Olestra is a mixture of hexa-, hepta- and octaesters of sucrose formed from long-chain fatty acids derived from edible oils. Olestra has physical properties and taste and cooking characteristics similar to regular fats and oils (Bernhardt 1988, Jandacek and Webb 1978, Kester 1993). However, it is not hydrolyzed by gastric enzymes (Mattson and Volpenhein 1972) or absorbed intact from the gastrointestinal (GI)⁵ tract (Miller et al. 1995). Because of these unique properties, olestra can serve as a replacement for conventional fats and oils, contributing no calories or fat to the diet. Olestra (Olean, Procter & Gamble, Cincinnati, OH) is approved for replacing 100% of the fat used to prepare savory snacks such as potato chips and crackers (Federal Register 1996).

Because olestra is lipophilic and is not absorbed, it can interfere with the absorption of lipophilic nutrients when eaten together with the nutrients. This interference occurs because olestra competes with the micelle-mediated absorptive process by which lipophilic nutrients such as the fat-soluble vitamins are absorbed. When olestra and fat-soluble nutrients are present together in the GI tract, a portion of the nutrients partitions into the olestra and thus becomes unavailable to the intestinal micelles for transport to absorptive sites (Jandacek 1982).

The potential for olestra to interfere with the availability of fat-soluble nutrients has been demonstrated in animal studies. For example, olestra increased fecal cholesterol excretion (Mattson et al. 1976) and reduced the concentration of vitamin A in the liver (Mattson et al. 1979) in rats. Also, olestra reduced tissue concentrations of vitamin A, vitamin E and 25-hydroxyergocalciferol in the pig, as discussed in other papers in this supplement (Cooper et al. 1997a-c, Daher et al. 1997a). On the basis of the partitioning mechanism, olestra would not be expected to affect the availability of water-soluble nutrients, and that was demonstrated in the pig studies.

Human studies have also shown that olestra reduces serum concentrations of cholesterol, carotenoids, vitamin E and 25-hydroxyergocalciferol (Crouse and Grundy 1979, Fallat et al. 1976, Glueck et al. 1979, Jones et al. 1991b, Koonsvitsky et al. 1997) but does not affect vitamin K status (Jones et al. 1991a, Koonsvitsky et al. 1997) or vitamin A absorption (Daher et al. 1997b). Recently, it has been shown that generic sucrose polyesters reduce the plasma concentrations of carotenoids and vitamin E (Weststrate and van het Hof 1995).

Tissue concentrations of fat-soluble vitamins can be restored to control concentrations in animals and humans eating olestra by adding extra amounts of the vitamins to the diet or to olestra (Koonsvitsky et al. 1997, Lafranconi et al. 1994, Mattson et al. 1979, Schlagheck et al. 1997). Studies in the pig have shown that the amounts of vitamins required to restore the tissue concentrations are linear functions of the amount of olestra in the diet (Cooper et al. 1997a and 1997b).

The purpose of this study was to determine the dose response of olestra on the status of the fat-soluble vitamins A, D, E and K, serum concentrations of carotenoids, vitamin B₁₂ absorption and the status of folate and zinc in normal, healthy subjects eating controlled diets. The study was conducted under conditions designed to maximize the opportunity for olestra and the nutrients to interact and thus maximize the opportunity for olestra to affect the availability of the nutrients. Foods containing olestra were eaten at every meal every day during the study and the subjects were asked not to eat between meals so that olestra and nutrients were always present simultaneously in the gut. In addition, the daily intakes of olestra were exaggerated relative to expected intakes from savory snacks.

SUBJECTS AND METHODS

Exclusion criteria included pregnancy or lactation, chronic use of drugs having the potential to interfere with vitamin absorption, use of tanning booths or high exposure to sunlight within the previous 2 mo, physician-recommended diet restrictions, or greater than average caloric need because of high activity level. Demographic and randomization parameters for the subjects entering the study are shown in Table 1.

Table 1. Demographics and randomization parameters for subjects entering the study [View Table]

Table 2. Specimen collection and measurement schedule for subjects consuming 0, 8, 20 and 32 g olestra/d [View Table]

Serum was analyzed for lipids, retinol, carotenoids, tocopherols, 25-hydroxyvitamin D metabolites [25(OH)D], 1,25-dihydroxyvitamin D [1,25(OH)₂D], phylloquinone, folate and zinc. Plasma was analyzed for prothrombin, des--carboxyprothrombin, PT and PTT. Clinical chemistry and urinalysis measurements were made at wk 0 and 8.

Urine collections (24-h) were made biweekly and total urine volume was determined. Aliquots were stored on dry ice until analyzed for -carboxyglutamic acid (Gla), zinc and creatinine. Vitamin B₁₂ absorption was measured at beginning and end of the study. The subjects were weighed and the females underwent a pregnancy test weekly.

