Olestra (Olean, Procter & Gamble, Cincinnati, OH), a mixture of hexa-, hepta- and octaesters of sucrose formed from long-chain fatty acids from vegetable oils, has the organoleptic and thermal properties of regular fats (Kester 1993). However, olestra is not hydrolyzed by gastric and pancreatic lipases (Mattson and Volpenhein 1972) and is not absorbed (Miller et al. 1995). Because of these properties, olestra can serve as a zero-calorie replacement of dietary fat. Olestra is approved by the U.S. Food and Drug Administration as a replacement for fat used in the preparation of savory snacks such as potato and corn chips and crackers (Federal Register 1996).

The presence of a nonabsorbed lipophilic substance such as olestra in the gastrointestinal (GI)⁵ tract can potentially interfere with the absorption of fat-soluble substances. A plausible mechanism proposed to explain how olestra interferes with fat-soluble substances is that the olestra in the GI tract competes with the intestinal mixed micelles for the fat-soluble molecules, making them unavailable for absorption (Jandacek 1982). The degree to which this interference might occur depends on several factors. A primary one is the degree of lipophilicity of the molecule. The more lipophilic the molecule, the more it will partition into the olestra. Water-soluble substances will not partition into olestra; therefore olestra has no effect on their absorption, a fact substantiated in a number of studies presented elsewhere in this supplement (Cooper et al. 1997a and 1997b, Schlagheck et al. 1997a and 1997b). Another important factor influencing the degree to which olestra can affect the absorption of fat-soluble molecules is the time between the consumption of olestra and the substance. The two must be present in the GI tract simultaneously for the interaction to occur. Finally, the amount of olestra eaten also affects the interaction.

Evidence that olestra can affect the absorption of lipophilic molecules comes from a number of animal and human studies. Olestra reduced serum cholesterol concentration in humans (Crouse and Grundy 1979, Fallat et al. 1976, Glueck et al. 1979) and increased cholesterol excretion in rats (Mattson et al. 1976) and in humans (Jandacek et al. 1980 and 1990). Reductions in plasma tocopherol concentration (Glueck et al. 1982, Mellies et al. 1983 and 1985) and serum 25-hydroxyergocalciferol concentration (Jones et al. 1991b) have been observed in humans consuming olestra. In a study in which free-living subjects ate 20 g/d olestra for 6 wk, olestra did not affect vitamin K function, as measured by prothrombin time (PT), partial thromboplastin time (PTT) or functional prothrombin concentration [Simplastin:Ecarin (S:E) ratio], although it had an effect on serum phylloquinone concentration (Jones et al. 1991a).

Results from short-term feeding studies and long-term safety studies in animals indicate that the effects of olestra on the absorption of fat-soluble vitamins can be offset by adding extra amounts of the vitamins to the diet. Mattson et al. (1979) measured the liver vitamin A content of rats fed 0, 677 or 1331 RE (µg retinol equivalents) of vitamin A total after feeding the rats a vitamin A-free diet containing 15% (wt/wt) olestra for 3 d. The liver contents of vitamin A were <50, 180 and 343 RE, respectively, indicating that addition of vitamin A offset the effect of olestra, in a roughly linear manner. In another study, mice were fed 0, 2.5, 5 or 10% olestra in diets supplemented with 2500 IU (1375 RE)/kg of vitamin A, 750 IU (18.8 µg)/kg of vitamin D, and from 160 to 640 IU [107 to 429 mg -tocopherol equivalents (-TE)]/kg of vitamin E for 2 y (Lafranconi et al. 1994). The liver concentrations of vitamins A and E for mice in the olestra-fed groups, measured periodically throughout the study, were comparable to control values. Serum 25-hydroxyvitamin D [25(OH)D] concentrations for the groups fed 2.5 or 5% olestra were comparable to control values; the concentration for the group fed 10% olestra was ~77% of control.

