点击显示 收起
1 From the Pennington Biomedical Research Center, Baton Rouge, LA (GAB, JCL, MM-W, SRS, JV, YD, LdJ, JR, and ML), and The Procter & Gamble Company, Cincinnati (ALE and JCP).
2 Supported in part by grant 96034323-3031 from the USDA and by the Procter & Gamble Co, Cincinnati. 3 Address reprint requests to GA Bray, Pennington Biomedical Research Center, 6400 Perkins Road, Baton Rouge, LA 70808. E-mail: brayga{at}pbrc.edu.
ABSTRACT
Background: Dietary fat has been implicated as a risk factor for cardiovascular disease and obesity.
Objective: We evaluated the effect on body weight, body fat, lipids, glucose, and insulin of replacing dietary fat with olestra in moderately obese men.
Design: Forty-five healthy overweight men were randomly assigned to 1 of 3 diets: control diet (33% fat), fat-reduced diet (25% fat), or fat-substituted diet (one-third of dietary fat replaced by olestra to achieve a diet containing 25% metabolizable fat). Body fat was measured by dual-energy X-ray absorptiometry and visceral and subcutaneous abdominal fat by computed tomography.
Results: Thirty-six men completed the 9-mo study. Body weight and body fat in the fat-substituted group declined by a mean (± SEM) of 6.27 ± 1.66 and 5.85 ± 1.34 kg, respectively, over 9 mo compared with 3.8 ± 1.34 and 3.45 ± 1.0 kg in the control group and 1.79 ± 0.81 and 1.68 ± 0.75 kg in the fat-reduced diet group. At 9 mo, the mean difference in body fat between the fat-reduced and fat-substituted groups was -4.19 ± 1.19 kg (95% CI: -6.57, -1.81), that between the control and fat-substituted groups was -2.55 ± 1.21 kg (-0.13, -4.97), and that between the control and fat-reduced groups was 1.63 ± 1.18 kg (3.96, -0.70). The men eating the fat-reduced diet asked for almost no extra foods, in contrast with the significantly higher requests (P < 0.05) from both of the other 2 groups.
Conclusion: Replacement of dietary fat with olestra reduces body weight and total body fat when compared with a 25%-fat diet or a control diet containing 33% fat.
Key Words: Body composition • dietary fat • obesity • feeding trial • olestra • men • Ole Study
INTRODUCTION
The relation of dietary fat to the development and treatment of obesity is controversial (1–5). Epidemiologic studies show a strong relation between the amount of dietary fat and the percentage of the population who are overweight (1). Similarly, in a population of Danish men, increased fat in the diet was associated with an increase in obesity (3). When dietary fat is reduced in clinical intervention studies, there is modest weight loss (1–5), although this weight loss may not be sustained (6). In contrast, the substantial increase in the prevalence of obesity in the United States over the past 20 y (7) has occurred in the face of a stable or declining percentage of energy from dietary fat, leading some to doubt the importance of dietary fat in the development of obesity (2). However, others have suggested that the decline in fat intake may be spurious, resulting from the selective underreporting of fat (8).
One problem with clinical intervention studies using lower-fat diets is that the composition of the diet must change to reduce the percentage of fat, thereby changing the diet’s energy density and hedonic properties. Thus, a blinded study of a low-fat diet has been nearly impossible. The availability of olestra, a fat substitute that cooks and has the mouth-feel of normal fats but cannot be digested in the intestine or distinguished in taste tests of snack foods (9), allows a method for the replacement of fat in the diet with foods that are indistinguishable from normal. This 9-mo feeding trial, called the Ole Study, was designed to test the hypothesis that substitution of olestra for dietary fat would significantly decrease body fat relative to a control diet containing 33% fat or to a reduced-fat diet containing 25% fat.
SUBJECTS AND METHODS
Subjects
A total of 164 subjects were screened for this protocol after they responded to an advertisement seeking men for 9 mo of employment in a feeding trial (Figure 1). To be eligible for the trial, the men had to be overweight [body mass index (BMI; in kg/m2) between 27 and 35] but otherwise healthy and able to participate in a 9-mo feeding study, be between the ages of 21 and 60 y, be nonsmokers, be relatively sedentary, be not consuming special diets, have no more than 3 bowel movements/d, have food preferences that would fit into the planned menus, and have a body weight that had not varied by > 3 kg in the past 6 mo. The major reason for ineligibility was not having 9 mo of uninterrupted time. The advertisements, protocol, and consent forms were approved by the Pennington Institutional Review Board, and subjects signed individual consent forms for screening and for inclusion in the protocol.
