Appetite, 1991, 17, 213-219
Effects of Pure Sugar vs. Mixed Starch Food Intake JUDITH
Fructose
Loads on
RODIN
Department of Psychology,
Yale University
Using a within-subject design, we gave subjects three different 520-530 kcal preloads in the form of puddings in a randomized fashion at weekly intervals. The puddings contained either 5Og of fructose or glucose as the sole carbohydrate source in a protein and fat mixture, or 50 g fructose plus 15 g of starch. Food intake was assessed 2.25 h after the preload was completed. Blood was drawn throughout and assayed for concentrations of glucose and insulin. When the
preload contained fructose alone as the sole source of carbohydrate, subjects ate significantly fewer calories and less fat than when the preload contained glucose alone. When starch was added to the fructose preload, there was no significant reduction in calorie and fat intake. Effects on food intake paralleled the rise in plasma insulin levels produced by the different preloads. Implications for use of fructose as an adjunct to weight control efforts are discussed.
Studies have demonstrated that ingested fructose as opposed to glucose is absorbed more slowly (Durnin & Womersley, 1974; Schutz et al., 1984), generates greater thermogenesis (Jarrett et al., 1979; Schwarz et al., 1989; Wall et al., 1975), is metabolized directly by the liver (Schutz et al., 1984; Tappy et al., 1986; Wotecki et al., 1982) and produces a reduction in subsequent food intake (Rodin, 1990; Rodin et al., 1988; Shafer et al., 1987; Spitzer 8z Rodin, 1987). These striking differences are
important since fructose is found in natural foods such as fruit and honey and in many processed foods and drinks in the form of high fructose corn syrup (Hue, 198 1; Nilsson & Hultman, 1974; Segal et al., 1985), and therefore is widely consumed. The observation of different effects of these sugars on food intake leads quite naturally to speculations about the possible use of fructose as a weight-control intervention. For example, our studies have demonstrated that when subjects drink fructose-sweetened beverages (50 g in 500 ml) in the fasted state, they subsequently (from 40min to 2 h later) eat fewer calories and less fat than when they &ink isocaloric glucose-sweetened beverages. Fructose may actually suppress intake under certain conditions since intake after fructose consumption appears even lower than after drinking non-caloric, but volumetrically equivalent water loads. Supported by NIH grant RR125, Division of Research Resources to the Yale Clinical Research Center and by a grant from Xyrofin-Finnsugar. The author is grateful to Nancy Cummings for her assistance in conducting the study and to Heinz Schneider, Ph.D., of R&D Wander Ltd, Neunegg, Switzerland for his assistance in obtaining the Alburone and Liprocil for the test preloads. Address correspondence to: Dr. Rodin, Department of Psychology, Yale University Box 11A Yale Station, New Haven, CT 06520, U.S.A. 0195-6663/91/060213
+07 $03.00/O
0
1991 Academic
Press Limited
214
J. RODIN
While in our laboratory the relationship between fructose ingestion and lowered calorie and fat intake at the subsequent meal appears highly replicable, the effect seemed to us to be confined to fructose-sweetened beverages. When we presented 50 g of fructose as compared to glucose as part of a mixed nutrient breakfast meal including egg, bread, muffin, margarine and milk, it did not suppress subsequent intake (Rodin et al., 1988). With the mixed meal, plasma glucose and insulin levels were not significantly different in the glucose and fructose conditions. These findings stand in contrast to our observations with pure sugar liquid preloads, where plasma glucose and insulin were significantly elevated after the glucose drink, with little rise after the fructose drink. Many variables, such as the physical form of the food, amount of fiber and protein and fat content, influence glucose tolerance to a mixed meal (Collier et al., 1984; Crapo et al., 1982; Jenkins et al., 1978; Nuttall et al., 1983; Nuttall et al., 1984). The mixed meal also contained a large amount of complex carbohydrate, and this source of carbohydrate may overshadow the effect of the different sugars on carbohydrate metabolism. If the effects of fructose on food intake are confined only to beverages with no other nutritive constituents, its potential therapeutic application for weight control would be limited. Recently, however, in an effort to generalize from previous studies examining the thermic effects of pure glucose vs. pure fructose, Jequier and his colleagues assessed the magnitude of the thermic effect of a mixed meal containing either glucose or fructose as the sole source of carbohydrate (Schwarz et al., 1989). They observed a greater dietary-induced thermogenesis with the meal containing fructose, accompanied by a higher respiratory quotient, which suggested an enhanced rate of carbohydrate oxidation. In addition, higher plasma lactate concentration but lower plasma insulin and glucose responses were observed with the fructose meal when compared to the glucose meal. Based on these encouraging results, we considered the effects on subsequent food intake of a mixed nutrient preload with either glucose or fructose as the only carbohydrate source, compared to the same preload in which another source of carbohydrate was also provided. METHODS
