Nutrient balance and energy expenditure during ad libitum feeding of high-fat and high-carbohydrate diets in humans13

Transcription

1 Nutrient balance and energy expenditure during ad libitum feeding of high-fat and high-carbohydrate diets in humans13 Cecilia D Thomas, John C Peters, George W Reed, Naji N Abumrad, Ming Sun, and James Hill ABSTRACT To study the influence ofdiet composition on regulation of body weight, we fed 21 weight-stable subjects (1 1 lean, 1 obese) high-carbohydrate (HC) and high-fat (HF) diets for 1 wk each. Although diet composition was fixed, total energy intake was unrestricted. Subjects had a higher energy intake on the HF ( ± 27 kj/d) than on the HC (1 672 ± 2617 kj/d) diet (P <.5), but energy expenditure was not different between diets. On day 7 of the HC diet, carbohydrate (CHO) oxidation was significantly related to CHO intake with the slope ofthe regression line.99, suggesting that overall CHO balance was near ero. However, the slope of the regression line was greater for obese than for lean subjects. On day 7 ofthe HF diet, fat oxidation was significantly related to fat intake but the slope ofthe line was.5, suggesting that overall fat balance was positive. However, this relationship was due entirely to lean subjects, with obese subjects showing no relationship between fat intake and oxidation. Am J Clin Nuir 1992;55:93-2. KEY WORDS Obesity, thermogenesis, diet composition Introduction Stability of body weight and body composition requires that over time, energy intake equals energy expenditure and also that intakes of protein, carbohydrate, and fat equal the oxidation of each (1, 2). Another way of stating this is that the respiratory quotient (RQ), which is the carbon dioxide produced divided by the oxygen consumed, must equal the food quotient (FQ), which is the carbon dioxide produced divided by the oxygen consumed if the diet were oxidied. If these conditions are not met, changes will occur in the body stores of protein, carbohydrate, and fat, which affect overall body weight and composition. Many studies have examined the regulation of energy intake and/or energy expenditure to understand how the two are balanced or, in the case ofdeveloping obesity, unbalanced (3). Few studies have examined the regulation of substrate oxidation to understand the extent to which the composition offuel oxidied is adjusted to the composition of energy ingested. If such adjustments do not occur, imbalances in one or more nutrients may lead to accumulation or depletion of stored energy and thus changes in body weight. Flatt (1, 2), has been a pioneer in this area and has proposed that difficulty in achieving fat balance on typical mixed diets may play a key role in the development of obesity. According to Flatt, this is because fluctuations in dietary protein and carbohydrate are compensated for by immediate changes in protein and carbohydrate oxidation, whereas changes in fat intake are not balanced by rapid responses in fat oxidation. The result is that excess dietary fat is almost entirely stored in the body whereas excess protein and carbohydrate are not. In addition, Flatt proposes that food intake is influenced by events related to carbohydrate utiliation so that a high-fat (HF) diet (low in carbohydrate) promotes increased total food intake. In support offlatt s model is the finding that HF diets promote obesity in rodents (, 5). In studies ofrats given ad libitum access to HF diets, we (6) and others (5) found that although most rats readily become obese, some avoid obesity. We further found that rats that avoided obesity on an HF diet had a lower daily RQ than did those that did become obese (6), suggesting that obesity-resistant rats increased fat oxidation in response to increased fat intake more than did the latter group. Thus in rats, differences in the ability to increase fat oxidation in response to high dietary fat intake may determine susceptibility to fat gain on an HF diet. We proposed that there are individual differences in the extent to which a high intake of dietary fat results in increased fat oxidation and body-fat gain, and thus there are differences in the susceptibility of individuals to the effects of diet composition in the development of obesity. The purpose of the present study was to determine whether alterations in the fat-carbohydrate ratio ofthe diet produce measurable changes in total energy expenditure, the composition of fuel oxidied by the body, and the total amount of energy consumed. Flatt s model ( 1, 2) would predict an increase in carbohydrate oxidation when subjects are given a high-carbohydrate, low-fat diet, but little or no increase in fat oxidation when subjects are given a high-fat, low-carbohydrate diet. This model would also predict a higher total food intake on an HF diet than on a high-carbohydrate (HC) diet. Additionally, our intent was to determine whether lean subjects differ from obese subjects either in their ability to increase fat oxidation when eating the highfat, low-carbohydrate diet or in the effect of diet composition on energy intake. To accomplish these aims, we fixed diet com- 1 From the Clinical Nutrition Research Center ofthe Departments of Pediatrics, Preventive Medicine, and Surgery, Vanderbilt University, Nashville, TN; and The Procter and Gamble Co, Cincinnati. 