The purposes of this study were as follows: 1 ) to determine the effects of olestra on serum concentrations of cholesterol, -tocopherol, 25-hydroxyvitamin D [25(OH)D], carotenoids and retinol; and 2 ) to determine whether addition of an extra amount of vitamin E (1.1 mg vitamin E/g olestra) would offset the effect of olestra on serum -tocopherol concentration in a free-living population consuming olestra in foods.

SUBJECTS AND METHODS

The study consisted of two treatment groups and a placebo control group. The treatment groups were given either 18 g/d olestra or 18 g/d olestra supplemented with 1.1 mg d--tocopheryl acetate (-TA)/g olestra. The olestra was delivered in cookies and a frozen dessert. The placebo groups were given the same food items prepared with triglyceride. The olestra-containing test foods were prepared by substituting olestra for triglyceride in the recipes and were identical in appearance and taste to the placebo foods. The frozen dessert contained 9 g olestra or placebo per serving; each cookie contained 3 g olestra or placebo. The subjects were instructed to consume one serving of the dessert and three cookies per day to yield the 18 g/d dose of olestra. The placebo groups was instructed to consume the same number of placebo food items. The items were to be consumed with meals at the subject's discretion throughout the day. The subjects were not specifically requested to divide the daily allocation among the meals, and there was no restriction on what other foods they could eat between meals. All other food items were self-selected and freely consumed.

The daily olestra intake, 18 g/d, was exaggerated relative to expected intake from savory snacks to invoke any measurable effects on nutrient absorption, if they exist. The estimated chronic intake of olestra from savory snacks by the average snack consumer (all ages, both sexes) is 3.1 g/d; for the 90th-percentile consumer, it is 6.9 g/d (Webb et al. 1997). The olestra, prepared as described by Rizzi and Taylor (1978), consisted of >99% octa- and heptaesters. The relative composition of the fatty acids making up the ester groups was 11% palmitic, 49% stearic, 31% oleic, 7% linoleic and 2% others.

The amount of vitamin E (Covitol 1360, Henkel, Cincinnati, OH) added to the olestra, 1.1 mg d--tocopheryl acetate/g olestra, was estimated from preliminary animal studies (data not shown). It was added to the olestra before the foods were prepared.

Two hundred and nineteen subjects (67 males and 152 females, 18-65 y of age) were admitted to the study. The subjects were in good health as determined by medical history, physical examination and clinical laboratory data. To be admitted, the subjects had to have a serum -tocopherol concentration > 11.6 µmol/L, a fasting serum triglyceride value < 3.1 mmol/L and a fasting serum cholesterol value of <7.1 mmol/L. Subjects who were pregnant or nursing, using vitamin supplements or regularly using oral contraceptives were excluded. The subjects were assigned randomly within strata of age, sex, body mass index (BMI) and smoking status. Within each stratum, treatment groups were balanced so that there were no significant differences in serum -tocopherol concentrations, normalized with respect to serum lipids, among the groups.

Urinalysis, clinical chemistry and hematology profiles were obtained at the beginning, midpoint (wk 8) and end of the study by standard accredited methods (The Medical Laboratory, Indianapolis, IN). Fasting blood samples were collected biweekly by venipuncture for analyses of tocopherol, carotenoid, 25(OH)D and lipid (cholesterol and triglyceride) concentrations. Prothrombin time, PTT and functional prothrombin concentration were measured at base line, wk 8 and wk 16. Serum was prepared from fresh blood. The samples were immediately frozen and stored at 20°C until analyzed.

Body weights were monitored weekly to ensure that the subjects maintained their weight within 2.25 kg of their base-line value. Any subject whose body weight changed beyond the ± 2.25 kg limit was dropped from the study. An assessment of dietary intake was made at wk 2 and wk 14 using a self-administered food-frequency questionnaire, which assesses food intake over the past year (Willett et al. 1985).