Any undesirable symptom or change in health, reported either voluntarily or in response to a daily question about whether any changes in health had occurred, was recorded as an adverse event and was followed until the condition resolved. If the subject requested or if the clinician judged it appropriate, the subject was seen by a physician. If the undesirable symptom or condition was gastrointestinal, the subject was given a form describing nine common GI symptoms and was asked to record the severity (1 = mild, 2 = moderate, 3 = severe), number of episodes and duration of the symptom(s).

The diet was based on a 6-d rotating menu. A core menu was designed to provide 9205 kJ/d (2200 kcal/d) of energy. This menu then was adjusted to provide more or less energy to meet the needs of individual subjects. Each subject was assigned to one of seven energy intake levels ranging from 7531 kJ/d (1800 kcal/d) to 12,552 kJ/d (3000 kcal/d) at intervals of 837 kJ (200 kcal/d), on the basis of resting energy requirement. Resting energy requirement was estimated by the Harris-Benedict equation and an activity factor ranging from 25 to 50% of the subject's basal metabolic rate (Alpers et al. 1983). Subjects were fed to maintain their weight within 5% of their weight at the start of the study. Adjustments to the diet were made on the basis of body weights obtained weekly.

Any subject who missed more than 10% of the meals or who consumed less than 90% of the olestra or placebo food items was disqualified. If a subject missed more than one meal during the 2 d before a scheduled blood draw, or missed the dinner of the evening before a blood draw, data from that subject were excluded from the database.

Dietary constituents, with the exception of phylloquinone, were calculated by using the University of Minnesota Nutrition Data System, version 2.3 (Nutrition Coordinating Center, University of Minnesota), which contains data on 32 dietary constituents including macronutrients, vitamins and minerals as well as fiber. Phylloquinone content of foods were taken from Booth et al. (1993). With these databases, group mean intakes were calculated for macronutrients, fat-soluble vitamins and selected water-soluble micronutrients. For analysis, daily nutrient intakes were averaged over the four 2-wk periods corresponding to the blood draws, as well as over the total 8-wk period.

The diet provided about 35% of energy from fat, 15% from protein and 50% from carbohydrates, and 80-120% of the RDA for vitamins A, D, E, K and B₁₂, folate and zinc. To ensure that the requirements of most of the subjects were met, the RDA of 25- to 50-y-old males were used to determine target nutrient intakes (National Research Council 1989). To maintain a constant digestible fat content in the test groups, the amount of triglyceride displaced by olestra was added back as butter, margarine or vegetable oil.

Because the assay used to measure vitamin B₁₂ absorption is not affected by chronic intake, the intake of vitamin B₁₂ was allowed to exceed the 80-120% RDA range to keep zinc intake within that range and to keep protein intake at about 15% of energy. Calcium and iron intakes were not controlled to the 80-120% RDA range but were maintained constant across the groups. Carotenoid intake also was held constant across the treatment groups.

The subjects were given a 20-µg supplement of ergocalciferol daily, one third with each meal, to ensure that serum 25-hydroxyergocalciferol [25(OH)D₂] concentration would be sufficient to allow detection of any olestra effect on the absorption of vitamin D₂. In a previous study (Jones et al. 1991b), a 20 µg/d supplement of ergocalciferol produced a serum 25(OH)D₂ concentration greater than ~25 nmol/L, a value large enough to permit accurate measurement of any change resulting from olestra intake.

Subjects given 32 g/d olestra ate about 112 g of total lipid (32 g olestra plus about 80 g of triglyceride). This represented a significant increase in dietary lipid for the female subjects, whose daily fat intake normally would be 70-77 g (McDowell et al. 1994).

Table 3. Olestra, energy, and nutrient intakes for subjects consuming 0, 8, 20 and 32 g olestra/d1 [View Table]

Table 4. Average severity and percentage of symptom-days for all gastrointestinal (GI) symptoms, and diarrhea and cramping, reported by subjects consuming 0, 8, 20 or 32 g/d olestra [View Table]

Table 5. Serum -carotene concentration for subjects consuming 0, 8, 20 and 32 g olestra/d1 [View Table]

Table 6. Serum -tocopherol concentration normalized with respect to serum total lipids for subjects consuming 0, 8, 20 or 32 g olestra/d1 [View Table]

Olestra was fed in potato chips, muffins, biscuits and cookies. The olestra foods, other than the potato chips, were prepared by substituting olestra for all or part of the triglyceride in the recipes. The potato chips were fried in 100% olestra. Before it was used to prepare the foods, the olestra was heated to ensure that the olestra consumed by the subjects would contain the same kinds and amounts of degradation products as might be found in olestra used in the commercial preparation of savory snacks. Under commercial frying conditions, olestra undergoes the same degradation processes as traditional frying oils and the same degradation products are formed (Gardner and Sanders 1990, Gardner et al. 1992, Henry et al. 1992).