2 Supported by NIH grants DK259, DK26657, and RR95. 3 Address reprint requests to JO Hill, Clinical Nutrition Research Center, D-13O MCN, Department ofpediatrics, Vanderbilt University, Nashville, TN Received August 5, Accepted for publication December 12, Downloaded from ajcn.nutrition.org at PENNSYLVANIA STATE UNIV PATERNO LIBRARY on February 21, Am J C/in Nuir l992;55:93-2. Printed in USA American Society for Clinical Nutrition

2 NUTRIENT BALANCE IN HUMANS 935 position (yielding a constant FQ) and allowed subjects to eat as much or as little of the diet as they wanted. Total energy cxpenditure and substrate oxidation were determined from data obtained with a whole-room respiration chamber. Methods Subjects Twenty-one adult subjects were recruited for this study, which was approved by the Vanderbilt University Committee for the Protection of Human Subjects. Subjects were recruited to fit into one of the following classifications: 1) nonobese (body fat < 25% for men; < 3% for women) with no obese first-degree relatives (determined from the subject s recall); or 2) obese (body fat > 25% for men; > 3% for women) with at least two obese first-degree relatives (determined by the subject s recall). Subjects were eliminated if they reported a personal or family history of diabetes or other endocrine disease. An attempt was made to recruit an equal number ofmen and women in each group. The characteristics ofthe 21 volunteers (1 obese, 1 1 nonobese) that met these criteria are shown in Table I. Procedures All subjects were studied during a 2-wk prestudy period, during a l-wk baseline period, and during 2 experimental wks of diet manipulations. Subjects were inpatients during their stays in the whole-room calorimeter and on all other days they received all oftheir food from the Clinical Research Center (CRC). The two experimental weeks were separated by a 1-mo washout period. During the prestudy period subjects recorded all food eaten, using diet diaries, which were analyed to determine the amount and composition ofenergy ingested. During the baseline period subjects were fed the amount and composition offood that they reported eating during the prestudy period. During the baseline week, subjects received all oftheir food from the CRC. However, adjustments in total food provided were made for some subjects who lost weight during the first few days of the baseline period. All subjects were studied in our whole-room calorimeter during days I and 7 of the baseline period. After the baseline period subjects were randomly assigned to receive either the HC or HF diet for 1 wk. This was followed by a 1-mo washout period during which the subject ate ad libitum at home, and then subjects were fed the remaining experimental diet for 1 wk. During both experimental weeks, subjects received TABLE 1 Characteristics of subjects all of their food from the Vanderbilt CRC. All subjects were studied in our whole-room calorimeter on day 7 of each of the two experimental weeks. Diets Diets were individualied for each subject to allow us to assess energy expenditure and nutrient balance at their usual diet composition (ie, during baseline) and with a relatively similar increase or decrease in the carbohydrate-fat ratio of the diet. This was accomplished as follows. Because we could not ensure that the usual diets ofall subjects were ofthe same composition, we tried to change diet composition by a similar magnitude for all subjects during the experimental weeks. We estimated the average RQ ofthe habitual diet on the basis offood-intake records, assuming that subjects were maintaining near-ero energy balance. To formulate the HF diet, we subtracted. from the measured baseline RQ and provided a diet with an FQ (RQ of the diet) equal to this value. For the HC diet we added. to the baseline RQ and provided a diet with an FQ equal to this value. Thus we produced a change in diet composition that was relatively uniform for each subject. Because altering voluntary food intake represents one way by which subjects could adjust to changes in diet composition, the goal of this study was to fix diet composition (ie, FQ), but not to restrict the amount of food eaten by the subjects. This was accomplished as follows. Each subject s maintenance