Subject compliance with respect to olestra consumption was monitored by means of diaries in which the subjects recorded the number of food items consumed and the time the items were consumed each day. These diaries were reviewed on a biweekly basis by study personnel to determine the daily amount of olestra test food item eaten. To be included in the database, a subject had to consume the test foods for 101 of the 112 study days, 90% of the total dose of olestra.

To determine the serum concentration of total 25-hydroxyvitamin D [25(OH)D], the sum of 25-hydroxycholecalciferol and 25-hydroxyergocalciferol, the hydroxylated metabolites were extracted from serum with acetonitrile, and the lipids were removed with a Sep-Pak C-18 column (Waters) using acetonitrile as the solvent. The metabolites were further purified by eluting the samples from a Sep-Pak silica column (Waters) with 96:4 hexane/isopropyl alcohol. The purified samples were assayed for 25(OH)D by a radio-binding assay (Nichols Institute, San Juan Capistrano, CA) using rat intestinal extract as the binder and ³H-25(OH)D as the ligand.

Vitamin K function was measured by the Simplastin(S)-Ecarin(E) assay (Suttie et al. 1988). This assay provides an indirect measure of circulating, biologically active, fully -carboxylated prothrombin by comparing the amount of fully -carboxylated prothrombin, activated by a commercial thromboplastin (Simplastin, Warner-Lambert, Morris Springs, NJ), with the amount of des--carboxy and partially carboxylated prothrombin, activated by the snake venom protease Echis carinatus (Ecarin, Sigma). A normal plasma reference sample with an S:E value of 0.99 ± 0.08 (N = 48) was assayed along with the test samples to monitor the specificity and potency of the reagents.

Serum lipids (total cholesterol and triglycerides), PT and PTT were measured by an accredited clinical laboratory (The Medical Laboratory, Indianapolis, IN) using standard clinical methods.

All statistical tests were performed at the two-sided 0.05 significance level with SAS software (Version 5, SAS Institute, Cary, NC). Treatment effects were assessed by one-way repeated-measures ANOVA using only those subjects with complete data sets. Analysis was conducted on absolute values, differences from base line and percentage of base-line values. Secondary analyses consisted of one-way ANOVA and Kruskal-Wallis tests performed on absolute values and paired t tests performed on differences from base-line values at each time point. Multiple comparisons were performed using the protected least significant difference test (Carmer and Swanson 1973, Welsch 1977). Unless specifically indicated otherwise, data are expressed as means ± SEM.

RESULTS

Table 1. Base-line demographics for subjects consuming 18 g/d olestra or placebo for 16 wk [View Table]

On the basis of the food-frequency questionnaire, there were no significant differences among the groups in daily intakes of vitamin E, vitamin D or carotenoids at wk 2 or 14 of the study with one exception. Vitamin E intake by the O +TA group at wk 14 was significantly less than that of either the O (24% less) or P (27% less) group (Table 2). At wk 2, one subject in the O + TA group had a vitamin E intake of 427 -TE/d and one subject in the P group had an intake of 234 -TE/d. The data for these two subjects were excluded from the data set. If these subjects are included, the mean ± SEM vitamin E intake for the O + TA group for wk 2 would be 13.7 ± 6.6 -TE/d and the mean ± SEM intake for the P group for wk 2 would be 10.8 ± 3.4 -TE/d. These subjects were probably taking a vitamin E supplement. If the data for these two subjects are included, there are still no significant differences among the groups at wk 2. 

Table 2. Daily intake of vitamin E, vitamin D, carotenoids and vitamin A for subjects consuming 18 g/d olestra or placebo for 16 wk1 [View Table]

Group mean body weights did not change for those subjects completing the study, and no significant differences were found among the groups. Base-line and wk-16 group mean (± SD) body weights were 77 ± 2 and 77 ± 2 kg for the O + TA group, 78 ± 2 and 78 ± 2 kg, for the O group, and 76 ± 2 and 77 ± 2 kg for the P group.

