Absorption and Bioconversion

Absorption of Vitamin A. Intestinal absorption of preformed vitamin A occurs following the processing of retinyl esters in the lumen of the small intestine. Within the water-miscible micelles formed from bile salts, solubilized retinyl esters as well as triglycerides are hydrolyzed to retinol and products of lipolysis by various hydrolases (Harrison, 1993). A small percentage of dietary retinoids is converted to retinoic acid in the intestinal cell. In addition, the intestine actively synthesizes retinoyl β-glucuronide that is hydrolyzed to retinoic acid by β-glucuronidases (Barua and Olson, 1989). The efficiency of absorption of preformed vitamin A is generally high, in the range of 70 to 90 percent (Sivakumar and Reddy, 1972). A specific retinol transport protein within the brush border of the enterocyte facilitates retinol uptake by the mucosal cells (Dew and Ong, 1994). At physiological concentrations, retinol absorption is carrier mediated and saturable, whereas at high pharmacological doses, the absorption of retinol is nonsaturable (Hollander and Muralidhara, 1977). As the amount of ingested preformed vitamin A increases, its absorbability remains high (Olson, 1972). Vitamin A absorption and intestinal retinol esterification are not markedly different in the elderly compared to young adults, although hepatic uptake of newly absorbed vitamin A in the form of retinyl ester is slower in the elderly (Borel et al., 1998).

Absorption and Bioconversion of Provitamin A Carotenoids. Carotenoids are also solubilized into micelles in the intestinal lumen from which they are absorbed into duodenal mucosal cells by a passive diffusion mechanism. Percent absorption of a single dose of 45 μg to 39 mg β-carotene, measured by means of isotopic methods, has been reported to range from 9 to 22 percent (Blomstrand and Werner, 1967; Goodman et al., 1966; Novotny et al., 1995). However, the absorption efficiency decreases as the amount of dietary carotenoids increases (Brubacher and Weiser, 1985; Tang et al., 2000). The relative carotene concentration in micelles can vary in response to the physical state of the carotenoid (e.g., whether it is dissolved in oil or associated with plant matrix materials). A number of factors affect the bioavailability and bioconversion of carotenoids (Castenmiller and West, 1998). Carotene bioavailability can differ with different processing methods of the same foods and among different foods containing similar levels of carotenoids (Boileau et al., 1999; Hume and Krebs, 1949; Rock et al., 1998; Torronen et al., 1996; Van den Berg and van Vliet, 1998) (also see Dietary Reference Intakes for Vitamin C, Vitamin E, Selenium, and Carotenoids [IOM, 2000]).

Absorbed β-carotene is principally converted to vitamin A by the enzyme β-carotene-15, 15′-dioxygenase within intestinal absorptive cells. The central cleavage of β-carotene by this enzyme will, in theory, result in two molecules of retinal. β-Carotene can also be cleaved eccentrically to yield β-apocarotenals that can be further degraded to retinal or retinoic acid (Krinsky et al., 1993). The predominant form of vitamin A in human lymph, whether originating from ingested vitamin A or provitamin A carotenoids, is retinyl ester (retinol esterified with long-chain fatty acids, typically palmitate and stearate) (Blomstrand and Werner, 1967; Goodman et al., 1966). Along with exogenous lipids, the newly synthesized retinyl esters and nonhydrolyzed carotenoids are transported from the intestine to the liver in chylomicrons and chylomicron remnants. Derived from dietary retinoids, retinoic acid is absorbed via the portal system bound to albumin (Blaner and Olson, 1994; Olson, 1991).

Vitamin A Activity of Provitamin A Carotenoids: Rationale for Developing Retinol Activity Equivalents. The carotene:retinol equivalency ratio (μg:μg) of a low dose (less than 2 mg) of purified β-carotene in oil is approximately 2:1 (i.e., 2 μg of β-carotene in oil yields 1 μg of retinol) (Table 4-1). This ratio was derived from the relative amount of β-carotene required to correct abnormal dark adaptation in vitamin A-deficient individuals (Hume and Krebs, 1949; Sauberlich et al., 1974). The data by Sauberlich et al. (1974) were given greater consideration because (1) the actual amount (μg) of vitamin A and β-carotene consumed was cited, (2) varied amounts of vitamin A or β-carotene were consumed by each individual, and (3) a greater sample size was employed (six versus two subjects). In addition to these studies, an earlier study by Wagner (1940) estimated a carotene:retinol equivalency ratio of 4:1; however, the method employed for measuring dark adaptation was not standardized and used an imprecise outcome measure.

TABLE 4-1

Relative Absorption of Vitamin A and Supplemental β-Carotene.

Studies have been performed to compare the efficiency of absorption of β-carotene after feeding physiological amounts of β-carotene in oil, in individual foods, and as part of a mixed vegetable and fruit diet. Many of the earlier studies analyzed the fecal content of β-carotene after the consumption of a supplement, fruit, or vegetable. Data from these studies were not considered because the portion of unabsorbed β-carotene that is degraded by the intestinal microflora is not known. The efficiency of absorption of β-carotene in food is lower than the absorption of β-carotene in oil by a representative factor of a. Assuming that after absorption of β-carotene, whether from oil or food, the metabolism of the molecule is similar and that the retinol equivalency ratio of β-carotene in oil is 2:1, the vitamin A activity of β-carotene from food can be derived by multiplying a by 2:1.

Until recently it was thought that 3 μg of dietary β-carotene was equivalent to 1 μg of purified β-carotene in oil (NRC, 1989) due to a relative absorption efficiency of about 33 percent of β-carotene from food sources. Only one study has compared the relative absorption of β-carotene in oil versus its absorption in a principally mixed vegetable diet in healthy and nutritionally adequate individuals (Van het Hof et al., 1999). This study concluded that the relative absorption of β-carotene from the mixed vegetable diet compared to β-carotene in oil is only 14 percent, as assessed by the increase in plasma β-carotene concentration after dietary intervention. Based on this finding, approximately 7 μg of dietary β-carotene is equivalent to 1 μg of β-carotene in oil. This absorption efficiency value of 14 percent is supported by the relative ranges in β-carotene absorption reported by others using similar methods for mixed green leafy vegetables (4 percent) (de Pee et al., 1995), carrots (18 to 26 percent) (Micozzi et al., 1992; Torronen et al., 1996), broccoli (11 to 12 percent) (Micozzi et al., 1992), and spinach (5 percent) (Castenmiller et al., 1999) (Table 4-2).

TABLE 4-2

Relative Absorption of Supplemental and Dietary β-Carotene.

Only one study has been published to assess the relative bioconversion of β-carotene from fruits versus vegetables by measuring the rise in serum retinol concentration after the provision of a diet high in vegetables, fruits, or retinol (de Pee et al., 1998). This study used methods similar to those employed by other researchers (Castenmiller et al. [1999], de Pee et al. [1995], Micozzi et al. [1992], Torronen et al. [1996], and Van het Hof et al. [1999]), and indicated that the vitamin A activity was approximately half the activity for dark, green leafy vegetables compared to equal amounts of β-carotene from orange fruits and some yellow tubers, such as pumpkin squash (de Pee et al., 1998) (Table 4-2). Because of the low content of fruits contained in the principally mixed vegetable diet of Van het Hof et al. (1999) and the low proportion of dietary β-carotene that is consumed from fruits compared to vegetables in the United States (16 percent from the 14 major dietary contributors of β-carotene which provide a total of 70 percent of dietary β-carotene) (Chug-Ahuja et al., 1993), it is estimated that 6 μg, rather than 7 μg, of β-carotene from a mixed diet is nutritionally equivalent to 1 μg of β-carotene in oil. Therefore, the retinol activity equivalency (μg RAE) ratio for β-carotene from food is estimated to be 12:1 (6 × 2:1) (Figure 4-2). Unfortunately, studies using a positive control group (preformed vitamin A) at a level equivalent to β-carotene from a mixed vegetable and fruit diet using levels similar to the RAE have not been conducted in healthy and nutritionally adequate individuals. An RAE of 12 μg for dietary β-carotene is supported by Parker et al. (1999) who reported that 8 percent of ingested β-carotene from carrots was absorbed and converted to retinyl esters contained in chylomicrons, resulting in a carotene:retinol equivalency ratio of 13:1.

