status may affect the absorption and bioconversion of ␤-carotene to vitamin A. A few animal studies have shown that We thank the study participants for their efforts in completing the study Downloaded from https://academic.oup.com/ajcn/article/80/3/705/4690550 by guest on 27 September 2021 bioconversion of ␤-carotene to vitamin A decreases when vita- protocol. min A intake increases (28–30). In contrast, the activity of in- MJH contributed to the study design, the training of field personnel, data testinal ␤-carotene 15,15'-oxygenase in rats is not affected by collection and analysis, and the writing of the manuscript. KMJ managed the depletion or excess feeding of ␤-carotene or retinol (31). Filipino clinical phase of the study and contributed to the training of field personnel, data collection, and the writing of the manuscript. FH and MIH each con- children with the lowest vitamin A status before intervention tributed to data collection during the clinical phase of the study. JMP con- showed the greatest improvement in vitamin A status in response tributed to the study design, statistical analyses, and the writing of the manu- to supplementation with provitamin A–rich fruit and vegetables. script. GJF contributed to the management of the clinical phase of the study. The inverse correlation was much stronger when vitamin A status KHB contributed to the study design, data analysis, and the writing of the 2 was assessed by using serum ratios of [ H4]retinol to retinol 3 d manuscript. None of the authors had any financial or personal interests in after the isotope dose than when serum retinol concentrations either of the 2 agencies that supported this study. were used (32). This suggests that absorption and bioconversion of ␤-carotene to vitamin A are greater when the initial vitamin A status is low. However, the disposal rate of vitamin A is known REFERENCES to vary with vitamin A status in rats (33). The lower retinol 1. World Health Organization. Global prevalence of vitamin A deficiency. Geneva: World Health Organization, 1995. disposal rate in rats with small vitamin A reserves presumably 2. Beaton GH, Martorell R, L’Abbe KA, et al. Effectiveness of vitamin A conserves the vitamin, whereas the higher disposal rate in rats supplementation in the control of young child morbidity and mortality in with large reserves prevents too much vitamin A from accumu- developing countries. Nutrition policy discussion paper no. 13. Geneva: lating in the body. Thus, in the present study, subjects with small ACC/SCN, 1993. ␤ 3. West K, Katz J, Khatry S, et al. Double blind, cluster randomised trial of initial vitamin A reserves who received synthetic -carotene may ␤ ␤ low dose supplementation with vitamin A or -carotene on mortality have absorbed and bioconverted more -carotene to vitamin A related to pregnancy in Nepal. BMJ 1999;318:570–5. and disposed of vitamin A to a lesser extent than did subjects with 4. World Health Organization/International Vitamin A Consultative large initial vitamin A reserves. In the vegetable groups, a rela- Group. Vitamin A supplements: a guide to their use in the treatment and tion between percentage change in vitamin A pool size and initial prevention of vitamin A deficiency. Geneva: World Health Organiza- tion, 2000. vitamin A pool size was not observed. One possible explanation 5. Underwood BA. Vitamin A in human and animal nutrition. In: Sporn M, for this is that less ␤-carotene was absorbed from the vegetables Roberts A, Goodman D, eds. The retinoids. New York: Academic Press, than from the synthetic ␤-carotene capsules because of food 1984:281–392. matrix effects. This seems likely because the plasma ␤-carotene 6. de Pee S, West CE, Muhilal, Karyadi D, Hautvast JGAJ. Lack of im- responses and mean changes in vitamin A pool size in the veg- provement in vitamin A status with increased consumption of dark-green leafy vegetables. Lancet 1995;346:75–81. etable groups tended to be lower than those in the synthetic 7. de Pee S, West C, Permaesih D, Martuti S, Muhilal, Hautvast J. Orange ␤-carotene group. Because stores increased to a lesser extent in fruit is more effective than are dark-green, leafy vegetables in increasing the vegetable groups, and the range of changes in pool size was serum concentrations of retinol and ␤-carotene in schoolchildren in smaller, it may not have been possible to detect a differential Indonesia. Am J Clin Nutr 1998;68:1058–67. 