Subjects Six subjects, three men and three women, participated in the study. The mean age of the subjects was 35.622.4 years and the mean percentage overweight was + 3 k3.6. Each subject was tested in three sessions, separated by 1 week. Subjects were assigned at random to a counterbalanced order of presentation of the three types of preload solutions-one at each session. Subjects were screened for symptoms of hypoglycemia or undiagnosed diabetes. Criteria for exclusion from the study included a fasting blood sugar >6*11 mmol/l, blood glucose concentrations 11.10 mmol/l during the study, and glucose 2 7.77 mmol/l at the conclusion of the study (Felig et al., 1981). No subjects were excluded on the basis of this evaluation. Procedure Subjects were recruited by a newspaper advertisement and were initially interviewed by telephone. A standardized screening procedure was followed to exclude
FRUCTOSE
AND FOOD INTAKE
215
potential subjects who reported medical problems, such as diabetes, heart disease, or hypertension, and people taking medications that might affect carbohydrate metabolism. Potential subjects were also excluded if they reported gaining or losing > 4.54 kg in the last 3 months or if they indicated that they were currently dieting. After the screening, subjects were scheduled for the study and were asked to eat no food and drink only water after 2100 hrs on the night preceding the study. They were also asked to eat 150-200 g of carbohydrate per day in the 3 days preceding the study. Bondy & Rosenberg (1980) suggested that only in circumstances where the patient was following a severe hypocaloric intake would this recommendation involve any change in the subject’s usual dietary habits. A 24-h food record was provided by each subject and was evaluated to establish whether subjects had eaten the recommended amount of carbohydrate. The mean amount eaten was 204 g. Participants were weighed upon admission as outpatients to the Clinical Research Center at Yale-New Haven Hospital. Blood pressure and pulse were measured and an intravenous (IV) line was inserted in the antecubital vein, which was kept open with a slow drip of isotonic saline. Subjects were next given one of three types of preloads. The composition of two of the preloads followed the test meal of Schwarz et al. with two changes: 1) 50g of fructose or glucose, rather than 75g, was used to make the amount of sugar comparable to all other preloads used in our laboratory; and 2) a pudding, rather than liquid meal, was constructed by allowing the mixtures to stand for several hours before serving. The formula contained 35g powder of Alburone (milk protein, Sopharga, France), 23g Liprocil (lipid containing 80% MCT, Sopharga, France), 50 g glucose or fructose, 7 g decaffeinated coffee, and 200 ml deionized mineral water. The pudding was 520 kcal with the glucose and 530 kcal with the fructose, and was 24% protein, 41% fat and 35% CHO. In the third preload, complex carbohydrate in the form of 15 g of cornstarch was added, and 14g of Alburone was deleted. The simple sugar was 50 g of fructose. The pudding was 530 kcal with 18% protein, 40% fat and 44% CHO. Subjects were given 5 min to eat the preload, and the rate was paced by having them take a spoonful every 30sec because rate of ingestion can influence glucose tolerance (Heine et al., 1983). Blood samples were taken 10min apart for baseline assessment prior to preload, and at 5 min intervals post preload up to + 30 and then at 15 min intervals to + 135 after the start of preload ingestion. After the IV lines were withdrawn, subjects were instructed to rest quietly for a few moments after which they were presented with a preweighed buffet lunch. The buffet consisted of 5OOg each of sliced ham, turkey, muenster cheese, and Swiss cheese; 8 slices of bread; an apple, pear, and banana; three bagels; 50 g cream cheese; 29.5 ml apple juice; two 35.48 ml sodas; two 236 ml containers of yogurt; and 20 chocolate-chip cookies (Rodin et al., 1988; Spitzer & Rodin, 1987). Subjects were told to eat until they were comfortably full because we needed to collect further data from them in a satiated state. Analytical methdds Plasma glucose concentrations were determined by the glucose oxidase method with an Astra glucose analyzer (Beckman Instruments, Fullerton, CA). This method excludes fructose in its analysis. Insulin was measured with radioimmunoassay by using a doubly labeled antibody technique (Morgan & Lazarow, 1963).