FIGURE 4-2

Absorption and bioconversion of ingested provitamin A carotenoids to retinol based on new equivalency factors (retinol activity equivalency ratio).

One RAE for dietary provitamin A carotenoids other than β-carotene is set at 24 μg on the basis of the observation that the vitamin A activity of β-cryptoxanthin and α-carotene is approximately half of that for β-carotene (Bauernfeind, 1972; Deuel et al., 1949). Therefore, the amount of vitamin A activity of provitamin A carotenoids in μg RAE is half the amount obtained if using μg RE (Table 4-3).

TABLE 4-3

Comparison of the 1989 National Research Council and 2001 Institute of Medicine Interconversion of Vitamin A and Carotenoid Units.

Example: A diet contains 500 μg retinol, 1,800 μg β-carotene and 2,400 μg α-carotene.

500 + (1,800 ÷ 12) + (2,400 ÷ 24) = 750 μg RAE.

Example: A diet contains 1,666 IU of retinol and 3,000 IU of β-carotene.

(1,666 ÷ 3.33) + (3,000 ÷ 20) = 650 μg RAE.

Example: A supplement contains 5,000 IU of vitamin A (20 percent as β-carotene).

5,000 ÷ 3.33 = 1,500 μg RAE.

The use of μg RAE rather than μg RE or international units (IU) is preferred when calculating and reporting the amount of the total vitamin A in mixed foods or assessing the amount of dietary and supplemental vitamin A consumed. Given the need to be able to calculate the intake of carotenoids, food composition data tables should report food content in amounts of each carotenoid whenever possible.

Metabolism, Transport, and Excretion

Retinyl esters and carotenoids are transported to the liver in chylomicron remnants. Apoprotein E is required for the uptake of chylomicron remnants by the liver. Some retinyl esters can also be taken up directly by peripheral tissues (Goodman et al., 1965). Several specific hepatic membrane receptors (low density lipoprotein [LDL] receptor, LDL receptor-related protein, lipolysis-stimulated receptor) have been proposed to also be involved with the uptake of chylomicron remnants (Cooper, 1997). The hydrolysis of retinyl ester to retinol is catalyzed by retinyl ester hydrolase following endocytosis. To meet tissue needs for retinoids, retinol binds to retinol-binding protein (RBP) for release into the circulation. In the blood, holo-RBP associates with transthyretin (a transport protein) to form a trimolecular complex with retinol in a 1:1:1 molar ratio. Retinol is transported in this trimolecular complex to various tissues, including the eye. The mechanism through which retinol is taken up from the circulation by peripheral cells has not been conclusively established. Retinol that is not immediately released into circulation by the liver is reesterified and stored in the lipid-containing stellate (Ito) cells of the liver until needed to maintain normal blood retinol concentrations.

Carotenoids are incorporated into very low density lipoproteins (VLDL) and exported from the liver into the blood. VLDL are converted to LDL by lipoprotein lipase on the surface of blood vessels. Plasma membrane-associated receptors of peripheral tissue cells bind apolipoprotein B100 on the surface of LDL, initiating receptor-mediated uptake of LDL and their lipid contents. The liver, lung, adipose, and other tissues possess carotene 15, 15′-dioxygenase activity (Goodman and Blaner, 1984; Olson and Hayaishi, 1965), and thus it is presumed that carotenes may be converted to vitamin A as they are delivered to tissues. The major end products of the enzyme's activity are retinol and retinoic acid (Napoli and Race, 1988). It is unclear, however, whether carotenoids stored in tissues other than the intestinal mucosa cells are cleaved to yield retinol. Thatcher et al. (1998) demonstrated that β-carotene stored in liver is not utilized for vitamin A needs in gerbils.

Typically, the majority of vitamin A metabolites are excreted in the urine. Sauberlich et al. (1974) reported that the percentage of a radioactive dose of vitamin A recovered in breath, feces, and urine ranged from 18 to 30 percent, 18 to 37 percent, and 38 to 60 percent, respectively, after 400 days on a vitamin A-deficient diet. Almost all of the excreted metabolites are biologically inactive.

Retinol is metabolized in the liver to numerous products, some of which are conjugated with glucuronic acid or taurine for excretion in bile (Sporn et al., 1984). The portion of excreted vitamin A metabolites in bile increases as the liver vitamin A exceeds a critical concentration. This increased excretion has been suggested to serve as a protective mechanism for reducing the risk of excess storage of vitamin A (Hicks et al., 1984).

Body Stores

The hepatic vitamin A concentration can vary markedly depending on dietary intake. When vitamin A intake is adequate, over 90 percent of total body vitamin A is located in the liver (Raica et al., 1972) as retinyl ester (Schindler et al., 1988), where it is concentrated in the lipid droplets of perisinusoidal stellate cells (Hendriks et al., 1985). The average concentration of vitamin A in postmortem livers of American and Canadian adults is reported to range from 10 to as high as 1,400 μg/g liver (Furr et al., 1989; Hoppner et al., 1969; Mitchell et al., 1973; Raica et al., 1972; Schindler et al., 1988; Underwood et al., 1970). In developing countries where vitamin A deficiency is prevalent, the vitamin A concentration in liver biopsy samples is much lower (17 to 141 μg/g) (Abedin et al., 1976; Flores and de Araujo, 1984; Haskell et al., 1997; Olson, 1979; Suthutvoravoot and Olson, 1974). A concentration of at least 20 μg retinol/g of liver in adults is suggested to be the minimal acceptable reserve (Loerch et al., 1979; Olson, 1982). The mean liver stores of vitamin A in children (1 to 10 years of age) have been reported to range from 171 to 723 μg/g (Flores and de Araujo, 1984; Mitchell et al., 1973; Money, 1978; Raica et al., 1972; Underwood et al., 1970), whereas the mean liver vitamin A stores in apparently healthy infants is lower, ranging from 0 to 320 μg/g of liver (Flores and de Araujo, 1984; Huque, 1982; Olson et al., 1979; Raica et al., 1972; Schindler et al., 1988).

With use of radio-isotopic methods, the efficiency of storage (retention) of vitamin A in liver has been estimated to be approximately 50 percent (Bausch and Rietz, 1977; Kusin et al., 1974; Sauberlich et al., 1974). More recently, stable-isotopic methods have shown an efficiency of storage of 42 percent for individuals with concentrations greater than or equal to 20 μg retinol/g of liver (Haskell et al., 1997). The efficiency of storage was lower in those with lower vitamin A status. The percentage of total body vitamin A stores lost per day was approximately 0.5 percent in adults consuming a vitamin A-free diet (Sauberlich et al., 1974).