Five subjects withdrew voluntarily because of illnesses unrelated to treatment. Seven subjects, four in the O + TA group, two in the O group, and one in the P group, withdrew voluntarily because of mild or moderate GI symptoms such as bloating, cramping, nausea and loose stools or diarrhea. Thirteen subjects were dropped or excluded from the data set because they did not consume 90% of the total dose of olestra, did not provide all blood samples or had unacceptable weight changes.

All laboratory values remained within normal laboratory ranges during the study. No changes indicative of an olestra-related effect were observed in any of the measures (data not shown). Sporadic intergroup significant differences occurred but were not consistent over time or with olestra ingestion.

Serum -tocopherol concentration remained essentially constant for the P group and decreased slightly for the O and the O + TA groups (Table 3). The decline occurred within 2 wk, after which the values remained essentially constant. Using the average of the values measured at wk 2 through 16 as the final steady-state value, the decrease was 7.3% in the O group and 5.4% in the O + TA group. Although the repeated-measures ANOVA showed no significant olestra effect, the test for parallelism indicated a significant difference in trend among the groups. ANOVA showed that serum -tocopherol values in the O group were significantly less than those in the P group at all time points except base line and wk 6 and 8. Values in the O + TA group were not significantly different than those in the P group at any time. The only significant differences between the O and the O + TA groups occurred at wk 10 and 14, when the values in the O + TA group were significantly greater than the values in the O group (Table 3). There was no significant olestra effect on serum -tocopherol concentration (data not shown).

Table 3. Serum -tocopherol concentration for subjects consuming 18 g/d olestra or placebo for 16 wk1 [View Table]

Because -tocopherol is transported as a constituent of lipoproteins (Behrens et al. 1982, Traber et al. 1988), serum -tocopherol concentration, normalized with respect to serum lipids, provides a more reliable measure of vitamin E status than the unnormalized concentration. This is the case especially when serum lipids change during a study, as they did here. Figure 1 shows -tocopherol concentrations, normalized with respect to serum cholesterol plus serum triglycerides and expressed as a percentage of base line for the three study groups. The value for the placebo group was essentially unchanged; only the wk-2 value was significantly different than base line. The value for the O group, averaged over wk 2 through 16, was decreased by 6% compared with the P group. This compared with a 4% decrease for the O + TA group. Repeated-measures ANOVA indicated a significant olestra effect on the normalized serum -tocopherol concentration. The mean value for the O group was significantly less than the value for the P group at all time points except wk 2, 8 and 16. The mean value for the O + TA group was not significantly different than that for either the P group or the O group at any time point.

Olestra, with or without added tocopheryl acetate, had no significant effect on the S:E ratio, PT or PTT, all measures of vitamin K status (Table 4). The S:E ratio indicated that vitamin K status was sufficient at the start of the study and did not change during the study for any of the groups.

Table 4. Simplastin-Ecarin ratio, prothrombin time (PT) and partial thromboplastin time (PTT) for subjects consuming 18 g/d olestra or placebo for 16 wk1,2 [View Table]

Serum 25(OH)D decreased slightly during the first 8 wk of the study and then increased in all groups (Table 5). There was no significant effect of olestra.

Table 5. Serum 25-hydroxyvitamin D concentration for subjects consuming 18 g/d olestra or placebo for 16 wk1 [View Table]

Repeated-measures ANOVA indicated a significant olestra effect on serum -carotene concentration. Serum -carotene concentration for the P group decreased slightly during the study; the values at wk 4, 6, 8, 12 and 14 were significantly less than the base-line value (Fig. 2). The values for the O and O + TA groups decreased within 2 wk and reached new steady-state values by wk 6. All biweekly values for both olestra groups were significantly less than base line. On the basis of the averages of the wk 6 through wk 16 values, the mean -carotene concentration decreased by 34% for the O group and by 36% for the O + TA group. ANOVA revealed that the values for the O and the O + TA group were significantly less than the value for the P group at all time points except base line. No significant differences between the values for the O and the O + TA groups occurred at any time. When adjustments were made for an 8% decrease in the serum carotenoid concentration for the P group, the net effect of olestra on serum -carotene concentration was a 26-28 % reduction.