J. RODIN
216
The buffet meal was weighed to the nearest 0.1 g on an electronic balance (model 3000, Mettler Instruments, Highstown, NJ) before and after presentation to the subject. Nutritional analyses, which included determination of the caloric and macronutrient content of the test buffet each subject consumed, were conducted using the University of Massachusetts database (Pennington & Church, 1985; U.S. Department of Agriculture, 1976-1986). In the statistical analyses, we controlled for gender by standardizing the data separately for men and women to a mean of 0 and an SD of 1. Then means of the standardized scores for each preload were obtained, and the standardized data were evaluated by using paired t tests.
RESULTS
First, we compared the effects of the two preloads in which the sugars were the only source of carbohydrate. The means for calories and per cent fat, protein and carbohydrate eaten after each of the preloads are shown in Table 1. Subjects ate significantly fewer calories [(t-=3.67, p
Effects of Pure Sugar vs. Mixed Starch Food Intake JUDITH
Fructose
Loads on
RODIN
Department of Psychology,
Yale University
Using a within-subject design, we gave subjects three different 520-530 kcal preloads in the form of puddings in a randomized fashion at weekly intervals. The puddings contained either 5Og of fructose or glucose as the sole carbohydrate source in a protein and fat mixture, or 50 g fructose plus 15 g of starch. Food intake was assessed 2.25 h after the preload was completed. Blood was drawn throughout and assayed for concentrations of glucose and insulin. When the
preload contained fructose alone as the sole source of carbohydrate, subjects ate significantly fewer calories and less fat than when the preload contained glucose alone. When starch was added to the fructose preload, there was no significant reduction in calorie and fat intake. Effects on food intake paralleled the rise in plasma insulin levels produced by the different preloads. Implications for use of fructose as an adjunct to weight control efforts are discussed.