Dark Adaptation Test

To perform a dark adaptation test, the eye is first dilated and the subject fixates on a point located approximately 15 degrees above the center of the test light. The test stimulus consists of light flashes of approximately 1-second duration separated by 1-second intervals of darkness. A tracking method is used with the luminance of the test light being increased or decreased depending upon the response of the subject. The ascending threshold is the intensity at which the subject first sees the test light as its luminance is increased. The descending threshold is the intensity at which the subject ceases to see the test light as its luminance is lowered. Each threshold intensity is plotted versus time and the values are read from the graph at the end of a test session. Testing is continued until the final threshold is stabilized. The final dark-adapted threshold is defined as the average of three ascending and three descending thresholds and is obtained after 35 to 40 minutes in darkness.

When the logarithm of the light perception is plotted as a function of time in darkness, the change in threshold follows a characteristic course. There is an initial rapid fall in threshold attributed to cones, followed by a plateau. A steeper descent, referred to as the rod-cone break, usually occurs at 3 to 9 minutes followed by a slower descent attributed to adaptation of the rods. The final threshold attained at about 35 to 40 minutes is the most constant indicator of dark adaptation. Among stable subjects, test results are reproducible over a 1- to 6-month interval with final threshold differences ranging from 0 to 0.1 log candela/meter². In one series, the dark adapted final threshold among 50 normal subjects (aged 20 to 60 years) was –5.0 ± 0.3 candela/ meter² (Carney and Russell, 1980).

Similar information on retinal function may be obtained by an electroretinogram or an electrooculorgram. However, these tests are more invasive than dark adaptation and there are not as many data relating these functional tests to dietary vitamin A levels.

There is literature relating dark adaptation test results to dietary levels of vitamin A under controlled experimental conditions (Table 4-4). Under controlled feeding conditions, dark adaptation, objectively measured by dark adaptometry, is one of the most sensitive indicators of a change in vitamin A deficiency status (Figure 4-3). Epidemiological evidence suggests that host resistance to infection is impaired at lesser stages of vitamin A deficiency, prior to clinical onset of night blindness (Arroyave et al., 1979; Arthur et al., 1992; Barreto et al., 1994; Bloem et al., 1990; Ghana VAST Study Team, 1993; Loyd-Puryear et al., 1991; Salazar-Lindo et al., 1993). Moreover, laboratory animals fed a vitamin A-deficient diet maintain ocular levels of vitamin A despite a significant reduction in hepatic vitamin A levels (Bankson et al., 1989; Wallingford and Underwood, 1987). Nevertheless, this approach can be used to estimate the average requirement for vitamin A but without assurance of adequate tissue levels to meet nonvisual needs for vitamin A.

TABLE 4-4

Correction of Abnormal Dark Adaptation with Vitamin A.

FIGURE 4-3

Serum vitamin A concentrations and dark adaptation final thresholds. Upper limit of normal final threshold = –4.6 log candela/m². Adapted from Carney and Russell (1980).

Pupillary Response Test

Another test of ability to dark adapt, one that avoids reliance on psychophysical responses, is the pupillary response test that measures the threshold of light at which a pupillary reflex (contraction) first occurs under dark-adapted conditions (Stewart and Young, 1989). The retina of one eye is briefly exposed to incremental pulses of light while a trained observer monitors the consensual response of the other pupil under dark conditions. A high scotopic (vision in dim light) threshold indicates low retinal sensitivity, a pathophysiological response to vitamin A deficiency. An early report of pupillary nonresponse to candlelight among night blind Confederate soldiers in the Civil War (Hicks, 1867) led to the development and validation of instrumentation for this test as a reliable, functional measure of vitamin A deficiency in Indonesian (Congdon et al., 1995) and Indian (Sanchez et al., 1997) children. However, data do not currently exist relating pupillary threshold sensitivity as determined by this test to usual vitamin A intakes, and so measures of pupillary response cannot be used at the present to establish dietary vitamin A requirements.

Dietary Fat

Dietary vitamin A is digested in mixed micelles and absorbed with fat. In some studies, increasing the level of fat in a low fat diet has been shown to improve retinol and carotene absorption (Reddy and Srikantia, 1966) and vitamin A nutriture (Jalal et al., 1998; Roels et al., 1963). Other studies, however, have not demonstrated a beneficial effect of fat on vitamin A absorption (Borel et al., 1997; Figueira et al., 1969).

For optimal carotenoid absorption, a number of research groups have demonstrated that dietary fat must be consumed along with carotenoids. Roels and coworkers (1958) reported that the addition of 18 g/day of olive oil improved carotene absorption from 5 to 25 percent. Jayarajan and coworkers (1980) reported that the addition 5 g of fat to the diet significantly improved serum vitamin A concentrations among children after the consumption of a low fat vegetable diet. The addition of 10 g of fat did not improve serum vitamin A concentrations any more than did 5 g of fat.

Infections

Malabsorption of vitamin A can occur with diarrhea and intestinal infections and infestations. Sivakumar and Reddy (1972) demonstrated depressed absorption of labeled vitamin A in children with gastroenteritis and respiratory infections. Malabsorption of vitamin A is also associated with intestinal parasitism (Mahalanabis et al., 1979; Sivakumar and Reddy, 1975).

The malabsorption of vitamin A that is observed in children with Ascaris lumbricoides infection was associated with an altered mucosal morphology that was reversed with deworming (Jalal et al., 1998; Maxwell et al., 1968).

Food Matrix

The matrix of foods affects the ability of carotenoids to be released from food and therefore affects intestinal absorption. The rise in serum β-carotene concentration was significantly less when individuals consumed β-carotene from carrots than when they received a similar amount of β-carotene supplement (Micozzi et al., 1992; Tang et al., 2000; Torronen et al., 1996). This observation was similar for broccoli (Micozzi et al., 1992) and mixed green leafy vegetables (de Pee et al., 1995; Tang et al., 2000) as compared with a β-carotene supplement. The food matrix effect on β-carotene bioavailability has been reviewed (Boileau et al., 1999).

Food Processing

The processing of foods greatly affects the absorption of carotenoids (Van het Hof et al., 1998). The absorption of carotene was 24 percent from sliced carrots, whereas the absorption of carotene from homogenized carrots was 56 percent (Hume and Krebs, 1949). Rock et al. (1998) reported that the rise in serum β-carotene concentration was significantly greater in subjects consuming cooked carrots and spinach as compared with those consuming an equal amount of raw carrots and spinach. Similarly, the rise in serum β-carotene concentration was greater after the consumption of carrot juice than after the same amount of raw carrots (Torronen et al., 1996).

Iron

A direct correlation between hemoglobin and serum retinol concentrations has been observed (Suharno et al., 1993; Wolde-Gebriel et al., 1993). Anemic rats have been shown to have reduced plasma retinol concentrations when fed a vitamin A-rich diet (Amine et al., 1970), although normal hepatic stores of vitamin A were observed (Staab et al., 1984). Rosales and coworkers (1999) reported that iron deficiency in young rats alters the distribution of vitamin A concentration between plasma and liver. In a cross-sectional study of children in Thailand, serum retinol concentration was positively associated with serum iron and ferritin concentrations (Bloem et al., 1989). Intervention studies among Indonesian girls demonstrated that combining vitamin A with iron supplementation was more effective in increasing hemoglobin concentrations than was giving iron alone (Suharno et al., 1993). As discussed in further detail in Chapter 9, various studies suggest that vitamin A deficiency impairs iron mobilization from stores and therefore vitamin A supplementation improves hemoglobin concentrations (Lynch, 1997).