The serum concentrations of -carotene, lycopene, lutein/zeaxanthin and total carotenoids responded similarly to that of -carotene (Table 6). All were reduced by 21-29% for the olestra groups when adjusted for small changes in the value for the P group. Normalization of the serum carotenoid concentrations with respect to total serum lipids did not affect the general results (data not shown). Serum retinol concentration was unaffected by olestra (data not shown).

Table 6. Serum -carotene, lycopene, lutein/zeaxanthin and total carotenoid concentrations for subjects consuming 18 g/d olestra or placebo for 16 wk1 [View Table]

Serum total cholesterol concentration remained constant for the P group and, on the basis of the average of the values at wk 12, 14 and 16, decreased by about 4.5%, relative to base line, for both olestra groups (Table 7). Scattered significant differences from base line occurred for both olestra groups, primarily during the last 4 wk of the study. Only one significant difference was found between the olestra groups and placebo; the value for the O group was significantly lower than the placebo value at wk 12. No significant differences were found between the two olestra groups. No consistent or significant olestra-related effects on serum triglyceride concentration were found although scattered significant differences occurred among the groups (Table 7).

Table 7. Serum total cholesterol and triglyceride concentrations for subjects consuming 18 g/d olestra or placebo for 16 wk1 [View Table]

DISCUSSION

The lipophilicity of a molecule can be determined from the degree to which it partitions between an oil and water; the octanol-water partition coefficient is commonly used as a measure of the lipophilicity of a molecule. This coefficient can be either measured experimentally or calculated from knowledge of molecular structure (Hansch and Leo 1979, Meylan and Howard 1995). Octanol-water partition coefficients for -carotene, -tocopherol and cholesterol are 17.6, 12.2 and 8.7, respectively (Cooper et al. 1997c). Because octanol-water partition coefficients are expressed in log units, -carotene is almost 100 times more lipophilic than cholesterol.

The circulating concentration of -tocopherol concentration is a reliable indicator of vitamin E status (Bieri 1990, Machlin 1991). Specifically, the ratio of serum -tocopherol concentration to the sum of serum cholesterol and serum triglyceride concentrations has been shown to be a sensitive measure of vitamin E status (Thurnham et al. 1986). In this study, serum -tocopherol concentration in both olestra groups fell to a constant percentage of control in about 2 wk. A decrease in serum -tocopherol concentration within 2 wk was found in an 8-wk study in which subjects ate 8, 20 or 32 g/d olestra as part of a controlled diet (Schlagheck et al. 1997b). These rapid changes in serum -tocopherol with olestra intake are consistent with changes that occur when intake of the vitamin is changed (Baker et al. 1986, Behrens and Madere 1990, Horwitt et al. 1972).

Although olestra reduced serum -tocopherol concentration, vitamin E status among the subjects remained adequate. The new steady-state concentrations in the placebo group and the olestra groups, 17-20 µmol/L, were within the 11-37 µmol/L range indicative of sufficient vitamin E status (Farrell 1980). The vitamin E intake of the three test groups was within the normal range reported for the general U.S. population (NRC 1989), i.e., about one Recommended Dietary Allowance (RDA), which is 8 and 10 mg -tocopherol equivalents for adult females and males, respectively.