Studies have demonstrated that ingested fructose as opposed to glucose is absorbed more slowly (Durnin & Womersley, 1974; Schutz et al., 1984), generates greater thermogenesis (Jarrett et al., 1979; Schwarz et al., 1989; Wall et al., 1975), is metabolized directly by the liver (Schutz et al., 1984; Tappy et al., 1986; Wotecki et al., 1982) and produces a reduction in subsequent food intake (Rodin, 1990; Rodin et al., 1988; Shafer et al., 1987; Spitzer 8z Rodin, 1987). These striking differences are
important since fructose is found in natural foods such as fruit and honey and in many processed foods and drinks in the form of high fructose corn syrup (Hue, 198 1; Nilsson & Hultman, 1974; Segal et al., 1985), and therefore is widely consumed. The observation of different effects of these sugars on food intake leads quite naturally to speculations about the possible use of fructose as a weight-control intervention. For example, our studies have demonstrated that when subjects drink fructose-sweetened beverages (50 g in 500 ml) in the fasted state, they subsequently (from 40min to 2 h later) eat fewer calories and less fat than when they &ink isocaloric glucose-sweetened beverages. Fructose may actually suppress intake under certain conditions since intake after fructose consumption appears even lower than after drinking non-caloric, but volumetrically equivalent water loads. Supported by NIH grant RR125, Division of Research Resources to the Yale Clinical Research Center and by a grant from Xyrofin-Finnsugar. The author is grateful to Nancy Cummings for her assistance in conducting the study and to Heinz Schneider, Ph.D., of R&D Wander Ltd, Neunegg, Switzerland for his assistance in obtaining the Alburone and Liprocil for the test preloads. Address correspondence to: Dr. Rodin, Department of Psychology, Yale University Box 11A Yale Station, New Haven, CT 06520, U.S.A. 0195-6663/91/060213
+07 $03.00/O
0
1991 Academic
Press Limited
214
J. RODIN
While in our laboratory the relationship between fructose ingestion and lowered calorie and fat intake at the subsequent meal appears highly replicable, the effect seemed to us to be confined to fructose-sweetened beverages. When we presented 50 g of fructose as compared to glucose as part of a mixed nutrient breakfast meal including egg, bread, muffin, margarine and milk, it did not suppress subsequent intake (Rodin et al., 1988). With the mixed meal, plasma glucose and insulin levels were not significantly different in the glucose and fructose conditions. These findings stand in contrast to our observations with pure sugar liquid preloads, where plasma glucose and insulin were significantly elevated after the glucose drink, with little rise after the fructose drink. Many variables, such as the physical form of the food, amount of fiber and protein and fat content, influence glucose tolerance to a mixed meal (Collier et al., 1984; Crapo et al., 1982; Jenkins et al., 1978; Nuttall et al., 1983; Nuttall et al., 1984). The mixed meal also contained a large amount of complex carbohydrate, and this source of carbohydrate may overshadow the effect of the different sugars on carbohydrate metabolism. If the effects of fructose on food intake are confined only to beverages with no other nutritive constituents, its potential therapeutic application for weight control would be limited. Recently, however, in an effort to generalize from previous studies examining the thermic effects of pure glucose vs. pure fructose, Jequier and his colleagues assessed the magnitude of the thermic effect of a mixed meal containing either glucose or fructose as the sole source of carbohydrate (Schwarz et al., 1989). They observed a greater dietary-induced thermogenesis with the meal containing fructose, accompanied by a higher respiratory quotient, which suggested an enhanced rate of carbohydrate oxidation. In addition, higher plasma lactate concentration but lower plasma insulin and glucose responses were observed with the fructose meal when compared to the glucose meal. Based on these encouraging results, we considered the effects on subsequent food intake of a mixed nutrient preload with either glucose or fructose as the only carbohydrate source, compared to the same preload in which another source of carbohydrate was also provided. METHODS
Subjects Six subjects, three men and three women, participated in the study. The mean age of the subjects was 35.622.4 years and the mean percentage overweight was + 3 k3.6. Each subject was tested in three sessions, separated by 1 week. Subjects were assigned at random to a counterbalanced order of presentation of the three types of preload solutions-one at each session. Subjects were screened for symptoms of hypoglycemia or undiagnosed diabetes. Criteria for exclusion from the study included a fasting blood sugar >6*11 mmol/l, blood glucose concentrations 11.10 mmol/l during the study, and glucose 2 7.77 mmol/l at the conclusion of the study (Felig et al., 1981). No subjects were excluded on the basis of this evaluation. Procedure Subjects were recruited by a newspaper advertisement and were initially interviewed by telephone. A standardized screening procedure was followed to exclude