The foods were provided to the subjects in units consisting of a bag of potato chips, a single muffin, a single biscuit or three cookies. For a specific treatment group, each unit contained the same amount of olestra. For example, each unit given to the 8 g/d olestra group contained 2.7 g olestra, each unit given to the 20 g/d olestra group contained 6.7 g olestra, and each unit given to the 32 g/d olestra group contained 10.7 g olestra. The bags of potato chips given to the 8 and 20 g/d olestra groups contained a mixture of olestra chips and regular chips in the proportions needed to deliver the desired amount of olestra. Bags given to the 32 g/d olestra groups contained only olestra chips. The amounts of olestra in the food items were determined by assaying a representative number of units for each group.

Serum concentrations of retinol, - and -tocopherols, and carotenoids (-carotene, -carotene, lutein and lycopene) were measured simultaneously by HPLC (Catignari and Bieri 1983, Driskell et al. 1982). Serum samples were deproteinated with ethanol, and the lipids were extracted with hexane. The individual analytes were separated and quantified by HPLC using a silica column (120A, ODS, YMC, Wilmington, DE) and a dual-channel UV-visible detector (Waters, Milford, MA). Retinol, tocopherols, - and -carotene and lycopene were detected on one channel at 325, 292 and 450 nm, respectively. Lutein was detected simultaneously on the other channel at 450 nm. Under the HPLC conditions used, zeaxanthin and lutein have essentially the same retention; therefore the two are reported together. Concentrations were determined from peak areas by using internal calibration standards (Sigma Chemical, St. Louis, MO).

Serum concentrations of 25(OH)D₂ and 25-hydroxycholcalciferol [25(OH)D₃] were measured simultaneously (Eisman et al. 1977, Kao and Heser 1984). Serum samples were acidified and the 25-hydroxy metabolites were extracted using prepacked octadecylsilano silica cartridges (C-18 Bond Elute, Analytichem International, Harbor City, CA). The extracts were purified further by extracting with aminopropyl cartridges (NH₂ Bond Elute, Analytichem International). The two 25-hydroxy metabolites, 25(OH)D₂ and 25(OH)D₃ , were quantified simultaneously by HPLC using a silica column (Zorbax, 10-µm, DuPont, Wilmington, DE) and a UV detector (Hitachi D2500, Hitachi Instruments, Danbury, CT). The serum concentrations were calculated from peak areas using a calibration curve established with a 25(OH)D₃ standard sample (Nichols Institute Diagnostics, San Juan Capistrano, CA). Recovery was determined for each sample by adding a 25-hydroxy-(25[27]-methyl-³H)cholecalciferol standard (Amersham, Arlington Heights, IL) to the sample before the initial extraction.

Table 7. Serum 25-hydroxyergocalciferol [25(OH)D2] concentration for subjects consuming 0, 8, 20 and 32 g olestra/d1 [View Table]

Table 8. Serum 25-hydroxycholecalciferol [25(OH)D3], total 25-hydroxyvitamin D [25(OH)D] and 1,25-dihydroxyvitamin D [1,25(OH)2D] concentrations for subjects consuming 0, 8, 20 and 32 g olestra/d1 [View Table]

Serum concentrations of 1,25(OH)₂D were measured by a radioreceptor assay (Hollis 1986, Reinhardt et al. 1984) using a thymus receptor specific for 1,25-dihydroxyergocalciferol (Nichols). The serum samples were spiked with 1,25(OH)₂(³H-26,27)D, extracted with acetonitrile and purified further with activated C₁₈OH cartridges. Bound and unbound hormones were separated by incubation with charcoal. After incubation, the supernatant containing the bound hormone was decanted into scintillation tubes and counted in a scintillation counter (LS-8000, Beckman Instruments, Fullerton, CA).

Serum phylloquinone concentration was measured by HPLC, following the method of Haroon et al. (1986). Serum was denatured with ethanol and extracted with hexane; the extract was purified by eluting from an activated silica Sep-Pak cartridge (Waters). The elute was reductively extracted with zinc ions and zinc metal and was ejected into an HPLC system equipped with a 5-µm Ultrasphere ODS (Beckman) column and a fluorescence detector (Model 1050, Hitachi). The concentration of phylloquinone was determined from comparison of peak heights with an internal standard, dihydrophylloquinone.

Plasma concentrations of prothrombin and des--carboxyprothrombin (PIVKA-II) were measured by a chromogenic assay (O'Donnell et al. 1987). (PIVKA is an acronym for protein induced by vitamin K absence, II refers to factor II, prothrombin.) For the PIVKA-II assay, prothrombin was separated from des--carboxyprothrombin by barium adsorption. The des--carboxyprothrombin was converted to thrombin with snake venom (Echis carinatus, Sigma). Activity was measured (405 nm) by using the chromogenic substrate Chromozyme TH (BCL, Lewes, UK). Control samples (A-Fact and B-Fact, George King Biomedical) were assayed with each 10 test samples.