Even though the serum -tocopherol concentration responded rapidly to olestra and reached a new steady-state concentration within about 2 wk, this response reflected changes in nutritionally important vitamin E stores. In young and adult animals, vitamin E concentrations in plasma and lean tissues were found to decline rapidly when the animals were placed on vitamin E-free diets; most of the change occurred within 2 wk (Bieri 1972, Machlin et al. 1979). In contrast, the mass of vitamin E in adipose tissue changed little, although the vitamin E adipose concentration declined as the amount of adipose tissue expanded with growth. These findings indicate that adipose stores of the vitamin, although large, are not mobilized in response to decreased intake and act to maintain serum concentrations of vitamin E. Similar effects occur in humans. For example, Schaefer et al. (1983) found that the amount of vitamin E per adipocyte remained constant in humans during periods of weight loss (i.e., declining adipose mass) despite a large depletion of triglyceride. These observations support the conclusion that adipose stores of vitamin E do not sustain the vitamin E nutritional needs of the body and do not mobilize to offset any decline in serum vitamin E concentration resulting from decreased availability of the vitamin.

The effect of olestra on serum -tocopherol measured in the present study was a reduction of ~6%, less than the effect measured in a study in which subjects consumed olestra for 8 wk as part of a controlled diet (Schlagheck et al. 1997b). In that study, the subjects ate olestra foods at each of the three daily meals and were not permitted to eat between meals. Under those dietary conditions, 20 g/d olestra reduced serum -tocopherol concentration by ~17%, almost three times the effect produced by 18 g/d in the present study. These results indicate that eating olestra every time vitamin E, or any other nutrient, is eaten exaggerates the olestra effect on absorption relative to dietary patterns in which olestra and vitamins may be consumed at different times. This is consistent with the partitioning mechanism, which requires that olestra and the vitamin be present in the GI tract at the same time for olestra to affect the absorption of the vitamin (Jandacek 1982).

Addition of 1.1 mg d--TA/g olestra partially restored the normalized serum -tocopherol concentration to control concentration, an indication that the effect of olestra on vitamin E status can be offset by adding adequate vitamin E to olestra, an expected result on the basis of the partitioning mechanism. However, an amount of vitamin E greater than that used in this study is required to completely offset the olestra effect and restore serum -tocopherol concentration to the control concentration. Studies in humans in which the vitamin E and other nutrient intakes were controlled showed that 2.1 mg d--TA/g olestra completely restored serum -tocopherol to control concentration (Schlagheck et al. 1997a).

Measurement of serum 25(OH)D is an accepted indicator of overall vitamin D status (Fraser 1984). Because hepatic 25-hydroxylase is not tightly homeostatically regulated, circulating concentrations of the 25-hydroxy metabolites of vitamin D₂ and vitamin D₃ are indices of the supply of vitamin D from dietary and endogenous sources (Holick et al. 1991). Serum 25(OH)D concentrations at the start of the study ranged from 20.1 to 24.1 nmol/L, typical of concentrations measured in free-living populations in the midwest and the northern U.S. in winter (Brazerol et al. 1988, Haddad and Hahn 1973, Jones 1978, Jones et al. 1991b). The lack of an effect on serum 25(OH)D concentration indicates that olestra did not measurably affect overall vitamin D status. The decline in serum 25(OH)D observed in all groups at wk 8, relative to base-line, and the subsequent increase at wk 16 were seasonal effects on the 25-hydroxyvitamin D₃ contribution to 25(OH)D. The study started in February; therefore, the contribution to 25(OH)D from sunlight-induced synthesis of 25-hydroxyvitamin D₃ was declining through wk 8 (April) before the subsequent increase with increased sun exposure during the last few weeks of the study.

In a previous 6-wk study in free-living subjects, 20 g/d olestra reduced serum 25(OH)D by ~5% (Jones et al. 1991b). Two factors help explain why the effect on serum 25(OH)D measured in this study was smaller than that. The first is a difference in the relative contribution of dietary vitamin D₂, the only portion expected to be affected by olestra, to serum 25(OH)D. That contribution probably was considerably lower in the present study than in the 6-wk study. Daily ergocalciferol intake was not measured in the 6-wk study, but 20 µg/d was given as a supplement, resulting in a contribution of dietary vitamin D₂ to serum 25(OH)D of ~50%. The dietary intake of dietary vitamin D in this study was 5-6 µg/d (200-250 IU/d). The dietary contribution to serum 25(OH)D, although not measured, was probably 10-15%.