Zinc

Zinc is required for protein synthesis, including the hepatic synthesis and secretion of retinol binding protein (RBP) and transthyretin; therefore, zinc deficiency influences the mobilization of vitamin A from the liver and its transport into the circulation (Smith et al., 1974; Terhune and Sandstead, 1972). In animal models, circulating and hepatic concentrations of retinol decline and rise with experimental zinc deficiency and repletion, respectively (Baly et al., 1984; Duncan and Hurley, 1978). In humans, cross-sectional studies and supplementation trials have failed to establish a consistent relationship between zinc and vitamin A status (Christian and West, 1998). Because zinc is important in the biosynthesis of RBP, it has been suggested that zinc intake may positively affect vitamin A status only when individuals are moderately to severely protein-energy deficient (Shingwekar et al., 1979).

Although the alcohol dehydrogenase enzymes involved in the formation of retinal from retinol in the eye are not zinc dependent (Duester, 1996; Persson et al., 1995), zinc-deficient rats had a significant reduction in the synthesis of rhodopsin (Dorea and Olson, 1986), which was postulated to be due to impaired protein (opsin and alcohol dehydrogenase) synthesis. Morrison and coworkers (1978) reported that dark adaptation improved after the provision of 220 mg/day of zinc to zinc-deficient patients.

Carotenoids

Competitive interactions among different carotenoids have been observed. When subjects were given purified β-carotene and lutein in a combined dose, β-carotene significantly reduced lutein absorption, and therefore serum lutein concentration, compared to when lutein was given alone (Kostic et al., 1995). However, lutein given in combination with β-carotene significantly increased β-carotene serum concentrations compared to when β-carotene was given alone. Johnson et al. (1997) reported that lycopene does not affect the absorption of β-carotene, and β-carotene improved the absorption of lycopene.

Alcohol

Because both retinol and ethanol are alcohols, there is potential for overlap in the metabolic pathways of these two compounds. Competition with each other for similar enzymatic pathways has been reported (Leo and Lieber, 1999), while other retinol and alcohol dehydrogenases show greater substrate specificity (Napoli et al., 1995). Ethanol consumption results in a depletion of hepatic vitamin A concentrations in animals (Sato and Lieber, 1981) and in humans (Leo and Lieber, 1985). Although the effect on vitamin A is due, in part, to hepatic damage (Leo and Lieber, 1982) and malnutrition, the reduction in hepatic stores is also a direct effect of alcohol consumption. Patients with low vitamin A stores, in the study by Leo and Lieber (1982), were otherwise well nourished. Furthermore, the reduction in hepatic vitamin A stores was reduced before the onset of fibrosis or cirrhosis of the liver (Sato and Lieber, 1981). Results suggest that vitamin A is mobilized from the liver to other organs (Mobarhan et al., 1991) with ethanol consumption. Chronic ethanol intake resulted in increased destruction of retinoic acid through the induction of P450 enzymes, resulting in reduced hepatic retinoic acid concentrations (Wang, 1999).

Method Used to Set the Adequate Intake

No functional criteria of vitamin A status have been demonstrated that reflect response to dietary intake in infants. Thus, recommended intakes of vitamin A are based on an Adequate Intake (AI) that reflects a calculated mean vitamin A intake of infants principally fed human milk.

Ages 0 through 6 Months. Using the method described in Chapter 2, the AI of vitamin A for infants ages 0 though 6 months is based on the average amount of vitamin A in human milk that is consumed. After rounding, an AI of 400 μg retinol activity equivalents (RAE)/day is set based on the average volume of milk intake of 0.78 L/day (see Chapter 2) and an average concentration of vitamin A in human milk of 1.70 μmol/L (485 μg/L) during the first 6 months of lactation (Canfield et al., 1997, 1998) (see Table 4-5). Because the bioconversion of carotenoids in milk and in infants is not known, the contribution of carotenoids in human milk to meeting the vitamin A requirement of infants was not considered.

TABLE 4-5

Vitamin A in Human Milk.

Ages 7 through 12 Months. Using the method described in Chapter 2 to extrapolate from the AI for infants ages 0 through 6 months fed human milk, the intake from human milk for the older infants is 483 μg RAE/day of vitamin A.

The vitamin A intake for older infants can also be determined by estimating the intake from human milk (concentration × 0.6 L/ day) and complementary foods (Chapter 2). Vitamin A intake data (n = 45) from complementary foods was estimated to be 244 μg/day based on data from the Third National Health and Nutrition Examination Survey. The average intake from human milk is approximately 291 μg/day (485 μg/L × 0.6 L/day). Thus, the total vitamin A intake is estimated to be 535 μg RAE/day (244 μg/day + 291 μg/day).

On the basis of these two approaches and rounding, the AI was set at 500 μg RAE/day. The AI for infants is greater than the Recommended Dietary Allowance (RDA) for young children because the RDA is based on extrapolation of adult data (see “Children and Adolescents Ages 1 through 18 Years”).

Vitamin A AI Summary, Ages 0 through 12 Months

Special Considerations

Concentrations of 520 to 590 μg/L of vitamin A in milk from Holstein cows have been reported (Tomlinson et al., 1976), which is significantly less than the levels observed in human milk (Table 4-5). The majority of vitamin A and carotenes are located in the fat globule and fat globule membrane in cow milk (Patton et al., 1980; Zahar et al., 1995). The concentrations of retinol and β-carotene in cow milk averaged 18 to 27 μg/g of milk fat in one study (Jensen and Nielsen, 1996). Retinol in cow milk is bound to β-lactoglobulin, which has a structure very similar to retinol binding protein (Papiz et al., 1986). There is minimal isomerization of trans-retinol to cis-retinol in unheated cow milk (Panfili et al., 1998), the latter being less well absorbed. Cow milk submitted to pasteurization resulted in 3 to 6 percent isomerization to cis-retinol. Greater isomerization was observed with severe heat treatments (16 percent in ultra high temperature milk and 34 percent in sterilized milk).

Method Used to Estimate the Average Requirement

No data are available to estimate an average requirement for children and adolescents. A computational method is used that includes an allowance for adequate liver vitamin A stores to set the Estimated Average Requirement (EAR) (see “Adults Ages 19 Years and Older”). The EAR for children and adolescents is extrapolated from adults by using metabolic body weight and the method described in Chapter 2. If total body weight is used, the RDA for children 1 through 3 years would be 200 μg RAE/day. If metabolic weight (kg^0.75) is used, the RDA would be 300 μg RAE/day. Studies conducted in developing countries indicate that xerophthalmia and serum retinol concentrations of less than 20 μg/dL exist among preschool children with daily intakes of up to 200 μg of vitamin A, whereas 300 μg/day of vitamin A is associated with serum retinol concentrations greater than 30 μg/dL (Reddy, 1985). Although similar data are lacking in developed countries, to ensure that the RDA will meet the requirement of almost all North American preschool children, metabolic weight was used to extrapolate from adults.

Vitamin A EAR and RDA Summary, Ages 1 through 18 Years

The RDA for vitamin A is set by using a coefficient of variation (CV) of 20 percent based on the calculated half-life values for liver vitamin A (see “Adults Ages 19 Years and Older”). The RDA is defined as equal to the EAR plus twice the CV to cover the needs of 97 to 98 percent of individuals in the group (therefore, for vitamin A the RDA is 140 percent of the EAR). The calculated values for the RDAs have been rounded to the nearest 100 μg.