FIGURE 1. . Recruitment and study design. The number of subjects who were screened, the number of subjects rejected, the number of men stratified to each group at baseline, the reasons for withdrawal, and the number of subjects completing each study arm are shown.
Study design
This was a randomized, parallel-arm, controlled feeding study with 15 subjects assigned to each dietary treatment. Initial energy intakes were individualized by using an averaged 3-d record of energy expenditure determined with Tritrac motion sensors (Reining International, Madison, WI). Activity was also measured with the Tritrac motion sensors at the end of each experimental period. During the 3-wk run-in period, all subjects ate a diet containing 33% fat (control diet). At the time of randomization, the men in one study arm were assigned to continue eating the control diet, the men in the second arm were assigned to eat a fat-substituted diet in which one-third of the fat was replaced with olestra to produce a diet with 25% available fat energy, and the third group of men was assigned to eat a fat-reduced diet containing 25% fat.
The subjects were told that the study was testing the effect of 2 low-fat diets on blood pressure and cardiovascular disease risk factors to divert attention from the interest in weight change. They were also told that one diet would use olestra and that it might produce gastrointestinal side effects. In addition, they were told that we would provide one meal at the Pennington Biomedical Research Center 5 d/wk and would give them the rest of their food in takeout containers. They could request additional food from us if they wished. They were asked to return all uneaten food each day. Finally, they were asked to not change their activity level. A daily questionnaire asked them about changes in activity, smoking, consumption of vitamin and mineral supplements, any symptoms, and any concomitant medications. All subjects ate the evening meal at the Center on weekdays and took all other meals with them. They were asked to obtain all of their food from items supplied by the Center and to report any nonstudy foods consumed. They were allowed 1 alcoholic beverage/d and up to 5 caffeine-containing beverages/d. They were given a 3-d break from eating study meals on 3 occasions (Easter, July 4, and Labor Day holidays).
Diets
Olestra was incorporated into a variety of baked goods, salted snacks, and entrées. The foods containing olestra were indistinguishable from regular foods prepared in the usual manner. Diets were prepared at levels of 7524, 9196, 10 868, 12 540, 14 212, and 15 466 kJ (1800, 2200, 2600, 3000, 3400, and 3700 kcal/d), and the menus were taste-tested for acceptability by the subjects just after random assignment and in the final period of the trial. For energy levels between these values, participants received 1, 2, or 3 mandatory "energy booster" unit foods of 418 kJ (100 kcal) each, including a chocolate cheesecake, chocolate muffin, or Mexican cornbread. Participants could also request additional "snack pack" foods if they were hungry. The snack packs consisted of graham crackers and yogurt, cheese and crackers, bean dip and tortillas, or bagel chips with cream cheese. The control diet provided 33% fat, 52% carbohydrate, and 15% protein. The fat-reduced and fat-substituted diets provided 25% digestible fat, 58% carbohydrate, and 17% protein. The energy level of the fat-substituted and fat-reduced diets was designed to be 11% less than was determined during the run-in phase. This was accomplished by reducing the number of unit foods, the basal diet energy level, or both. However, all subjects were allowed to request additional snack packs if they felt hungry. They could also reduce the number of unit foods consumed if they were too full. The number of snack packs requested by each subject was recorded daily and provided one way of evaluating food intake. Ten meal plans were used in 5 meal rotations.
Food analysis
Daily menus at each energy level were collected over the course of the study, composited, and analyzed for protein, fat, moisture, and ash by standard methods. Carbohydrate was determined by difference. When fat-substituted diets containing olestra were analyzed, they had, as expected, amounts of fat comparable with those in the control diet because the analytic methods did not distinguish olestra from dietary triacylglycerol. To correct for this, the kitchen prepared identical meals in which the olestra was replaced by an equal volume of water. These meals had amounts of fat comparable with the fat-reduced diet (data not shown).