8. Institute of Medicine. Dietary reference intakes: vitamin A, vitamin K, response in percentage change in vitamin A pool size by initial arsenic, boron, chromium, copper, iodine, iron, manganese, molybde- pool size in these groups. num, nickel, silicon, vanadium, and . Washington, DC: National Academy Press, 2001. Relation between estimated vitamin A pool size and 9. Van het Hof K, Brouwer I, West C, et al. Bioavailability of lutein from ␤ vegetables is five times higher than that of ␤-carotene. Am J Clin Nutr plasma -carotene concentrations 1999;70:261–8. Changes in plasma ␤-carotene concentration were signifi- 10. Micozzi MS, Brown ED, Edwards BK, et al. Plasma carotenoid response ␤ cantly related to changes in vitamin A pool size. Estimates of to chronic intake of selected foods and -carotene supplements in men. Am J Clin Nutr 1992;55:1120–5. vitamin A equivalency factors based on relative changes in 11. Furr HC, Amedee-Manesme O, Clifford AJ, et al. Vitamin A concen- plasma ␤-carotene concentration between the synthetic trations in liver determined by isotope dilution assay with tetradeuterated ␤-carotene group and the vegetable groups were similar to those vitamin A and by biopsy in generally healthy adult humans. Am J Clin based on relative changes in vitamin A pool size between the Nutr 1989;49:713–6. 12. Haskell MJ, Handelman GJ, Peerson JM, et al. Assessment of vitamin A retinyl palmitate group and the vegetable groups. This suggests status by the deuterated retinol dilution technique and comparison with that absorption and bioconversion of ␤-carotene to retinol are hepatic retinol concentration in Bangladeshi surgical patients. Am J Clin directly related within the range of pool sizes observed in the Nutr 1997;66:67–74. 714 HASKELL ET AL

13. Haskell M, Mazumder R, Peerson J, et al. Use of the deuterated retinol concentrations in children are affected by food sources of ␤-carotene, fat dilution (DRD) technique to assess total body vitamin A stores of adult intake, and anthelmintic drug treatment. Am J Clin Nutr 1998;68:623–9. volunteers consuming different levels of vitamin A. Am J Clin Nutr 24. Hume EM, Krebs HA. Vitamin A requirement of human adults. Medical 1999;70:874–80. Research Council special report series, no. 264. London: Her Majesty’s 14. National Research Council, Committee on Dietary Allowances. Recom- Stationery Office, 1949:1–145. mended dietary allowances. Washington, DC: National Academy of 25. Sauberlich HE, Hodges HE, Wallace DL, et al. Vitamin A metabolism Sciences, 1980. and requirements in the human studied with the use of labelled retinol. 15. Darnton-Hill I, Hassan N, Karim R, Duthie MR. Tables of nutrient Vitam Horm 1974;32:251–75. composition of Bangladesh foods. Dhaka, Bangladesh: Helen Keller 26. Tang G, Qin J, Dolnikowski G, Russell R. Vitamin A equivalence of International, 1988. ␤-carotene in a woman as determined by a stable isotope reference 16. Adams C. Nutritive value of American foods. Agriculture handbook no. method. Eur J Nutr 2000;39:7–11. 456. Washington, DC: Agricultural Research Service, US Department of 27. van Lieshout M, West CE, Permaesih D, et al. Bioefficacy of ␤-carotene Agriculture, 1988. dissolved in oil studied in children in Indonesia. Am J Clin Nutr 2001; 17. Spiller G. CRC handbook of dietary fiber in human nutrition. Boca 73:949–58. Raton, FL: CRC Press, 1986. 28. van Vliet T, van Vlissingen M, van Schaik F, van den Berg H. Beta- 18. Handelman G, Shen B, Krinsky N. High resolution analysis of carote- carotene absorption and cleavage in rats is affected by the vitamin A

noids in human plasma by high-performance liquid chromatography. Downloaded from https://academic.oup.com/ajcn/article/80/3/705/4690550 by guest on 27 September 2021 concentration of the diet. J Nutr 1996;126:499–508. Methods Enzymol 1992;213:336–46. 19. Handelman GJ, Haskell MJ, Jones AD, Clifford AJ. An improved pro- 29. Dua P, Day E, Tipton H, Hill J. Influence of dietary vitamin A on tocol for determining ratios of retinol-d4 to retinol isolated from human carotenoid utilization, nitrogen retention and energy utilization by the plasma. Anal Chem 1993;65:2024–8. chick. J Nutr 1966;90:117–22. 20. Olson JA. Recommended dietary intakes (RDI) of vitamin A in humans. 30. Villard L, Bates C. Carotene dioxygenase activity in rat intestine: effects Am J Clin Nutr 1987;45:704–16. of vitamin A deficiency and of pregnancy. Br J Nutr 1986;56:115–22. 21. Castenmiller J, West C, Linssen J, van het Hof K, Voragen A. The food 31. Gronowska-Senger A, Wolf G. Effect of dietary protein on the enzyme matrix of spinach is a limiting factor in determining the bioavailability of from rat and human intestine which converts beta-carotene to retinal. J ␤-carotene and to a lesser extent of lutein in humans. J Nutr 1999;129: Nutr 1970;100:300–8. 349–55. 32. Ribaya-Mercado J, Solon F, Solon M, et al. Bioconversion of plant 22. Rock C, Lovalvo J, Emenhiser C, Ruffin M, Flatt S, Schwartz S. Bio- carotenoids to vitamin A in Filipino school-aged children varies in- availability of ␤-carotene is lower in raw than in processed carrots and versely with vitamin A status. Am J Clin Nutr 2000;72:455–65. spinach in women. J Nutr 1998;128:913–6. 33. Green MH, Green JB, Lewis KC. Variation in retinol utilization rate with 23. Jalal F, Nesheim M, Zulkarnain A, Sanjur D, Habicht J. Serum retinol vitamin A status in the rat. J Nutr 1987;117:694–703.