Serum and RBC folate concentrations were measured by a radioassay following the method of Dunn and Foster (1973). The assay is based on competition between radiolabeled and unlabeled folic acid for binding sites on a folate-binding substrate. ¹²⁵I-labeled folic acid was added to a sample of whole blood; the sample then was heated and the substrate, -lactoglobulin (Sigma), was added. After incubation, the supernatant was decanted and the precipitate was counted in a dual-channel gamma counter (Model 880708, Nuclear Data, Farmingdale, NY). Pterylglutamic acid (Sigma) was used as a calibration standard. Six control samples (Lyphochek, Bio-Rad Laboratories, Richmond, CA) were included for each 125 test samples.

The concentration of zinc in serum and urine was measured by atomic absorption spectrophotometry using a zinc hollow-cathode lamp (Perkin-Elmer, Norwalk, CT) and an air-acetylene oxidizing flame (Dawson et al. 1968, Parker et al. 1967). Absorption was measured at 214 nm.

The urinary concentration of Gla was measured by HPLC (Ferland et al. 1993, Haroon 1984). Gla and other amino acids were derivatized with o-phthalaldehyde (Sigma), and Gla was separated by gradient elution from a 3-µm C-18 column (Rainin Instruments, Woburn, MA) connected to a fluorescence detector (Model 980, Kratos Analytical, Ramsey, NJ). Fluorescence was determined by excitation at 334 nm and emission at 425 nm. Quantification was from ratios of test and standard (L--carboxyglutamic acid, Sigma) peak heights.

Urinary creatinine concentration was measured by a colorimetric assay (Fabiny and Ertingshausen 1971) using a commercial kit (EM Diagnostic Systems, Gibbstown, NJ). The kit employs the Jaffé reaction, in which alkaline picrate and creatinine react to form a red complex (Butler 1975). The concentration of creatinine is proportional to the rate of change of absorbency at 505/570 nm.

Urine was collected for 24 h, the samples were pooled, and the ⁵⁷Co content of a 4-mL aliquot was determined by gamma counting (Model 880708, Nuclear Data, Cleveland, OH). The percentage of the administered dose of ⁵⁷Co-labeled cyanocobalamin excreted in the urine was determined by comparing the counts/min per milliliter of urine to the counts/min from a reference standard (Cobatope-57, Squibb). The results were normalized with respect to urinary creatinine excretion.

Table 9. Serum phylloquinone concentration for subjects consuming 0, 8, 20 and 32 g olestra/d1 [View Table]

Table 10. Plasma des--carboxyprothrombin concentration, prothrombin concentration and excreted urinary -carboxyglutamic acid (Gla) for subjects consuming 0, 8, 20 and 32 g olestra/d1 [View Table]

Repeated-measures ANOVA was conducted as the first step in analyzing nutrient concentrations in blood and urine samples collected as a function of time. Age and BMI were used as covariates. The p-value of each F-test in the repeated ANOVA table was corrected by using the Huynh-Feldt adjustment (Huynh and Feldt 1970). This adjustment corrects for the fact that correlation between observations on a subject may be induced by remeasuring the observation.

If the Huynh-Feldt adjusted p-values indicated significance for either a dose effect or a covariate effect, a separate ANOVA was run at each time point. If neither covariate had a significant effect on the response, a two-way ANOVA with no covariate was performed, and the olestra effect was determined via the protected LSD test. If the covariate effect was significant, or if the covariate relationships were not the same in all treatment groups, the protected LSD test was conducted on the treatment means adjusted for the covariate effect.

Whenever the gender-by-treatment interaction was significant, comparisons were performed on the group means within each gender. Otherwise the comparisons were conducted on the combined data.

Trend tests were conducted to assess the strength of apparent dose responses, using standard linear regression analysis.

All analyses were carried out at the two-tailed 0.05 significance level. The data were analyzed using either SAS for Windows (Version 6.08) or PC SAS (Version 6.05) software (SAS Institute, Cary, NC).

RESULTS

Mean intakes of olestra, energy, macronutrients and key micronutrients, averaged across 8 wk, are shown in Table 3. Olestra intake was on target; on average, the groups ate 98.8% of the target amounts. Energy intake among the groups differed by <5%. Contributions to total energy from protein, fat and carbohydrate were near the 15:30:55 target and did not differ among the groups. Energy from saturated, monounsaturated and polyunsaturated fat was in the ratio of 1.1:1.0:0.7, near the targets of 1:1:1 (data not shown).

Intake of nutrients targeted to be within 20% of the RDA fell within that range (Table 3). Vitamin B₁₂ was an exception, as expected, because it was considered more important to keep zinc intake and energy from protein within targets. Intakes of vitamins A and D increased slightly with increasing doses of olestra. Vitamin A intake increased as a result of adding corn oil and corn oil margarine, which contained vitamin A, to the diet to replace the fat displaced by olestra. Vitamin D intake increased with olestra dose because the added-back corn oil margarine was fortified with vitamin D.

A significant olestra dose response was observed for serum iron concentration for males but not for females (Table A in the appendix). The dose response was not accompanied by changes in hematocrit, hemoglobin or RBC. The serum concentrations of other minerals, including calcium, were not affected by olestra.