The second factor that helps explain the smaller effect of olestra on serum 25(OH)D in the present study relative to the 6-wk study is the difference in the way olestra and vitamin D were eaten in the two studies. In the 6-wk study, the 20 µg/d ergocalciferol supplement was always eaten simultaneously with olestra, thus providing maximum opportunity for olestra to affect absorption. In the present study, the subjects had the opportunity to eat vitamin D-containing foods at times other than when they ate the olestra foods.

Olestra did not affect vitamin K function as evidenced by the lack of any significant effect on the S:E ratio, PT or PTT. The ratio of the amount of des--carboxylated and partially -carboxylated prothrombin to that of fully -carboxylated prothrombin, the S:E ratio, is a direct and sensitive measure of the adequacy of vitamin K for clotting factor synthesis (Suttie et al. 1988). Full vitamin K sufficiency results in a S:E value of ~1, depending on the specific batches of reagents used in the assay. When the intake of vitamin K by a vitamin K-replete subject falls below the amount needed for complete -carboxylation of the vitamin K-dependent proteins over a period of several days, the S:E ratio declines. The S:E ratio is more reliable than serum phylloquinone concentration as a measure of vitamin K functional status because the circulating concentration of phylloquinone fluctuates rapidly as a result of its short half-life in the plasma (~2 h) and because of the lack of tissue stores of the vitamin (Schlagheck 1997b, Shearer et al. 1974).

The absence of change in measures of vitamin K function agrees with the results of a previous 6-wk free-living study in which the same measures were unaffected by 20 g/d olestra (Jones et al. 1991a). It is also consistent with the results of studies in which olestra doses as high as 32 g/d had no effect on the urinary excretion of -carboxyglutamic acid or the plasma concentration of des--carboxyprothrombin (Schlagheck et al. 1997a and 1997b), both sensitive indicators of vitamin K function (Suttie 1992).

The half-lives of the vitamin K-dependent proteins and clotting factors, 60 h, are such that any effect of olestra on vitamin K function would have been manifested during this study (Ferland et al. 1993). For example, the S:E ratio has been shown to decline within 2 wk in response to a reduction in vitamin K₁ intake (Allison et al. 1987, Suttie et al. 1988). Other measures of vitamin K function, such as urinary excretion of -carboxyglutamic acid, also have been shown to change within 2-3 wk of restricted vitamin K intake (Allison et al. 1987, Ferland et al. 1993).

Comparison of the carotenoid intake of the present study population with values reported by others indicates that the intake was similar to that of the general population (Forman et al. 1993, Stryker et al. 1988). In addition, the base-line serum -carotene concentrations, 0.27-0.28 µmol/L, were in agreement with values reported by others for normal, healthy adults (Forman et al. 1993, Henderson et al. 1989, Weststrate and van het Hof 1995).

Because serum carotenoid concentration is directly proportional to carotenoid intake, changes in serum carotenoid concentration are reliable indicators of changes in absorption (Forman et al. 1993, Henderson et al. 1989, Ribaya-Mercado et al. 1989, Stryker et al. 1988). Nutritionally active carotenoids provide at least 25% of the vitamin A in the U.S. diet; the remainder comes from preformed vitamin A (Olson 1987). Therefore, a decrease in the absorption of either dietary source can result in a decrease in total vitamin A stores. However, the 27% reduction in serum carotenoid concentration measured in this study does not translate into a 27% reduction in body vitamin A stores because olestra has no significant effect on the absorption of retinyl palmitate, the major dietary source of vitamin A stores (Daher et al. 1997b).

Because liver vitamin A stores maintain serum vitamin A concentration when dietary intake is reduced until liver stores become depleted (Olson 1984), the lack of change in serum retinol concentration observed in this study is not surprising.