Outcome measures
At baseline, 3 mo, 6 mo, and 9 mo, we measured body fat, lean body mass, and bone mineral content by dual-energy X-ray absorptiometry (DXA) with a Hologic QDR 2000 absorptiometer (Hologic Inc, Waltham, MA). DXA was performed with the subjects wearing hospital gowns and after they had fasting overnight. At the same time intervals, intraabdominal fat area was measured at the L4-L5 intervertebral disc space with a GE High-Light Computed Tomographic instrument (Milwaukee). Data were stored on digital tape and analyzed by using ANALYZE software (AnalyzeDirect, Inc, Lenexa, KS) on a Sun Sparc 20 computer with twin RISC processors (Sun Microsystems, Inc, Santa Clara, CA).
Data analysis
The study was designed as a single cohort experiment. To estimate the sample size for each treatment group, a power analysis was done by using the variance in weight loss from 2 previous unpublished short-term trials with olestra that provided an SD of 2.96. A sample of 12 subjects in each cell at the end of the study was estimated to detect a difference in change from baseline to 9 mo of 4 kg between the treatment groups with an of 0.05 and power of 80%. Because BMI, age, and total cholesterol are associated with weight loss, we stratified the participants with respect to these 3 variables to achieve balance at baseline between the 3 treatment groups. All the screening variables were available at the time of random assignment for all 45 subjects. The 3 variables, BMI, age, and cholesterol, were used as continuous, noncategorized variables. The first principal component was calculated and subjects were then sorted with respect to it. A random number generator was then used to allocate the subjects into the 3 treatment groups. Body fat was analyzed as change from baseline at 3, 6, and 9 mo by using baseline body fat as a covariate in each model. To minimize the variance of the response in the case of independent replicates, the averages at each time point were used in calculations. The repeated-measures fixed design approach was used in the analysis with treatments as fixed effects and time points as repeated factors. All analyses were performed with SAS (version 8.0; SAS Institute Inc, Cary, NC). The medium variability of the means was assumed at the end of the study (10). The actual SD of weight loss at the end of the Ole Study was 3.46 kg.
RESULTS
Subjects
A total of 164 men contacted the center about the study. The largest group of ineligible men were unwilling to devote 9 mo to the study requirements. The 45 men enrolled in the study were selected from the 68 eligible participants by stratification of BMI, age, and screening cholesterol concentration. These men were randomly assigned into 3 groups of 15 participants each (Figure 1). There were 8 withdrawals during the trial: 2 from the fat-reduced diet group, 2 from the control diet group, and 4 from the fat-substituted diet group. Four subjects (2 in the fat-substituted group and 2 in the fat-reduced group) moved out of the area. Two subjects (1 in the fat-substituted and 1 in the control diet group) withdrew for personal reasons. One subject in the control diet group simply failed to return during the first period. One subject in the fat-substituted group withdrew because of loose stools. Mild gastrointestinal complaints occurred in all 3 groups throughout the study. During the third 3-mo period of the trial, one participant in the control group had a serious adverse event of rectal bleeding and diarrhea after a course of antibiotic therapy. The subject was seen in the emergency ward and Clostridium difficile was cultured. The diarrhea subsided and the subject returned to the study with no further problems. Because this problem occurred near the end of the trial, the data for the third experimental period for the control group were analyzed without this subject.
The baseline characteristics of the 3 randomly assigned groups are shown in Table 1. The mean (± SEM) age of the entire group was 36.8 ± 1.34 y, and the participants were on average obese with a BMI of 30.7 ± 0.37. There were no significant differences between the groups at baseline. Total cholesterol at screening was 5.27 ± 0.14 mmol/L (204 ± 5.28 mg/dL), but after 3 wk of the 33%-fat diet it dropped to 4.55 ± 0.12 mmol/L (176 ± 4.56 mg/dL) (P < 0.0001).