energy requirement was assumed to be 1. X measured resting metabolic (RMR) rate. An equivalent amount of energy was provided to the subject each day in the form of a diet with a fixed FQ. Additionally, subjects could request an unlimited number of food modules, all of 836 Id and all having the target FQ at any time ofthe day. The FQ ofthe food modules was also individualied for each subject. This protocol was in effect during the time spent in the calorimeter as well as during the outpatient portion of the study. Thus, within the limits of this procedure subjects were allowed to eat ad libitum a diet with a constant FQ. Subjects received all of their food from the Vanderbilt CRC dietitians during the baseline week and the two experimental weeks. CRC dietitians weighed all food provided and all food not eaten. The composition of foods used to prepare diets was determined from food tables (7). The digestible energy contents of protein, carbohydrate, and fat were also estimated from the food tables. The average digestible energy ofthe protein used in this study was estimated to be 85% of gross protein energy. Digestible intakes of carbohydrate and fat were estimated to be 95% of the gross energy of these food components. This value Downloaded from ajcn.nutrition.org at PENNSYLVANIA STATE UNIV PATERNO LIBRARY on February 21, 213 Age Height Weight BMI t Percent fat Insulin sensitivity y cm kg % X1 Lean men (n = 6) 2 ± ± ±. 2 ± 1 21 ± ±.53 Lean women (n = 5) 27 ± ± ± ± I 26 ± ±.68 Obesemen(n=5) 27± ± ±. 31 ± 1 35± ±. Obesewomen(n=5) 28± ± ± ±2 3±3 1.56±.69 * I ± SE. t Body mass index [in kg (wt)/m2 (ht)]. t SI value from Bergman s minimal model.

3 936 THOMAS ET AL was obtained by averaging the digestible energy of representative carbohydrate and fat sources used in the present study. Body composition Body composition was determined once during baseline for each subject from measurements of body density estimated by underwater weighing. Body weights in air and underwater were measured to the nearest 25 g by using Detecto platform (Webb City, MO) and Chatillon spring (Kew Gardens, NY) scales, respectively. Residual lung volume was determined (simultaneously with underwater weighing) with a closed-circuit, nitrogen-dilution method (8). Nitrogen concentration during rebreathing was measured with a Med-Science 55-D Nitralier (St Louis). Percent body fat was estimated from body density by using the revised equation oflohman et al (9). We performed many studies in which body composition, determined by using this technique, is determined in the same subjects on many different occasions. From these studies we estimated our absolute error as 2-5%. Energy expenditure Total 2-h energy expenditure was measured with a wholeroom indirect calorimeter, which was described previously (1). While in the calorimeter, the subjects were free to move around but were not provided with exercise equipment or given specific instructions to exercise. The calorimeter is located within the CRC and is connected via an intercom to the nursing station. Oxygen consumption and carbon dioxide production are determined from the flow rate and the differences in gas concentrations between entering and exiting air. Values are corrected for temperature, barometric pressure, and humidity. Energy expenditure is calculated from the oxygen consumption and RQ. The operation of the chamber is controlled by a personal computer by using a software program written in Turbo C. The program was based on calculations described by Jequier et al (1 1). Values for all indices were averaged over 3-mm intervals and recorded in a data file. In addition to total 2-h energy expenditure, the individual components of energy expenditure were estimated during each 2-h stay in the whole-room calorimeter. Energy expenditure due to activity or movement (EEAcr) was estimated with the assistance ofa Doppler radar system (12), installed in the wholeroom calorimeter. The instrument records relative activity, which is significantly correlated with metabolic rate. It is possible to calculate the caloric cost of activity (the slope of the regression line describing the relationship between activity and metabolic rate) and total EEA,. (cost of activity times amount of activity by using linear regression. The metabolic rate at ero activity would represent RMR and the thermic effect of food (TEF). After leaving the calorimeter at 7 after a 23-h stay, subjects were moved to an adjacent room and allowed to rest for 5 mm, and RMR was measured for 3 mm with a ventilated-hood systern (Sensormedics 29 Oxygen Uptake System, Anaheim, CA). In addition to RMR, we measured sleeping metabolic rate (SMR), which we defined as the average metabolic rate measured during sleep. Periods of sleep were determined from an activity