The 27% reduction in serum carotenoid concentration seen in this study was less than the 61% reduction produced by 20 g/d olestra in studies in which subjects ate olestra at every meal for 8 wk and were not permitted to eat between meals (Schlagheck et al. 1997a and 1997b). This is explained primarily by the difference between the studies in the frequency of co-consumption of olestra and carotenoid-containing foods. In the 8-wk studies, olestra was eaten at all meals and the subjects were not permitted to eat between meals, which meant that olestra and carotenoids were always eaten together. This pattern provides the maximum opportunity for olestra to affect the absorption of the carotenoids. In the present study, the subjects were requested to eat olestra at meals but were allowed to eat between meals, which meant that substantial amounts of carotenoids were probably eaten at times when olestra was not eaten. Separating the time between olestra intake and nutrient intake removes or greatly decreases the effect of olestra on nutrient absorption (Daher et al. 1997a).

The effect of olestra on carotenoid absorption measured in this study or in the 8-wk study is greater than the effect that will occur when olestra snack foods are eaten under real-life, free-living dietary patterns because of the exaggerated frequency of co-consumption as well as the daily intake of olestra. Menus census data show that snacks and carotenoid-containing foods are eaten together only about four times in 14 d by the average snack consumer and about six times by the 90th-percentile consumer (Cooper et al. 1997c). In the 8-wk study in which 20 g/d olestra reduced serum -carotene by 61%, olestra and carotenoid-containing foods were eaten together 42 times in 14 d. In the present study, olestra and carotenoid-containing foods could have been eaten together as many as 42 times in 14 d. Further, 18 g/d olestra represents an exaggerated intake of olestra from savory snacks, almost six times the estimated average chronic intake, 3.1 g/d, for the total population of snack eaters (Webb et al. 1997). From menu census data on the amount of -carotene eaten by the general population of snack eaters, the frequency at which snacks and carotenoid-containing foods are eaten together and the effect of olestra on carotenoid absorption when the two are eaten together (Schlagheck et al 1997a and 1997b), it has been calculated that eating olestra snacks will lead to a reduction of <10% in carotenoid intake for the general population (Cooper et al. 1997c).

The effect of olestra on total vitamin A stores might be slightly higher in subgroups of the population such as lacto-ovovegetarians and vegans than in the general population because these subgroups receive a higher proportion of their vitamin A stores from carotenoids (U.S. Department of Commerce 1988). However, these individuals also have a higher overall intake of vitamin A than the general population (Calkins et al. 1984, Marsh et al. 1988, Tylavsky and Anderson 1988, U.S. Department of Commerce 1988); therefore, any reduction in vitamin A stores they might have from eating olestra snacks is unlikely to be nutritionally significant. In any case, the effect of olestra on vitamin A stores can be offset by the addition of vitamin A to olestra (Cooper et al. 1997a and 1997b), and marketed olestra snacks will contain additional amounts of vitamin A to offset these effects.

The observation that olestra reduces serum cholesterol concentration is in agreement with results reported by others (Crouse and Grundy 1979, Fallat et al. 1976). However, the magnitude of the effect measured here, a 4.5% reduction, is less than the 7-14% effects reported because of the lower olestra dose and because the dietary habits of the subjects were not as controlled in this study. The subjects in the present study could, and probably did, consume considerable amounts of cholesterol at times when they did not eat the olestra foods.

The results of this study demonstrated that it is feasible to offset the effects of olestra on vitamin E status in a free-living population by adding the vitamin to olestra. This result applies to fat-soluble vitamins in general because the mechanism by which olestra affects vitamin E status is general for fat-soluble nutrients. Complete restoration of tissue concentrations of vitamin A, vitamin E and vitamin D has been demonstrated in pigs fed olestra mixed in the diet (Cooper et al. 1997a and 1997b) and, for vitamins E and D, in humans eating olestra in foods such as potato chips, biscuits and cookies (Schlagheck et al. 1997a). Extra amounts of all of these vitamins, as well as vitamin K, will be added to marketed olestra snacks.

ACKNOWLEDGMENTS