View this table:
TABLE 1 . Baseline characteristics of the 3 treatment groups1
Body fat
The change in body fat from baseline to each of the 3 measurement points is shown in Figure 2. During the first 3 mo, all 3 groups showed a parallel and significant loss of body fat, without a significant change in lean body mass (Table 2). During the second and third periods, however, there was a significant separation among the groups. At 9 mo, body fat was significantly lower in the fat-substituted group than in the control diet group (P < 0.05) or in the fat-reduced group (P = 0.001), in whom body fat had actually increased from the nadir (Figure 2). Subjects in the fat-substituted group lost 6.27 ± 1.66 kg body weight (6.70% of body weight; P < 0.0001; Table 2) and 5.85 ± 1.34 kg (P < 0.001) body fat over the 9 mo. The men in the control diet group lost 3.81 ± 1.34 kg body weight (4.29% of body weight; P < 0.0001) and 3.45 ± 1.0 kg (P < 0.003) body fat, whereas the fat-reduced group lost only 1.79 ± 0.81 kg body weight (1.94% of body weight; NS) and 1.68 ± 0.75 kg (P = 0.06) body fat. At 9 mo, the difference in body fat between the control and fat-reduced groups was 1.63 ± 1.18 kg; that between the control and fat-substituted groups was -2.55 ± 1.21 kg; that between the fat-reduced and fat-substituted groups was -4.19 ± 1.19 kg. There were no significant changes in lean body mass during the 9 mo in any of the groups (Table 2).
View this table:
TABLE 2 . Changes in body composition, intraabdominal fat distribution, and energy intake from baseline to each time point1
FIGURE 2. . Mean (± SEM) effect of diet on body fat determined by dual-energy X-ray absorptiometry. The baseline data for all 45 subjects in each group and for those completing each of the periods were used to calculate the changes in body fat. Completing the first period: n = 14, 14, and 15 for the control, fat-reduced, and fat-substituted groups, respectively. For the second period, n = 14, 14, and 13, and for the third period, n = 12, 13, and 11, respectively. There was a significant time x group interaction (P = 0.008) and an effect of time in the fat-substituted group (P = 0.04) and the fat-reduced group (P = 0.01). At 9 mo, there was a significant difference in body fat between the control and fat-substituted groups (P = 0.038) and between the fat-reduced and fat-substituted groups (P = 0.0008).
Changes in total abdominal cross-sectional fat area by computed tomography, including both visceral and subcutaneous fat, are shown in Table 2. Cross-sectional fat area declined by 3 mo in all 3 groups, but subsequently paralleled the pattern of change observed with total fat content measured by DXA. At 9 mo, the difference in subcutaneous body fat between the control and fat-reduced groups was 15.4 ± 16.2 cm2; that between the control and fat-substituted groups was -45.8 ± 16.7 cm2; and that between the fat-reduced and fat-substituted groups was -61.2 ± 16.6 cm2.
Diets and food intake
The analytic data for the menus fed at the 12 540-kJ/d (3000-kcal/d) level are shown in Table 3. There were 6–8 menus for each of the diets, with 1 menu prepared in duplicate. The analytic values were close to the targets, and the same was true at all other energy intakes. The 2 lower-fat diets had the same available fat, but the 25% fat-reduced diet provided a smaller weight (g) of food but a similar energy density. Twenty-nine men had energy intakes between 10 868 and 13 376 kJ/d (2600 and 3200 kcal/d), 4 ate < 10 450 kJ/d (2500 kcal/d), and 11 ate > 13 794 kJ/d (3300 kcal/d). Each day the kitchen recorded what was given to the subjects and what was returned. Subjects were asked about any nonstudy foods they ate. A total of 57 deviations were reported by men in the fat-substituted group, 50 by men in the fat-reduced group, and 61 by men in the control group. During the last week of run-in (baseline), the recorded mean energy intake for all men was 12 290 ± 280 kJ/d (2940 ± 67 kcal/d) (Table 2).
View this table:
TABLE 3 . Diet analysis of menus from each group1
Because the reporting of food intake is notoriously unreliable, the average number of extra unit foods requested by each of the 3 groups during each period is plotted in Figure 3. Almost no extra foods were requested by the men in the fat-reduced group (9.28 ± 1.33 kJ/d, or 2.22 ± 0.32 kcal/d), which was significantly lower than the request for additional foods by each of the other 2 groups. Although the fat-substituted group asked for more unit foods (275 ± 9.6 kJ/d, or 65.7 ± 2.3 kcal/d) than did the control group (129.6 ± 5.4 kJ/d, or 31.0 ± 1.3 kcal/d), this difference was not significant.