Evidence Considered in Estimating the Average Requirement

The calculation described below can be used for estimating the vitamin A requirement and is calculated on the basis of the amount of dietary vitamin A required to maintain a given body-pool size in well-nourished subjects. Olson (1987) determined the average requirement of vitamin A by this approach using the calculation:

A × B × C × D × E × F

A = Percent of body vitamin A stores lost per day when ingesting a vitamin A-free diet

B = Minimum acceptable liver vitamin A reserve

C = The liver weight:body weight ratio

D = Reference weight for a specific age group and gender

E = Ratio of total body:liver vitamin A reserves

F = Efficiency of storage of ingested vitamin A.

By using this approach, a daily vitamin A intake can be determined that will assure vitamin A reserves to cover increased needs during periods of stress and low vitamin A intake. That value can be used for estimating the average requirement for vitamin A.

The portion of body vitamin A stores lost per day has been estimated to be 0.5 percent based on the rate of excretion of radio-activity from radiolabeled vitamin A and by the calculation of the half-life of vitamin A. The minimal acceptable liver reserve is estimated to be 20 μg/g and is based on the concentration at which (1) no clinical signs of a deficiency are observed, (2) adequate plasma retinol concentrations are maintained (Loerch et al., 1979), (3) induced biliary excretion of vitamin A is observed (Hicks et al., 1984), and (4) there is a protection against a vitamin A deficiency for approximately 4 months while the person consumes a vitamin A-deficient diet. The liver weight:body weight ratio is 1:33 (0.03) and is an average of ratios for infants and adults. The reference weights for adult women and men are 61 and 76 kg, respectively (see Chapter 1). The ratio of total body:liver vitamin A reserves is 10:9 (1.1) and is based on individuals with adequate vitamin A status. Finally, the efficiency of storage can be determined by isotope dilution methods following the administration of either radioactive or stable-isotopically labeled vitamin A to subjects adequate in vitamin A (Bausch and Reitz, 1977; Haskell et al., 1997). Recent studies by Haskell and coworkers (1997) suggest that the efficiency of storage is approximately 40 percent, rather than the 50 percent that was previously reported (Olson, 1987). Based on these current estimations, the EAR of preformed vitamin A required to assure an adequate body reserve in an adult man is 0.005 × 20 μg/g × 0.03 × 76 kg × 1.1 × 2.5, or 627 μg RAE/day. With a reference weight of 61 kg for women, the EAR would be 503 μg RAE/day.

Based on the study of Sauberlich and coworkers (1974), Olson (1987) estimated that the liver vitamin A concentration was less than 10 μg/g at the time the first clinical signs of vitamin A deficiency appeared. From this assumption, it was estimated that the half-life of vitamin A is approximately 128 days, and the CV is 21 percent. Because the portion of this variability that is due to experimental error is not known, a CV of 20 percent is used for setting the RDA.

Vitamin A EAR and RDA Summary, Ages 19 Years and Older

The RDA for vitamin A is set by using a CV of 20 percent (see Chapter 1) using the EAR for adequate body stores of vitamin A. The RDA is defined as equal to the EAR plus twice the CV to cover the needs of 97 to 98 percent of the individuals in the group (therefore, for vitamin A the RDA is 140 percent of the EAR). The calculated values for the RDAs have been rounded to the nearest 100 μg.

Evidence Considered in Estimating the Average Requirement

Direct studies of the requirement for vitamin A during pregnancy are lacking. The model used to establish the EAR is based on the accumulation of vitamin A in the liver of the fetus during gestation and an assumption that liver contains approximately half of the body's vitamin A when liver stores are low, as in the case of newborns. Liver vitamin A concentrations for full-term stillborn infants (Dorea et al., 1984; Hoppner et al., 1968; Montreewasuwat and Olson, 1979; Olson, 1979) have ranged from less than 10 to greater than 100 μg/g liver, with values tending to be skewed towards the lower range (Olson, 1979). A vitamin A concentration of 1,800 μg per liver for 37 to 40 week gestation age (Montreewasuwat and Olson, 1979) was used to calculate a concentration of 3,600 μg per fetus. Assuming the efficiency of maternal vitamin A absorption to average 70 percent and vitamin A to be accumulated mostly in the last 90 days of pregnancy, the mother's requirement would be increased by approximately 50 μg/day during the last trimester. Because vitamin A in the mother's diet may be stored and mobilized later as needed and some vitamin A may be retained in the placenta, the EAR is estimated to be ~50 μg/day in addition to the EAR for nonpregnant adolescent girls and women for the entire pregnancy period.

Vitamin A EAR and RDA Summary, Pregnancy

The RDA for vitamin A is set by using a CV of 20 percent based on the calculated half-life values for liver vitamin A (see “Adults Ages 19 Years and Older”). The RDA is defined as equal to the EAR plus twice the CV to cover the needs of 97 to 98 percent of individuals in the group (therefore, for vitamin A the RDA is 140 percent of the EAR). The calculated values for the RDAs have been rounded up to the nearest 10 μg.

Evidence Considered in Estimating the Average Requirement

As indicated earlier in the section on infants, human milk-fed infants consume on average 400 μg/day of vitamin A in the first 6 months of life. The carotenoid content of human milk has been summarized in Dietary Reference Intakes for Vitamin C, Vitamin E, Selenium, and Carotenoids (IOM, 2000). Because the bioconversion of carotenoids in milk and in infants is not known, the contribution of carotenoids in human milk to meeting the vitamin A requirement of infants was not considered. To set an EAR during pregnancy, 400 μg RAE/day is added to the EAR for nonpregnant adolescent girls and women to assure adequate body stores of vitamin A.

Vitamin A EAR and RDA Summary, Lactation

The RDA for vitamin A is set by using a CV of 20 percent based on the calculated half-life values for liver vitamin A (see “Adults Ages 19 Years and Older”). The RDA is defined as equal to the EAR plus twice the CV to cover the needs of 97 to 98 percent of individuals in the group (therefore, for vitamin A the RDA is 140 percent of the EAR). The calculated values for the RDAs have been rounded to the nearest 100 μg.

Alcohol Consumption

Excessive alcohol consumption results in a depletion of liver vitamin A stores (Leo and Lieber, 1985). Depletion is partly due to the reduced consumption of foods. Furthermore, mobilization of vitamin A out of the liver may be increased with excessive alcohol consumption (Lieber and Leo, 1986). Because alcohol intake has been shown to enhance the toxicity of vitamin A (Leo and Lieber, 1999) (see “Tolerable Upper Intake Levels”), individuals who consume alcohol may be distinctly susceptible to the adverse effects of vitamin A and any increased intake to meet one's needs should be in the context of maintaining health.

Developing Countries and Vegetarian Diets

A number of factors can influence the requirement for vitamin A, including iron status, the presence and severity of infection and parasites, the level of dietary fat, protein energy malnutrition, and the available sources for preformed vitamin A and provitamin A carotenoids.

Parasites and Infection. Malabsorption of vitamin A can occur with diarrhea and intestinal infestations (Jalal et al., 1998; Sivakumar and Reddy, 1972). Furthermore, the urinary excretion of vitamin A is increased with infection, and especially with fever (Alvarez et al., 1995; Stephensen et al., 1994). For these reasons, with parasitic infestation and during infection, the requirement for vitamin A may be greater than the requirements set in this report, which are based on generally healthy individuals.

Protein Energy Malnutrition. Protein synthesis generally, and specifically retinol binding protein synthesis, is reduced with severe protein energy malnutrition (PEM) (marasmus and kwashiorkor), and therefore release of retinol from the liver (assuming stores are present) is also reduced (Large et al., 1980). With successful dietary treatment of PEM, growth and tissue weight gain will be stimulated, and the relative requirement of vitamin A will increase during the recovery period.