Common GI symptoms were reported by subjects in all groups, including placebo. These included loose stools, fecal urgency, bloating, nausea, diarrhea, abdominal gas and cramping, and flatulence. The symptoms abated without cessation of placebo or olestra consumption but often recurred. They did not worsen with continued olestra consumption, nor were they more severe when they recurred. The average severity of the symptoms and the percentage of possible symptom-days are shown in Table 4 for all GI symptoms combined, and for diarrhea and cramping, broken out because diarrhea is the symptom of most concern. A symptom-day, used because of the intermittent nature of the symptoms, is defined as a day on which one or more GI symptoms were experienced with more than usual frequency, as reported by the subjects. For a given symptom, the maximum possible number of symptom-days is obtained by multiplying the number of subjects in the test group by the number of days of the study. The percentage of symptom days was dose responsive with respect to olestra intake; however, the average severity of the symptoms was essentially the same regardless of olestra intake and not different than the average severity of the symptoms reported by the subjects in the placebo group.

Table 11. Urinary excretion of 57Co-cyanocobalamin, normalized with respect to creatinine, for subjects consuming 0, 8, 20 and 32 g olestra/d1 [View Table]

Olestra had no significant effects on the serum concentrations of cholesterol, triglycerides, or total lipids (Table D in the Appendix) or on the serum concentration of retinol (Table E in the Appendix). The serum concentration of -carotene was reduced by olestra in a dose-responsive manner. At wk 8, the concentration for the subjects given 8 g/d olestra was 38% of control (Table 5). The effect was evident by wk 2. The effects of olestra on the serum concentrations of -carotene, lycopene and total carotenoids were similar to the effect on serum -carotene concentration (Tables F-H in the Appendix); however, the effect on serum concentration of lutein (+ zeaxanthin) was less. Serum lutein (+ zeaxanthin) concentration for the group given 8 g/d olestra was 86% of control (Table I in the Appendix). Serum concentrations of the carotenoids normalized with respect to serum lipids showed responses to olestra similar to those of nonnormalized concentrations (data not shown).

Serum concentrations of -tocopherol normalized with respect to total lipids decreased in a dose-responsive manner with increasing olestra dose. At wk 8, serum -tocopherol concentrations for the groups given 8, 20 or 32 g/d olestra were 94, 82 and 82% of control, respectively (Table 6). The dose response was essentially complete within the first 2 wk. Nonnormalized serum -tocopherol concentrations and normalized or non normalized serum concentrations of total tocopherol showed similar responses to olestra (data not shown).

Olestra reduced the serum concentration of 25(OH)D₂ (Table 7). At wk 4, 6 and 8, the mean 25(OH)D₂ concentrations for the subjects given 8 or 32 g/d olestra were consistently about 78 and 73% of the control value, respectively. The concentration for the group given 20 g/d olestra was more variable; values ranged from 73 to 86% of control at the different time points.

No olestra dose response was observed on the serum concentrations of 25(OH)D₃ (Table 8 and Table J in the Appendix), 25(OH)D (Table 8 and Table K in the Appendix) or 1,25(OH)₂D (Table 8 and Table L in the Appendix).

Olestra reduced serum phylloquinone concentration in a dose-responsive manner. At wk 8, serum phylloquinone concentrations for subjects given 8, 20 or 32 g/d olestra were 72, 67 and 56% of control, respectively (Table 9). Serum phylloquinone concentration for the control group varied substantially from week to week because of differences in the amount of phylloquinone eaten at the evening meal before the blood draw, as will be discussed in the following section.

No effect of olestra was found for plasma concentration of des--carboxyprothrombin (PIVKA-II assay) or prothrombin, or for urinary Gla excretion (Table 10 and Tables M, N, and O in the Appendix). In addition, there was no effect on PT or PTT (Tables P and Q in the Appendix).

Table 12. Serum folate and red blood cell (RBC) folate for subjects consuming 0, 8, 20 and 32 g olestra/d1 [View Table]

Table 13. Serum zinc and 24-h urinary zinc concentrations, normalized with respect to creatinine, for subjects consuming 0, 8, 20 and 32 g olestra/d1 [View Table]

No effect of olestra was observed on the absorption of vitamin B₁₂. The excretion of ⁵⁷Co-cyanobalamin at the end of the study, after a breakfast containing the total daily dose of olestra, showed no significant differences from the excretion measured after a placebo breakfast at the beginning of the study (Table 11).

Olestra did not affect the serum or RBC concentration of folate (Table 12) or the serum or urinary concentration of zinc (Table 13). Serum and RBC folate concentrations and serum and urine zinc concentrations are shown in Tables R, S, T, and U in the Appendix for all time points.

DISCUSSION

The effects on the fat-soluble vitamins measured here are exaggerated relative to those likely to occur when snacks foods made with olestra are eaten in real life for two reasons. First and most important, olestra was eaten 42 times during each repeating 14-d period in the study. It is estimated that the average consumer will eat olestra snack foods about five times in 14 d; the 90th-percentile consumer will eat them about 10 times in that period (Webb et al. 1997). In addition, it is estimated that only 8% of meals contain a snack food for the average consumer, and only 18% of the meals for the 90th-percentile consumer. The dietary pattern used in this study provided a substantially greater opportunity for olestra to interfere with the absorption of the fat-soluble nutrients than is likely to occur in real life.