FIGURE 3. . Mean (± SEM) intake of units foods requested by all men in each group over each of the 3-mo intervals. See the legend to Figure 2 for the number of subjects in each period. There were almost no requests for food in the fat-reduced group but an increasing number with time in each of the other 2 groups. There was a main effect of treatment (P = 0.02) and a main effect of time (P = 0.01). At 9 mo, there was a significant difference in snack pack intake between the control and fat-reduced groups of 6479 ± 281 kJ/d (155 ± 67.4 kcal/d; P = 0.02) and between the fat-reduced and fat-substituted groups of -1447 ± 507 (-346.1 ± 121.4 kcal/d; P = 0.006).
The palatability of the diets was assessed using a 7-point Likert scale at the beginning of the randomization period and again during the final diet period. As shown in Table 4, there was little change in preference for the control diet over time, a deterioration of 1 unit for the fat-substituted diet, and a worsening of 1–2 units for the fat-reduced diet.
View this table:
TABLE 4 . Evaluation of the diets during the Ole Study1
DISCUSSION
This study compared the effect on body fat and visceral fat of 3 different diets: a control diet containing 33% fat, a fat-reduced diet containing 25% fat, and a fat-substituted diet containing the artificial fat olestra to achieve a diet containing 25% metabolizable fat. Body weight and body fat in the fat-substituted group declined significantly more (6.3 and 5.9 kg, respectively) over 9 mo than did body weight and body fat in the control (4.0 and 3.5 kg, respectively) and the fat-reduced (1.8 and 1.7 kg, respectively) groups.
During the first 3 mo, all 3 groups lost 3 kg fat (ie, 112 900 kJ, or 27 000 kcal, from fat stores), with no significant loss of lean body mass. This effect was unexpected for the control group because the foods that we provided were intended to maintain body weight. The similar loss of body weight over 3 mo implies a similar negative energy balance in all 3 groups. This response in the control group could have resulted if there was an increase in physical activity, if we supplied less food than was needed, or if the subjects were not eating all of the foods provided. From the daily records provided by the men and from Tritrac measurements (data not shown), there was no evidence of any change in activity level during the study. The minimal requests for supplemental foods (Figure 3) during the first 3 mo suggests that all 3 diet groups were eating less food than they needed. This may reflect the fact that even the 33%-fat diet given during the run-in period provided less energy than the subjects’ prestudy diets. Using a value of 6-g/d weight loss for each 1% decrease in dietary fat (1), we calculated that the dietary fat level in the prestudy diet eaten by these men was close to 39%. A decrease in fat from 39% to the 33% in the experimental diets would be consistent with the decline in serum cholesterol from 5.27 mmol/L (204 mg/dL) at screening to 4.55 mmol/L (176 mg/dL) at the end of the run-in period.
After the first 3 mo, body fat diverged in the 3 groups (Figure 2). The body weight and body fat of the men eating the control diet stabilized after 3 mo, when they had lost 3.34 kg or 11% of their body fat. The body fat of the men eating the fat-substituted diet, on the other hand, declined by 4.11 kg in the first 3 mo and continued to decline to a minimum of 5.85 kg or 19% below baseline by the end of 9 mo. During the last 6 mo (periods 2 and 3), the men eating the fat-reduced diet gained 1.62 kg body fat. This occurred even though the fat-reduced group did not ask for additional unit foods, whereas the other 2 groups increased their requests for unit foods. Because the unit foods were identical in the fat-reduced and fat-substituted groups, this suggests that the subjects eating the fat-reduced diet may have perceived that the overall diet was lower in palatability, particularly after the first 3 mo. Initial taste-testing confirmed that individual products (eg, muffins or cookies) made with olestra, regular fat, or low-fat recipes had similar palatability ratings. However, it is clearly possible that an overall diet that is low in fat may be perceived as less palatable over time than a higher-fat diet or one in which fat is substituted by olestra.