Vegetarianism. Preformed vitamin A is found only in animal-derived food products. A clinical sign of vitamin A deficiency, night blindness, is prevalent in developing countries where animal and vitamin A-fortified products are not commonly available. Although carotenoids such as β-carotene are abundant in green leafy vegetables and certain fruits, because it takes 12 μg of dietary β-carotene to provide 1 retinol activity equivalent (RAE) (as compared to previous recommendations where 1 μg of retinol was thought to be provided by 6 μg of β-carotene [NRC, 1989 and Table 4-3]), a greater amount of fruits and vegetables than previously recommended are required to meet the daily vitamin A requirement for vegetarians and those whose primary source of vitamin A is green leafy vegetables.

Analyzing intakes of vitamin A and β-carotene and using an RAE of 12 μg for dietary β-carotene indicate that the RDA for vitamin A can be met by those consuming a strict vegetarian diet containing the deeply colored fruits and vegetables (1,262 μg RAE) that are major sources of β-carotene in the United States (Chug-Ahuja et al., 1993) (Table 4-6). The United States has several vitamin A-fortified foods, including milk, cereals, and infant formula. Furthermore, certain food products, such as sugar, are being fortified with vitamin A in some developing countries. If menus are restricted in the amounts of provitamin A carotenoids consumed and such fortified products are not part of routine diets, then vitamin A supplements may be required.

TABLE 4-6

Vitamin A Intake from a Vegan Diet High in Carotene-Rich Fruits and Vegetables.

Populations Where Consumption of Vitamin A-Rich Foods is Limited. Three major intervention trials have been conducted in developing countries to evaluate the efficacy of provitamin A carotenoids in maintaining or improving vitamin A status in lactating women (de Pee et al., 1995), preschool children (Jalal et al., 1998), and young children (Takyi, 1999). Vitamin A status, as determined by serum retinol concentration, was not improved in Indonesian lactating women after the consumption of dark green leafy vegetables (de Pee et al., 1995). These women had hemoglobin concentrations less than 13 mg/dL. There is evidence that iron deficiency impairs the metabolism of vitamin A in laboratory animals (Jang et al., 2000; Rosales et al., 1999). In some, but not all, studies (Suharno et al., 1993), iron supplementation improved vitamin A status in humans (Munoz et al., 2000). Therefore the presence of iron deficiency, which is prevalent in developing countries, may impair the efficacy of dark green leafy vegetables. Jalal and coworkers (1998) reported that the addition of β-carotene-rich foods to the diets of preschool children improved vitamin A status, however, vitamin A status improved almost as well when fat was added to the diet and an anthelmintic drug to destroy parasitic worms was provided. This finding demonstrates the importance of dietary fat, which is often low in the diets of developing countries and the importance of intestinal parasites on vitamin A status. Takyi (1999) reported that the vitamin A status of young children improved similarly when fed either a pureed β-carotene-rich diet or provided a similar amount of β-carotene as a supplement. Here, in contrast to the findings of Jalal et al. (1998), dietary fat and anthelmentic drugs did not appear to have a beneficial effect on vitamin A status, possibly because the carotenoid was already provided in a highly absorbable, pureed form.

The EARs that have been set for the North American population are achievable through diet because of the abundance of vitamin A-rich foods. Populations of less developed countries may have difficulty in meeting the EAR that ensures adequate vitamin A stores. Therefore, an EAR that does not assure adequate vitamin A stores has been determined on the basis of the level of vitamin A for correction of abnormal dark adaptation in adults. This approach does not assure adequate stores of vitamin A because animal studies indicate that vitamin A depletion of the eye occurs after the depletion of hepatic vitamin A reserves (Bankson et al., 1989; Lewis et al., 1942). Furthermore, epidemiological studies in children suggest impaired host resistance to infection, presumably reflecting compromised immunity and represented by increased risk of morbidity and mortality at lesser stages of depletion (Arroyave et al., 1979; Arthur et al., 1992; Barreto et al., 1994; Bloem et al., 1990; Ghana VAST Study Team, 1993; Loyd-Puryear et al., 1991; Rahmathullah et al., 1990; Salazar-Lindo et al., 1993; West et al., 1991).

An EAR of 300 μg RAE/day can be calculated based on the dark adaptation data obtained from 13 individuals from four studies on adults (Table 4-4). The duration of depletion and repletion varied among these four studies and the majority of the studies were conducted on men. Interpolation of the level of vitamin A at which dark adaptation of each individual was corrected in these four studies results in a median intake of 300 μg RAE/day, which can be used to set an EAR based on dark adaptation for adults. Using this method, there was insufficient evidence to support setting a different EAR for men and for women, as there were too few women studied. EARs using dark adaptation as the indicator for children (1–3 years, 112 μg RAE/day; 4–8 years, 150 μg RAE/day; 9–13 years, 230 μg RAE/day) and adolescents (14–18 years, 300 μg RAE/day) are based on extrapolation from the adult EAR as described in Chapter 2.

Adverse Effects in Adults

Bone Mineral Density. Chronic, excessive vitamin A intake has been shown to lead to bone mineral loss in animals (Rohde et al., 1999), making such a consequence in humans biologically plausible. Most human case reports are not well described and epidemiological studies are inadequate in design. However, four studies provide interpretable evidence relating changes in bone mineral density (BMD) and risk of hip fracture with variation in dietary intake of preformed vitamin A (Freudenheim et al., 1986; Houtkooper et al., 1995; Melhus et al., 1998). The studies are distinguished by their well-described study designs and populations, adequate dietary intake estimates, and accurate methods for measuring BMD at multiple sites.

One two-part study (Melhus et al., 1998) suggests that a chronic intake of 1.5 mg/day of preformed vitamin A is associated with osteoporosis and increased risk of hip fracture. The first part, a cross-sectional multivariate regression analysis in 175 Swedish women 28 to 74 years of age, showed a consistent loss in BMD at four sites and in total BMD with increased preformed vitamin A intake. Numerous nutritional and non-nutritional exposures were concurrently assessed, allowing substantial control of potential confounders. With the use of stratified estimates of retinol intake in a univariate regression analysis, BMD was shown to increase with each 0.5 mg/day increment in intake above a reference intake of less than 0.5 mg/day, until intakes exceeded 1.5 mg/day. Above this level, mean BMD decreased markedly at each site. In a multivariate model, adjusting for effects of 14 other covariates, similar results were found. It is not clear whether the findings are equally applicable to pre- and postmenopausal women.

The second part was a nested case-control study on the risk factors for hip fracture. Cases were mostly postmenopausal women with first hip fracture within 2 to 64 months after entry into the large cohort study, or 5 to 67 months after the mid-point of the recalled dietary assessment. Four matched control subjects were selected for each case. A total of 247 cases and 873 control subjects completed the study. Univariate and multivariate conditional logistic regression analysis showed a dose-dependent increase in the risk of hip fracture with each 0.5 mg/day increment in reported retinol intake above 0.5 mg/day (baseline). The odds ratio was 2.05 (95 percent confidence interval, 1.05–3.98) at intakes above 1.5 mg/day.

In contrast to the results of Melhus and coworkers (1998), which suggest that risk of bone mineral loss and hip fracture occurs at estimated intakes above 1.5 mg/day, two U.S. studies provide no evidence of increased bone mineral loss in women with intakes of preformed vitamin A up to 1.5 to 2.0 mg/day (Freudenheim et al., 1986; Houtkooper et al., 1995). Freudenheim and coworkers (1986) evaluated the correlation between mean 3-year vitamin A intakes ranging from approximately 2 to 3 mg/day and rates of change in BMD in 84 women, 35 to 65 years of age (17 pre- and 67 postmenopausal). No consistent relationship was reported between vitamin A intake and the rate of bone mineral content loss in pre- and postmenopausal women. The single subject who showed rapid bone mineral loss with very high vitamin A intake also appeared to have consumed large amounts of other micronutrients as well, obscuring the significance of this relationship. Further, this study suffers from a small sample size in each of the four key groups (i.e., pre- and postmenopausal women by calcium supplement status), making correlations of potential nutritional or pathological importance indeterminate.