diary maintained by the subject while in the whole-room calorimeter as well as from measures of activity obtained from the radar detector. SMR was taken to be the average of all 3-mm periods when the subject reported having slept and when sleep could be inferred from the subject s diary and when activity by radar was < 1%. Substrate-oxidation rates and daily nutrient balance Daily rates ofoxidation ofprotein, carbohydrate, and fat were determined for each 2-h stay in the whole-room calorimeter. Protein oxidation was determined from 2-h urinary nitrogen excretion (measured with the Kjeldahl technique), with correction for any change in the blood-urea-nitrogen pool (13); carbohydrate and fat oxidation were determined from total energy expenditure and the nonprotein RQ (1). Nutrient balance was calculated as the difference between intake and oxidation of each nutrient over 2 h. Insulin sensitivity We obtained a measure of each subject s insulin sensitivity, using Bergman s minimal model program with a modified frequent sampling intravenous glucose tolerance test (I 5). This procedure was performed on all subjects during the baseline period. Three baseline blood samples were drawn and then.3 g 5% dextrose/kg body wt was infused over 2 mm. Additional blood samples were taken at 3, 6, 1, 1, 19, 23, 25, 26, 29, 3, 32, 35, 5, 7, 1, 1, and 18 mm. Tolbutamide (5 mg/kg) was given to subjects 2 mm after glucose infusion. Blood samples were analyed for glucose (autoanalyer) and insulin (16). SI (equivalent to insulin sensitivity) was calculated by using a minimal model software program obtained from Bergman (15). Statistical methods All dependent variables were analyed by using repeatedmeasures analysis of variance with subject phenotype (lean vs obese) and gender (male vs female) as the between-groups factors and responses to the three experimental diet conditions as the within-subjects factor. Post hoc tests were performed when appropriate by using the Newman-Keuls method (17). Linearregression analysis was used to describe the relationship between selected variables. The 95% confidence limits were calculated for the slopes of each regression line to determine if they were different from and from each other. Results Body weight Individual body weights fluctuated (by <.25 kg) throughout the study, but we did not observe any systematic increase or decrease in the body weight of any subject. Additionally, there was not a significant change in the average body weight of the group as a whole over the course of the study. Average dailyfood intake Table 2 shows the average daily intake for each group throughout the study. The first column in this table shows the average intake during a 1. period preceding the study and was based on self reports of food intake by the subjects. The remaining columns reflect average food intake as determined by the CRC dietitians. During the baseline period, the average diet composition was 13.8% protein, 8.3% carbohydrate, and 37.9% fat. Obese subjects tended to have a slightly higher fat intake as percent of calories than did lean subjects. During the HC diet the average diet composition was 12.7% protein, 6 1.6% Downloaded from ajcn.nutrition.org at PENNSYLVANIA STATE UNIV PATERNO LIBRARY on February 21, 213

4 NUTRIENT BALANCE IN HUMANS 937 TABLE 2 Daily energy intake Diet diariest HC wk HF wk Leanmen(n=6) 11718± ± ± ±1216 Lean women (n = 5) ± ± ± ± 1 32 Obesemen(n=5) 11873± ± ± ±891 Obese women (n = 5) ± ± ± ± 976 * I ± SE. HC, high carbohydrate; HF, high fat. kj/d t Measured over 1 d. carbohydrate, and 25.7% fat. During the HF diet the average diet composition was 12.6% protein, 35.% carbohydrate, and 52.% fat. Total energy intake was higher on the HF diet (by 69 kj/d) than on the HC diet (P <.5). Total energy ingested during the three measured periods was higher for men than for women (P <.1) and was higher for obese subjects than forlean subjects (P <.1). Energy expenditure Table 3 shows the average daily rates of energy expenditure measured at baseline and on day 7 of each feeding period. Changes in diet composition did not produce measurable changes in energy expenditure. Energy expenditure was higher in males than in females (P <.1) and higher in obese subjects than in lean subjects (P <.1). Both differences would be expected based on differences in fat-free mass between subject groups. In addition, neither SMR nor RMR was influenced by the composition ofthe diet eaten (data not shown). Similarly, there were no significant differences due to diet composition in the estimated energy expended in physical activity while in the calorimeter (Table ). Table 5 shows the difference between average energy consumed during each period during the study and measured energy expenditure in the room calorimeter. If