We know from studies with doubly labeled water that reports of food intake usually underestimate actual intake (11–14). The fat gain in the fat-reduced group from 3 to 9 mo could be achieved with an additional 85 kcal/d, which corresponds to only one nonstudy food per week, such as a typical fast-food quarter-pound cheeseburger. The most logical conclusion is that the subjects eating the fat-reduced diet found their diet less palatable after the first 3 mo and were eating more nonstudy foods.
Depletion of energy stores provides important signals for food ingestion. The present study indicates, however, that under some circumstances these signals can be bypassed. The amounts of carbohydrate and protein in the diets of the 3 groups were similar because the energy reduction was accomplished by removing fat from the diet or by replacing fat with olestra. Yet, the patterns of weight change were divergent, indicating that the changes in body fat cannot be attributed to changes in the amount of dietary protein or carbohydrate in the foods provided.
The fat-substituted diet contained 8% less available fat than did the control diet. Using the value of 6 g/d as the rate of weight loss for each 1% reduction in dietary fat, we can calculate a loss of body fat of 48 g/d. Over the last 180 d (periods 2 and 3) when the body weight of the control group was essentially stable, we would anticipate a weight loss in the fat-substituted group of 8.6 kg. About two-thirds of this, or 5.9 kg, would be from fat stores. This is approximately twice the 2.27 kg fat that was actually lost by the fat-substituted group relative to the control group. The slowing rate of weight loss seen in Figure 2 between 6 and 9 mo, along with the increased request for unit foods, suggests that deprivation signals from energy depletion were beginning to initiate the intake of more food. Previous short- and intermediate-term studies with olestra substitution showed that 25% of the energy reduction is made up for by increased food intake (15–24). In studies lasting 12–20 d, lean and obese men and women failed to compensate completely for the reduced fat and energy in the diet when they consumed olestra-containing food throughout the day (22–24). In the Ole Study, the plateau in weight loss in the fat-substituted group occurred after 9 mo, when these men had lost about one-half of the predicted amount of fat, suggesting that they compensated for 50% of the energy deficit.
Food mass and gastrointestinal distention may also be involved in the weight loss produced by the fat-substituted diet. The mass of food in the diet provided for the control and the fat-substituted groups was the same. As food is digested and absorbed, however, the fat-substituted diet leaves undigested olestra in the intestine as an unabsorbed mass. This residual fat mass in the middle and lower small intestine may generate signals that override the signals produced by energy depletion and thus allow the extra weight loss. The existence of gastrointestinal signals for feeding is well known. These include distension of the stomach, distension of the intestine, the release of peptides from the gastrointestinal tract, and absorption of nutrients. Data from several studies support the hypothesis that gastrointestinal distension is an important control mechanism for food intake and intermediate-term regulation of body fat (25–28). Using covert manipulation of dietary fat, Stubbs et al (25) found that men ate for the mass of food, not for its energy or carbohydrate content. Sparti et al (26) found that men ate a similar quantity (mass) of food when the choices were high in carbohydrate rather than high in fat, even though the energy intake was > 3 times higher when high-fat foods were eaten. Tremblay et al (27) also observed that food intake was not increased when a high-fat, very-low-carbohydrate diet was fed compared with a high-fat diet containing normal amounts of carbohydrate. They concluded, as do we, that food intake may be regulated by the quantity (mass) of food in the gastrointestinal tract. Finally, Bell et al (28) showed that the mass of food is similar even when energy intake diverges. Finally, a gastric bypass operation for obesity in which food is drained from a small gastric pouch into the jejunum produces more weight loss than when the food from a similar-sized pouch drains into the lower stomach in the vertically banded gastroplasty (29, 30). Loss of weight in patients treated with the lipase inhibitor orlistat may also be explained by undigested nutrients in the intestine (31, 32).
In conclusion, the results of this study showed that long-term reduction of dietary fat through substitution of fat with olestra produces an 18% loss of body fat in moderately overweight men. The diets appear to be well tolerated (33, 34).
ACKNOWLEDGMENTS
We thank Susan Mancuso, the clinical coordinator; Terri Keller, the research dietitian; and Heather Wiseman, the clinical trial monitor for their important work on this trial. We also thank the men who participated in the trial.
REFERENCES