Houtkooper and coworkers (1995), in a longitudinal study of 66 women 28 to 39 years of age, showed that vitamin A intake was significantly associated with the increased annual rate of change in total body BMD. The mean rate of change in total body BMD over the 18-month study was negative, although several sites (lumbar spine, trochanter, and Ward's triangle) showed small positive slopes. The estimated mean intake of preformed vitamin A from the diet was 1,220 ± 472 (standard deviation [SD]) μg/day. The estimated vitamin A intake from provitamin A carotenoids was 595 ± 352 (SD) μg/day. In multivariable regression models that included covariables for body composition and treatment (exercise versus sedentary) status, the slopes for vitamin A and carotene (two separate models) were both positive [b = 0.007 and 0.008 mg/(cm²-year)] with r² values of ~0.30 for each model. While the positive association between vitamin A and carotene intake and change in BMD may not be causal, the data provide evidence that vitamin A does not adversely affect premenopausal bone health within this range of intake.

The findings from these studies are provocative but conflicting, and therefore, they are not useful for setting a UL for vitamin A. More research is needed to clarify whether chronic vitamin A intake, at levels that characterize upper-usual intake ranges for many American and European populations, may lead to loss in BMD and consequent increased risk of hip fracture in certain population groups, particularly among pre- and postmenopausal women.

Teratogenicity. Concern for the possible teratogenicity of high vitamin A intake in humans is based on the unequivocal demonstration of human teratogenicity of 13-cis-retinoic acid (Lammer et al., 1985) after supplementation with high doses of vitamin A (Eckhoff and Nau, 1990; Eckhoff et al., 1991). Numerous studies in experimental animals clearly establish the teratogenic potential of excessive intakes of vitamin A (Cohlan, 1953, 1954; Geelen, 1979; Hutchings and Gaston, 1974; Hutchings et al., 1973; Kalter and Warkany, 1961; Pinnock and Alderman, 1992).

Epidemiological data show the possibility of teratogenic effects with high intakes of preformed vitamin A (Table 4-8). The critical period for susceptibility appears to be the first trimester of pregnancy and the primary birth defects associated with excess vitamin A intake are those derived from cranial neural crest (CNC) cells such as craniofacial malformations and abnormalities of the central nervous system (except neural tube defects), thymus, and heart. Examination of the data suggests a likely dose-dependent association between vitamin A intake at excessive levels and the risk of birth defects. One case-control report showed a statistically nonsignificant association between a reported maternal intake of greater than 12,000 μg/day and malformations, but not below that level (Martinez-Frias and Salvador, 1990). Two other large case-control studies showed no relationship between risk of malformation and likely supplemental daily doses of 2,400 to 3,000 μg by mothers (Khoury et al., 1996; Shaw et al., 1996). An observational study by Rothman and coworkers (1995) involving 22,748 pregnant women found that those who ingested greater than 4,500 μg/day of preformed vitamin A from food and supplements were at greater risk of delivering infants with malformations of CNC cell origin (e.g., cleft lip or palate) than were women consuming less than 1,500 μg/day. But questions have been raised about the accuracy of intake estimates and birth defects diagnosed. It has been argued further that the limited number of excess cases used to identify a toxicity threshold of 4,500 μg/day of preformed vitamin A (or 3,000 μg from supplements) permits the study's findings to be consistent with a larger threshold than other studies would suggest (Brent et al., 1996; Mastroiacovo et al., 1999; Watkins et al., 1996; Werler et al., 1996). Thus, while few dispute a causal association between excessive periconceptual vitamin A intake and risk of malformation, the threshold at which risk increases remains a matter of debate. However, in the context of the totality of data on vitamin A and birth defects, the data of Rothman and coworkers (1995) provide supportive evidence of a causal association. Case reports of malformations exist to support an increased risk of birth defects above a maternal intake of 7,800 μg/day of vitamin A (Bauernfeind, 1980; Bernhardt and Dorsey, 1974). Human case reports support a temporal association between maternal exposure to elevated vitamin A intakes and birth defects (Bernhardt and Dorsey, 1974; Von Lennep et al., 1985).

Liver Abnormalities. There is a strong causal association shown by human and animal data between excess vitamin A intake and liver abnormalities because the liver is the main storage site and target organ for vitamin A toxicity. The wide spectrum of vitamin A-induced liver abnormalities ranges from reversibly elevated liver enzymes to widespread fibrosis, cirrhosis, and sometimes death. Table 4-9 shows consistency and specificity of the following effects in liver pathology: spontaneous green fluorescence of sinusoidal cells, perisinusoidal fibrosis, hyperplasia, and hypertrophy of Ito cells. Human data are potentially confounded by other factors related to liver damage such as alcohol intake, hepatitis A, B, and C, hepatotoxic medications, or preexisting liver disease. A thorough evaluation of the liver data is provided in the later section, “Dose-Response Assessment”.

Adverse Interactions. Alcohol intake has been shown to enhance the toxicity of vitamin A (Leo and Lieber, 1999). In particular, the hepatotoxicity of vitamin A may be potentiated by alcohol use. Therefore, alcohol drinkers may be distinctly susceptible to the adverse effects of vitamin A.

Adverse Effects in Infants and Children

There are numerous case reports of infants (Table 4-10), toddlers, and children who have demonstrated toxic effects due to ex cess vitamin A intakes for months to years. Of particular concern are intracranial (bulging fontanel) and skeletal abnormalities that can result in infants given vitamin A doses of 5,500 to 6,750 μg/day (Persson et al., 1965). The clinical presentation of vitamin A toxicity in infants and young children varies widely. The more commonly recognized signs and symptoms include skeletal abnormalities, bone tenderness and pain, increased intracranial pressure, desquamation, brittle nails, mouth fissures, alopecia, fever, headache, lethargy, irritability, weight loss, vomiting, and hepatomegaly (Bush and Dahms, 1984). Furthermore, tolerance to excess vitamin A intake also appears to vary (Carpenter et al., 1987). Carpenter and coworkers (1987) described two boys who developed hypervitaminosis A by age 2 years for one and by age 6 years for the other. Both were given chicken liver that supplied about 690 μg/day of vitamin A and various supplements that supplied another 135 to 750 μg/day. An older sister who had been treated similarly remained completely healthy.

TABLE 4-8

The Relationship Between Reproductive Risk and Excess Preformed Vitamin A in Humans.

Summary

Based on considerations of causality, quality, and completeness of the database, teratogenicity was selected as the critical adverse effect on which to base a UL for women of childbearing age. For all other adults, liver abnormalities were the critical adverse effect. Abnormal liver pathology, characteristic of vitamin A intoxication (or grossly elevated hepatic vitamin A levels), was selected rather than elevated liver enzymes because of the uncertainties regarding other possible causes such as concurrent use of hepatotoxic drugs, alcohol intake, and hepatitis B and C. Bone changes were not used because of the conflicting findings and the lack of other data confirming the findings of Melhus et al. (1998).