The second factor that exaggerated the effects was the daily olestra intake. For example, the highest olestra dose tested, 32 g/d, is more than four times the estimated 90th-percentile chronic olestra intake of olestra from savory snacks, 6.9 g/d (Webb et al. 1997).

The observation that olestra did not affect the status or absorption of water-soluble micronutrients is consistent with results from studies in the domestic pig (Cooper et al. 1997a-c) and with the partitioning mechanism (Jandacek 1982). Neither serum folate, which responds rapidly to changes in folate intake (Herbert 1967, Wagner 1984), nor RBC folate, which reflects the total body pool of folate at the time of red blood cell formation (Bailey 1990, Jacobs et al. 1984), was affected by olestra. Similarly, serum zinc, which is depressed in cases of zinc deficiency, and the urinary excretion of zinc, which has been shown to decline with the development of zinc deficiency in healthy subjects (Gibson 1990), were not affected. The excretion of ⁵⁷Co-labeled cyanocobalamin within 24 h of a single low oral dose of ⁵⁷Co-labeled cyanocobalamin (Schilling test) was not changed significantly as a result of olestra consumption.

The apparent dose-responsive change in serum iron concentration observed for male subjects is unlikely to represent an effect of olestra on iron absorption for several reasons. It was not accompanied by changes in hematocrit, hemoglobin or RBC and there was not a similar trend for the females whose average iron intake, 0.8 RDA, was about half that of the males, 1.6 RDA. Further, zinc absorption was unaffected. If olestra affects the absorption of one mineral, it would be expected to affect the absorption of others.

Further evidence that olestra does not affect iron availability from the diet comes from other studies. In another 8-wk human study, serum total iron, serum ferritin concentration or total iron-binding capacity was not affected by 8, 20 or 32 g/d olestra (Schlagheck et al. 1997). Serum ferritin is a particularly sensitive indicator of iron status and is the first to change when iron status begins to decline (Cook and Skikne 1989). Finally, olestra did not affect liver iron concentration, measured in pigs fed up to 7.7% olestra for 12 wk (Cooper et al. 1997a and 1997c) or 5.5% olestra for 26 wk (Cooper et al. 1997b).

Effects of olestra on serum concentrations of carotenoids have been seen in other studies (Koonsvitsky et al. 1997, Weststrate and van het Hof 1995). With the exception of lutein, all carotenoids were affected similarly by olestra. Eight g/d olestra reduced serum lutein (+ zeaxanthin) concentration by 14%, compared with about 60% for -carotene, -carotene and lycopene. The smaller effect on lutein is consistent with the partitioning mechanism, which leads to the prediction that the more lipophilic the molecule, the greater the olestra effect (Jandacek 1982). Lutein contains two hydroxyl groups and therefore is less lipophilic than the other carotenes. The octanol-water partition coefficient, generally expressed in log units (log₁₀ p_c ), of a molecule is a measure of its lipophilicity; the log₁₀ p_c value of lutein is 14.8 compared with 17.6 for -carotene (Cooper et al. 1997e), indicating that -carotene is almost 30 times more lipophilic than lutein.

The effects on carotenoid availability observed in this study were greater for a similar daily intake of olestra than effects measured by Koonsvitsky et al. (1997). This difference was most likely a result of the different dietary patterns used in the studies. In this study, olestra was present in the gut whenever carotenoids were present, providing maximum opportunity for olestra and carotenoids to interact. Under these conditions, 20 g/d of olestra reduced serum -carotene to 38% of control. In the study reported by Koonsvitsky et al. (1997), the subjects ate olestra with meals, at their discretion, but were not restricted from eating other foods between meals, which means that some portion of their carotenoid intake was in the absence of olestra, thereby reducing the opportunity for the interaction to occur. Under those conditions, 18 g/d olestra reduced serum -carotene concentration to about 73% of control, an effect less than one half of that measured in the present study.

The effects measured in either of the two studies are exaggerated relative to what is likely to occur in real life. The degree to which carotenoid availability may be affected by eating snack foods prepared with olestra in real life was assessed by Cooper et al. (1997e). From a consideration of the frequency at which snack foods and carotenoid-containing foods are eaten together, obtained from a menu census survey, and the reduction in absorption when the two are eaten at the same time, measured in this study, it was calculated that carotenoid availability may be reduced by 6-10%. This is still a conservative estimation because it is based on the assumption that all snacks eaten by the consumer will be olestra snacks.

The absence of an effect of olestra consumption on serum retinol concentration is not surprising because serum retinol concentration does not reflect short-term change in vitamin A intake unless tissue stores of the vitamin are severely depleted, a process requiring months of inadequate vitamin A intake (Olson 1984). However, olestra has been shown in another study not to significantly affect the absorption of preformed vitamin A (Daher et al. 1997b).