FRUCTOSE
AND FOOD INTAKE
215
potential subjects who reported medical problems, such as diabetes, heart disease, or hypertension, and people taking medications that might affect carbohydrate metabolism. Potential subjects were also excluded if they reported gaining or losing > 4.54 kg in the last 3 months or if they indicated that they were currently dieting. After the screening, subjects were scheduled for the study and were asked to eat no food and drink only water after 2100 hrs on the night preceding the study. They were also asked to eat 150-200 g of carbohydrate per day in the 3 days preceding the study. Bondy & Rosenberg (1980) suggested that only in circumstances where the patient was following a severe hypocaloric intake would this recommendation involve any change in the subject’s usual dietary habits. A 24-h food record was provided by each subject and was evaluated to establish whether subjects had eaten the recommended amount of carbohydrate. The mean amount eaten was 204 g. Participants were weighed upon admission as outpatients to the Clinical Research Center at Yale-New Haven Hospital. Blood pressure and pulse were measured and an intravenous (IV) line was inserted in the antecubital vein, which was kept open with a slow drip of isotonic saline. Subjects were next given one of three types of preloads. The composition of two of the preloads followed the test meal of Schwarz et al. with two changes: 1) 50g of fructose or glucose, rather than 75g, was used to make the amount of sugar comparable to all other preloads used in our laboratory; and 2) a pudding, rather than liquid meal, was constructed by allowing the mixtures to stand for several hours before serving. The formula contained 35g powder of Alburone (milk protein, Sopharga, France), 23g Liprocil (lipid containing 80% MCT, Sopharga, France), 50 g glucose or fructose, 7 g decaffeinated coffee, and 200 ml deionized mineral water. The pudding was 520 kcal with the glucose and 530 kcal with the fructose, and was 24% protein, 41% fat and 35% CHO. In the third preload, complex carbohydrate in the form of 15 g of cornstarch was added, and 14g of Alburone was deleted. The simple sugar was 50 g of fructose. The pudding was 530 kcal with 18% protein, 40% fat and 44% CHO. Subjects were given 5 min to eat the preload, and the rate was paced by having them take a spoonful every 30sec because rate of ingestion can influence glucose tolerance (Heine et al., 1983). Blood samples were taken 10min apart for baseline assessment prior to preload, and at 5 min intervals post preload up to + 30 and then at 15 min intervals to + 135 after the start of preload ingestion. After the IV lines were withdrawn, subjects were instructed to rest quietly for a few moments after which they were presented with a preweighed buffet lunch. The buffet consisted of 5OOg each of sliced ham, turkey, muenster cheese, and Swiss cheese; 8 slices of bread; an apple, pear, and banana; three bagels; 50 g cream cheese; 29.5 ml apple juice; two 35.48 ml sodas; two 236 ml containers of yogurt; and 20 chocolate-chip cookies (Rodin et al., 1988; Spitzer & Rodin, 1987). Subjects were told to eat until they were comfortably full because we needed to collect further data from them in a satiated state. Analytical methdds Plasma glucose concentrations were determined by the glucose oxidase method with an Astra glucose analyzer (Beckman Instruments, Fullerton, CA). This method excludes fructose in its analysis. Insulin was measured with radioimmunoassay by using a doubly labeled antibody technique (Morgan & Lazarow, 1963).
J. RODIN
216
The buffet meal was weighed to the nearest 0.1 g on an electronic balance (model 3000, Mettler Instruments, Highstown, NJ) before and after presentation to the subject. Nutritional analyses, which included determination of the caloric and macronutrient content of the test buffet each subject consumed, were conducted using the University of Massachusetts database (Pennington & Church, 1985; U.S. Department of Agriculture, 1976-1986). In the statistical analyses, we controlled for gender by standardizing the data separately for men and women to a mean of 0 and an SD of 1. Then means of the standardized scores for each preload were obtained, and the standardized data were evaluated by using paired t tests.
RESULTS
First, we compared the effects of the two preloads in which the sugars were the only source of carbohydrate. The means for calories and per cent fat, protein and carbohydrate eaten after each of the preloads are shown in Table 1. Subjects ate significantly fewer calories [(t-=3.67, p