Women of Reproductive Age

Data Selection. Epidemiological studies evaluating the teratogenicity of vitamin A intake shortly before or during pregnancy (Table 4-8) were used to derive a UL for women of reproductive age. Because adequate human data were available, animal data were not used to derive a UL.

Identification of a No-Observed-Adverse-Effect Level (NOAEL). A NOAEL of approximately 4,500 μg/day of preformed vitamin A from food and supplements was based on a critical evaluation of the data in Table 4-8. There are numerous reports showing no adverse effects at doses below 3,000 μg/day of vitamin A from supplements (Czeizel and Rockenbauer, 1998; Dudas and Czeizel, 1992; Khoury et al., 1996; Rothman et al., 1995) or 4,500 μg/day of preformed vitamin A from food and supplements (Rothman et al., 1995). Rothman and coworkers (1995) showed a significantly increased risk of birth defects at the cranial neural crest sites among women who consumed greater than 4,500 μg of preformed vitamin A/day from food and supplements during the first trimester compared to those who took 1,500 μg/day or less. Most of the human data on teratogenicity of vitamin A involve doses equal to or greater than 7,800 μg/day. There are limited epidemiological data to clearly define a dose-response relationship in the dose range of 3,000 to 7,800 μg/day. Nevertheless, 4,500 μg/day represents a conservative value for a NOAEL in light of the evidence of no adverse effects at or below that level.

TABLE 4-9

Evidence of Liver Abnormalities After Excess Preformed Vitamin A Intakes (< 30,000 μg/day), Based on Increasing Dose.

Uncertainty Assessment. An uncertainty factor (UF) of 1.5 was selected on the basis of inter-individual variability in susceptibility. Because there are substantial data (Table 4-8) showing no adverse effects at doses up to 3,000 μg/day of vitamin A supplements, a higher UF was not justified.

Derivation of a UL. The NOAEL of 4,500 μg/day was divided by the UF of 1.5 to obtain a UL value of 3,000 μg/day for women of reproductive age.

TABLE 4-10

Cases of Subchronic and Chronic, Low-Dose Vitamin A Toxicity in Infants, Based on Increasing Dose.

The UL for adolescent girls was adjusted on the basis of relative body weight as described in Chapter 3 with the use of reference weights from Chapter 1 (Table 1-1).

Vitamin A UL Summary, Adolescent Girls and Women Ages 14 through 50 Years, Pregnancy, Lactation

All Other Adults Ages 19 Years and Older, Excluding Women of Childbearing Age

Data Selection. Data on liver abnormalities in humans were used to derive a UL. Because clear toxicity has been demonstrated in numerous studies at doses above 15,000 μg/day, only data involving doses less than 30,000 μg/day of vitamin A were included in Table 4-9. Data were thoroughly evaluated for other potential causes of liver abnormalities. The following criteria for selecting the data sets were used: (1) data must show grossly elevated liver vitamin A levels or hypertrophy of Ito cells, (2) no alcoholism, (3) no concomitant liver hepatitis, and (4) no hepatotoxic drug use. While hepatitis A and B status are known in most cases, testing for hepatitis C did not begin until the early 1990s and is unknown in most cases. Therefore, hepatitis C was not used as a criterion for exclusion.

Two case studies reported hypertrophy of Ito cells in a 63-year-old woman after vitamin A intake of 14,000 μg/day for 10 years (Minuk et al., 1988) and in a 36-year-old man who took about 15,000 μg/ day for 12 years (Zafrani et al., 1984). Neither of these reports appear to be confounded by hepatitis A or B viral infections or concomitant exposure to other hepatotoxic agents including alcohol. Reports of vitamin A-induced hepatotoxicity at doses less than 14,000 μg/day were found (Eaton, 1978; Hatoff et al., 1982; Kowalski et al., 1994; Oren and Ilan, 1992). However, as Table 4-9 shows, these studies fail to provide information on other predisposing or confounding factors such as alcohol intake, drugs and medications used, and history of viral hepatitis infection.

Uncertainty Assessment. A UF of 5.0 was selected to account for the severe, irreversible nature of the adverse effect, extrapolation from a lowest-observed-adverse-effect level (LOAEL) to a NOAEL, and interindividual variation in sensitivity.

Derivation of a UL. Hepatotoxicity was reported at vitamin A supplement doses of 14,000 μg/day. A LOAEL of 14,000 μg/day was divided by a UF of 5 to obtain a UL after rounding of 3,000 μg/day for adults other than women of reproductive age. This UL is the same as that set for women of reproductive age, given that the UL is defined as the highest level of daily nutrient intake likely to pose no risk of adverse health effects to almost all of the general population.

Vitamin A UL Summary, Ages 19 Years and Older, Excluding Women of Childbearing Age

Infants, Children, and Adolescent Boys

Data Selection. Case reports of hypervitaminosis A in infants were used to identify a LOAEL and derive a UL. Data were not available to identify a NOAEL.

Identification of a LOAEL. A LOAEL of 6,460 μg/day of vitamin A (which was rounded to 6,000 μg/day) was identified by averaging the lowest doses of four case reports (Persson et al., 1965). Four cases of hypervitaminosis A occurred after doses of 5,500 to 6,750 μg/day of vitamin A for 1 to 3 months (Table 4-10). The age of onset of symptoms ranged from 2.5 to 5.5 months and included anorexia, hyperirritability, occipital edema, pronounced craniotabes, bulging fontanels, increased intracranial pressure, and skin lesions and desquamation. The lowest dose associated with a bulging fontanel involved a 4-month-old girl given a daily dose of 24 drops of AD-vimin (about 5,500 μg of vitamin A) for 3 months. Her fontanels bulged 0.5 centimeters above the plane of the skull. The other three cases involved a dose of 6,750 μg/day of vitamin A for 1 to 2.5 months. Increased intracranial pressure and bulging fontanels were observed in these cases as well. Other effects observed at the higher dose included anorexia, hyperirritability, occipital edema, pronounced craniotabes, skin lesions, skin desquamation, epiphyseal line changes, and cortical hyperostosis on x-rays.

Uncertainty Assessment. A UF of 10 was selected to account for the uncertainty of extrapolating a LOAEL to a NOAEL for a nonsevere and reversible effect (i.e., bulging fontanel) and the interindividual variability in sensitivity.

Derivation of a UL. The LOAEL of 6,000 μg/day was divided by a UF of 10 to calculate a UL of 600 μg/day of preformed vitamin A for infants.

Children and Adolescent Boys. There are limited case report data of hypervitaminosis A (e.g., bulging anterior fontanels, increased intracranial pressure, hair loss, increased suture markings on the skull, and periosteal new bone formation) in children and adolescents after doses ranging from 7,000 μg/day in young children to 15,000 μg/day in older children and adolescents (Farris and Erdman, 1982; Siegel and Spackman, 1972; Smith and Goodman, 1976). Given the dearth of information and the need for conservativism, the UL values for children and adolescents are extrapolated from those established for adults. Thus, the adult UL of 3,000 μg/day of preformed vitamin A was adjusted for children and adolescents on the basis of relative body weight as described in Chapter 2 with use of reference weights from Chapter 1 (Table 1-1). Values have been rounded.

Vitamin A UL Summary, Infants, Children, and Adolescent Boys

Special Considerations

A review of the literature revealed that individuals with high alcohol intake, pre-existing liver disease, hyperlipidemia, or severe protein malnutrition may be distinctly susceptible to the adverse effects of excess preformed vitamin A intake (Ellis et al., 1986; Hathcock et al., 1990; Leo and Lieber, 1999). These individuals may not be protected by the UL for vitamin A for the general population.