the energy expenditure measured in the room calorimeter reflects sedentary energy expenditure, the difference between these values might reflect 1) the difference in the energy expended in spontaneous activity between a sedentary day and a usual day; and 2) the degree of positive energy balance created by the change in diet cornposition. We will refer to the value as positive energy balance (PEB). Without independent measures ofspontaneous physical TABLE 3 Twenty-four-hour energy expenditure in the calorimeter* activity, it is impossible to determine whether changes in PEB simply reflect changes in spontaneous physical activity. However, if the assumptions above are reasonable, this allows for some interesting speculation. First, during the 2-wk period of keeping diet diaries, subjects should be near energy balance and, if so, PEB would most likely reflect energy expended in spontaneous physical activity. Ifthis assumption is correct and ifdiet diaries accurately reflect food intake during this period, lean males and females expended significantly (P <.1) more energy during a usual day than during a sedentary day in the room calorimeter. However, the value for PEB in obese subjects was not different from ero, which suggests either that activity during a usual day was not different than activity seen in the room calorimeter, an unlikely event, or that diet diaries underestimated actual food intake in these subjects. For the first day of the baseline period there was a negative correlation between PEB and percent body fat that was nearly statistically significant(r = -.2, P =.575). We found that PEB was greater on the HF diet than on the HC diet, and this difference (673 kj/d) also approached statistical significance (P <.6). Nutrient oxidation in relation to nutrient intake Table 6 shows values for the 2-h RQ during each stay in the whole-room calorimeter. The RQ was significantly higher on day 7 of the HC week than it was on day 7 of the HF week (P <.1) for all subjects. The RQ measured during sleep is shown in Table 7 for the various groups during baseline and diet treatments. Obese subjects had a lower sleeping RQ on the HF than on the HC diet (P <.3) but there was no difference for lean subjects. Figures 1-3 show intakes ofprotein, carbohydrate, and fat in relation to oxidation of each on the first day of the baseline Downloaded from ajcn.nutrition.org at PENNSYLVANIA STATE UNIV PATERNO LIBRARY on February 21, 213 (day 1) HC wk HF wk Lean men (n = 6) ± ± ± ± 352 Lean women (n = 5) 7 5 ± ± ± ± 7 Obesemen(n=5) 11678± ± ± ±3 Obesewomen(n=5) 887± ± ±526 81±S2 * I ± SE. HC, high carbohydrate; HF, high fat. kj/d

5 938 THOMAS ET AL TABLE Energy expended in physical activity in room calorimeter (day 1) HC wk HF wk Leanmen(n=6) 1316± ± ± ±229 Lean women (n = 5) 896 ± ± ± ± 218 Obese men (n = 5) 158 ± ± ± ± 163 Obese women (n = 5) 1521 ± ± ± ± 33 * : SE. HC, high carbohydrate; HF, high fat. kj/d period and on day 7 of the HC and HF treatment periods. For each plot we show the regression line for all 2 1 subjects (solid line) as well as the regression lines for lean and obese subjects separately. On day I of baseline (Fig 1), protein intake and oxidation were correlated (r =.5, P <.25) when all subjects were considered. This relationship did not differ between lean and obese subjects and the slopes of the regression lines were similar and significantly different from. Carbohydrate intake and oxidation were also positively correlated when all subjects were considered (r =.6, P <. 1). The correlations for lean and obese subjects were also high, although statistically significant only with lean subjects. There was not a significant relationship between fat intake and fat oxidation in all subjects (r =.9, NS) or in either lean or obese subjects when considered separately. The slopes ofthe regression lines describing fat oxidation in relation to fat intake were not different from. Figure 2 shows the results for day 7 ofthe HC feeding. Protein intake and oxidation were positively correlated for all subjects (r =.68, P <.1) and also for lean and obese subjects when considered separately. Slopes ofall ofthese regression lines were significantly different from. Carbohydrate intake and oxidation were also significantly correlated for all subjects (r.7 1, P <.1). The equation for the regression line was y (CHO oxidation) =.99x + 2. Carbohydrate intake and oxidation were also significantly correlated for lean and obese subjects when considered separately. Slopes ofall regression lines were different from. The slope of the regression line for obese subjects (y = l.2lx - 12) was significantly (P <.5) higher than that for lean subjects (y =.9x - 6). There was no significant relationship between fat intake and oxidation for all subjects (r = -.3) or