The rapid effect of olestra on serum -tocopherol observed in this study has also been seen in other human studies (Glueck et al. 1982, Koonsvitsky et al. 1997, Mellies et al. 1983 and 1985) and in pig studies (Cooper et al. 1997a, 1997b and 1997c). Rapid responses of serum -tocopherol concentration to changes in vitamin E intake, not produced by olestra, have also been reported by others. For example, Dimitrov et al. (1991) reported that the plasma concentration of -tocopherol increased to a new steady-state value within 10 d when subjects were given supplements of 440-1320 mg of vitamin E daily. When the supplements were withdrawn, plasma -tocopherol concentrations declined to the presupplement levels within 5 to 10 d.

Even though serum -tocopherol responded rapidly to olestra intake and reached a new steady-state level within 2 wk, this response reflects changes in nutritionally important vitamin E stores. In young and adult animals, vitamin E concentrations in plasma, liver, spleen, testis, brain, lung and heart were found to decline rapidly when subjects 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.

As with the carotenoids, the effect of olestra on vitamin E status measured in this study was greater than that measured in free-living subjects. In the study reported by Koonsvitsky et al. (1997), 18 g/d olestra reduced serum -tocopherol concentration by ~6%. This compares to an 18% reduction in the 20 g/d olestra group in the present study. Again, individuals eating olestra snacks in real life will likely experience reductions in vitamin E absorption considerably smaller than those observed in either of these studies because of the decreased opportunity for olestra and vitamin E to interact under real-life dietary patterns. In any case, the effect of olestra on vitamin E status can be offset by incorporating additional vitamin E into olestra foods (Cooper et al. 1997a, Schlagheck et al. 1997).

Serum total 25(OH)D concentrations in this subject population ranged from 64 to 71 nmol/L at base line, consistent with values of 39-66 nmol/L reported by others (Arnaud et al. 1977, Delvin et al. 1979, Jones 1978, Jones et al. 1991a). As a result of the 20 µg/d vitamin D₂ supplement, the contribution of dietary vitamin D₂ to total vitamin D status was about 68% by the end of the study, greater than the 11-49% contribution typical of people living in North American locations in winter (Arnaud et al. 1977, Delvin et al. 1979, Haddad and Hahn 1973, Jones 1978, Jones et al. 1991a).

The finding that olestra reduced serum 25(OH)D₂ concentration agrees with findings from other studies in humans (Jones et al. 1991a) and in pigs (Cooper et al. 1997b and 1997c). In contrast to the effect on carotenoids and vitamin E, the effect of olestra on serum 25(OH)D₂ measured in this study was about the same as that measured in free-living subjects (Jones et al. 1991a). When free-living subjects ate 18 g/d olestra for 6 wk, serum 25(OH)D₂ concentration was 81% of control. In this study, serum 25(OH)D₂ concentration for the group given 20 g/d olestra was ~80% of control when averaged over wk 4, 6 and 8. The responses of serum 25(OH)D₂ concentration to olestra were similar in the two studies because dietary vitamin D₂ in both studies came primarily from a vitamin D₂ supplement that was eaten with meals that included olestra foods. The effect of olestra on serum 25(OH)D₂ can be offset by adding the vitamin to foods made with olestra (Schlagheck et al. 1997).

Olestra did not significantly affect overall vitamin D status of the subjects even though the dietary contribution to total vitamin D status was ~68%. There were no significant changes in serum 25(OH)D, 1,25(OH)₂D, calcium or phosphorus concentrations, or serum alkaline phosphatase activity. These findings suggest that olestra will have no significant effect on the vitamin D status of people who obtain a significant amount of their vitamin D from the diet.

Olestra had a rapid effect on serum phylloquinone; the dose response was essentially complete in 2 wk. This agrees with findings by others (Allison et al. 1987, Ferland et al. 1993, Suttie et al. 1988). Circulating phylloquinone has a short half-life, ~104 min (Shearer et al. 1974). Because of this rapid clearance, serum phylloquinone concentration primarily reflects recent intake of the vitamin. This is illustrated by a comparison of the serum phylloquinone concentration for the control group with the amount of phylloquinone contained in the evening meal of the days preceding the blood draws. For example, the subjects ate 210 µg of phylloquinone at dinner on the evenings before the wk-2 and the wk-8 blood draws; serum phylloquinone concentrations on the following days were 1.33 and 1.60 nmol/L, respectively (Table 9). At dinner on the day before the wk-4 blood draw, the subjects ate 17 µg of phylloquinone; at dinner before the wk-6 blood draw they ate 51 µg. Serum phylloquinone concentrations at wk 4 and 6 were 0.51 and 0.64 nmol/L, respectively. Because serum phylloquinone concentration changes so rapidly with intake, its use as a measure of vitamin K status over the long term is restricted unless intake is controlled carefully.