for either lean or obese subjects. The slopes of these regression lines were not different from. On day 7 of the HC TABLE 5 diet, note that obese subjects tended to oxidie proportionally more carbohydrate and less fat than did lean subjects over the range of intakes presented in Figure 2. Figure 3 shows the results for day 7 ofthe HF feeding. Protein intake and oxidation were significantly related for all subjects (r =.72, P <.1) and for lean and obese subjects separately. There was not a significant relationship between carbohydrate intake and oxidation either for all subjects together (r =.22) or for lean or obese subjects considered separately. There was a significant relationship between fat intake and fat oxidation (r =.57, P <.1) for all subjects. The regression line was y(fat oxidation) =.5x + 6). However the relationship between intake and oxidation of fat was entirely explained by lean subjects, with obese subjects not demonstrating a significant correlation between fat intake and oxidation. The slope of the regression line for lean subjects (y =.65x + 17) was significantly different from but the slope of the regression line for obese subjects (y =.2x + 12) was not. Insulin sensitivity Insulin sensitivity (estimated from S) was slightly but nonsignificantly (P <.8) higher in lean than in obese subjects (Table 1). We were unable to show any significant relationship between S and any measure of substrate balance on either the HC or HF diets. Discussion The results of this study provide support for the basic tenet of Flatt s model (1, 2), that the ability ofan individual to match fat intake and oxidation plays a key role in body-weight regulation. We extended this model to suggest that there are mdi- Downloaded from ajcn.nutrition.org at PENNSYLVANIA STATE UNIV PATERNO LIBRARY on February 21, 213 Difference between average energy in take and energy expended in the room calorimete r (E1 - (day 1) HC wk HF wk kj/d Lean men (n = 6) Leanwomen(n = 5) Obese men (n = 5) Obese women (n = 5) 1956 ± ± ± 92-9 ± ± ± ± ± ± ± ± ± ± ± ± ± 656 * Determined as average daily intake during each specified diet; i ± SE. HC, high carbohydrate; HF, high fat.

6 TABLE 6 Average 2-h respiratory quotient in the calorimeter* NUTRIENT BALANCE IN HUMANS 939 HC wk HF wk (day 1) Lean men (n = 6).855 ± ±.1.97 ±.7.82 ±. 15 Lean women (n = 5).869 ± ±.1.91 ± ±.13 Obese men (n = 5).875 ±..853 ± ± ±.9 Obese women (n = 5).866 ± ± ±..83 ±.13 * I ± SE. HC, high carbohydrate; HF, high fat. vidual differences in the ability to achieve fat balance when humans consume an HF diet and that lean subjects may have a better ability to achieve fat balance under these conditions than obese subjects. Additionally, on day 7 of the HC diet, obese subjects showed a proportionally higher oxidation of carbohydrate and lower oxidation of fat than did lean subjects. We hypothesie that differences in the proportion of carbohydrate vs fat oxidied under some dietary conditions may be important in determining who is susceptible to dietary-induced obesity. We further speculate that a poor ability to match fat oxidation to fat intake may play a role in the development of obesity in obese subjects and a good ability to match fat oxidation to fat intake may play a role in the avoidance ofobesity in lean subjects. Flatt s model (1, 2), if applied to humans, would suggest that alterations in carbohydrate intake are quickly adjusted for by changes in carbohydrate oxidation but that changes in fat intake are poorly compensated for by appropriate alterations in fat oxidation. Overall, our results provide four lines of support for this notion. First, when subjects were studied in the room calorimeter after 2 wk ofeating their usual diets, both protein and carbohydrate oxidation were correlated with intake ofeach; the slope ofthe line describing this relationship was near unity. The slope of the line plotting fat intake and oxidation did not approach unity, even in lean individuals. It is important to point out that all subjects were in slight positive energy balance in the room calorimeter because of low physical activity. Second, all subjects in this study increased carbohydrate oxidation when given the HC diet, the increase in oxidation being roughly equivalent to the increase in carbohydrate intake. The slope of the regression line was significantly greater for obese than for lean subjects (P <.5), but the y intercepts did not differ between groups. Third, although there was an increase in fat oxidation in lean subjects after 7 d of feeding the HF diet, this increase in oxidation was much less than the increase in fat intake, resulting in a state of positive fat balance. Finally, the difference between energy ingested and energy expended in the whole-room calorimeter was greater during HF