Despite the effect on serum phylloquinone, a lack of change in the functional assays, i.e., urinary Gla excretion and plasma concentration of des--carboxyprothrombin, indicates that olestra did not affect vitamin K functional status. Urinary Gla excretion is a sensitive measure of vitamin K nutritional status. Changes in urinary Gla concentration reflect changes in vitamin K function in the liver as well as in other tissues.

The Simplastin-Ecarin (E. carinatus) assay, an indirect measure of circulating, biologically active, fully -carboxylated prothrombin, showed that olestra doses of 18 and 20 g/d did not affect vitamin K status in free-living subjects (Jones et al. 1991a, Koonsvitsky et al. 1997). The PIVKA-II assay measures incomplete carboxylation of plasma prothrombin (des--carboxyprothrombin) and is as sensitive, or more so, as urinary Gla excretion to changes in vitamin K status. The PIVKA-II assay is several orders of magnitude more sensitive than measurements of plasma prothrombin concentration or PT and an order of magnitude more sensitive than the Simplastin-Ecarin (E. carinatus) assay (Suttie 1992). Recently, it has been shown that the serum concentration of undercarboxylated osteocalcin is also a sensitive measure of vitamin K nutritional status (Sokoll et al. 1997). At the time this study was conducted, the response of serum undercarboxylated osteocalcin to changes in dietary vitamin K was not fully documented. The absence of any significant effect of olestra on either PIVKA-II or urinary Gla is strong evidence that olestra did not affect vitamin K-dependent proteins.

The length of the study was sufficient to reveal effects on vitamin K function, if any had occurred. The turnover times of vitamin K clotting factors and other vitamin K-dependent proteins, 60 h, are such that 95% of any effect on vitamin K functional parameters would be seen within 2 wk (Ferland et al. 1993). Other researchers have shown that vitamin K status changes within 2-3 wk in response to changes in phylloquinone intake (Allison et al. 1987, Suttie et al. 1988). Sokoll et al. (1997) observed that urinary Gla excretion declined significantly after 10 d consumption of a diet providing 100 µg/d phylloquinone.

The effects on micronutrients observed in this study are in general agreement with those observed in the pig (Cooper et al. 1997a-c) and are consistent with the idea that olestra affects the absorption of lipophilic dietary components via the partitioning mechanism. The absorption of water-soluble nutrients was unaffected; the magnitude of the effect on fat-soluble nutrients correlated roughly with the lipophilicity of the nutrient. The 8 g/d olestra amount reduced serum -carotene by 62% (log₁₀ p_c = 17.6), serum lutein (+ zeaxanthin) by 14% (log₁₀ p_c = 15), and serum -tocopherol by 6% (log₁₀ p_c = 12.2). Because partition coefficients are in log units, a difference of one means that the lipophilicity differs by a factor of 10.

In this study, the occurrence and severity of GI symptoms were monitored by means of questionnaires completed by the subjects. GI symptoms were reported to some extent by all groups. The symptoms were mild to moderate in severity and transitory in nature. A significant number of the GI symptoms were related to stool consistency. For example, there was a dose-responsive increase in the frequency of reports of loose stools and/or "diarrhea." It is expected that olestra would affect the consistency of the stool because of its properties, highly lipophilic and nonabsorbed. Large amounts of a lipophilic substance in the bowel can disrupt the fecal matrix and prevent the formation of firm, formed stools as evidenced by the stool-softening effects of liquid petrolatum and oils such as olive oil (Curry 1986). This disruption of the stool matrix produces stools that are looser than normal and that, in extreme cases, may be perceived as diarrhea, but would not be accompanied by the clinical features of true diarrhea such as water and electrolyte loss. This stool softening property of olestra also occurred in the pig studies in which increased incidences of pasty feces were seen (Cooper et al. 1997a-d).

The GI symptoms did not affect compliance with the study protocol. None of the subjects dropped or were withdrawn from the study because of symptoms. Only three subjects, one in the 20 g/d group and two in the 32 g/d group, stopped eating olestra foods temporarily (48 h or less). The percentage of meals eaten by the subjects met protocol requirements and the amounts of olestra eaten were on target.

The GI symptoms also did not affect the ability to assess the nutritional effects of olestra. There was no evidence of general malabsorption; the observed effects on nutrient availability were consistent with the partitioning mechanism (Jandacek 1982) and with effects seen in the pig (Cooper et al. 1997a-c). The availability of water-soluble nutrients was not affected; the availability of fat-soluble nutrients was effected in a dose-responsive manner; the more lipophilic ones were more affected than the less lipophilic ones.

The reports of GI symptoms in this study are exaggerated in relation to what is likely to occur when individuals eat snacks prepared with olestra in real life because the dietary pattern used in the study resulted in large amounts of olestra being in the digestive tract at all times. In a study in which more than 3000 subjects ate potato chips prepared with either olestra or triglyceride, at their choosing, over a 5-mo period, in amounts of their choosing, < 1% of the subjects who ate either kind of chips voluntarily reported GI symptoms (Lawson et al. 1997). There was no significant difference between the groups in the number of reports of symptoms.

ACKNOWLEDGMENTS