feeding than during HC feeding. Unless voluntary physical activity increased during HF as compared with HC feeding, this is further mdication that there was greater positive energy balance on the HF diet, suggesting that the HF diet was more obesity producing than was the HC diet. These results provide additional evidence that HF diets promote obesity in humans. Both lean and obese subjects were in positive fat balance during feeding ofthe HF diet. Additionally, HF diets also appeared to disrupt the usual relationship between carbohydrate oxidation and carbohydrate intake in both lean and obese subjects. This leaves obese individuals particularly prone to body-weight (and fat) gain because rates of intake of both carbohydrate and fat may exceed rates ofoxidation of each. Lean subjects may have a better (or quicker) ability than obese subjects to increase total fat oxidation in response to the increased fat intake. The greater insulin sensitivity of lean subjects may be a key to the ability to adjust nutrient oxidation to nutrient intake. For example, at least under some conditions studied in this study, obese subjects differed from lean subjects in that they tended to oxidie proportionally more carbohydrate and less fat than did the latter group. Reduced insulin sensitivity in obese subjects would require increased insulin concentrations in order to sustain a high degree ofcarbohydrate oxidation (18). Because insulin has antilipolytic effects (19), increased circulating insulin concentrations may limit lipolysis, an effect that may also limit total fat oxidation (2). If one projects these short-term results over a longer time period, both lean and obese subjects should gain body fat when eating an HF diet, but this should be greater for obese subjects. It is not possible to provide any reasonable estimate of how much difference in fat storage would occur over time between lean and obese subjects because it could be that the difference in response to the HF diet between lean and obese subjects only Downloaded from ajcn.nutrition.org at PENNSYLVANIA STATE UNIV PATERNO LIBRARY on February 21, 213 TABLE 7 Respiratory quotient during sleep (day 1) HC wk HF wk Lean men (n = 6).827 ± ± ± ±.36 Lean women (n = 5).826 ± ± ±.22.8 ±.22 Obese men (n = 5).816 ± ± ± ±. 12 Obese women (n = 5).79 ± ± ± ±.9 * I ± SE. HC, high carbohydrate; HF, high fat.

7 9 THOMAS ET AL x Co C) x I _ 2 CO 2? U. BASELINE DA Y 1 This is a lesser difference than that reported by others, who re- ported average differences between intake on HF and HC diets Lean, r =.8. n.s. of799 (25), (26), and (27) kj/d. This lesser difference. Obese; r =.62. p <.5 #{19} may be due to the design of the present study where the com #{19}L#{19}- eat subject total position However, a energy snack onofthese additional containing intake HF results diet on food an do more whohf provide available desired fathanthanfurther ona was pure an in fixed. evidence HCfat thediet. snack diet For for as example, could a greater whole. not #{19}L#{19}.:#{19} a We did not find any differences in total daily energy expenditure due to diet condition. This finding agrees with previous work in humans (1, 28). However, this does not necessarily PROTEIN INTAKE (g/day) mean that diet composition does not affect energy expenditure, CHO INTAKE (g/day) Lean; r= -.1, n.s.. Obese; r= -.9, n.s.. o #{176} FAT INTAKE (g/day) FIG I. Relationship between intake (x axis) and oxidation (y axis) of protein, carbohydrate (CHO), and fat for all subjects on the first day of the baseline period. (-), regression line for all 21 subjects; (---), regression line for lean subjects; and (.... ), regression line for obese subjects. reflects a temporal difference in the rate at which the body adjusts fat oxidation to match fat intake. Although it will be important to determine whether this is the case, this situation would still tend to promote obesity more in obese than in lean subjects. Other investigators showed that adding dietary fat to a mixed diet does not produce an increase in fat oxidation over a 6-h (2 1, 22) or 2-h (22, 23) period, whereas adding carbohydrate does produce an increase in carbohydrate oxidation as early as 1 d later (2). Our results suggest that switching subjects to an HF diet can produce changes in fat oxidation after 7 d but only in lean subjects. Flatt s model also predicts that food intake is linked to a mechanism related to carbohydrate utiliation, so that total energy intake should be higher on HF diets than on HC diets (unless the regulated amount ofglycogen or ofglucose utiliation drops). Subjects are kj/d more when given the HF vs HC diet. it may be that the difference in energy expenditure between diet conditions would be less than can be reliably measured with.2oo. Lean;r=.79,p<.1 2OO. >. 21 x I1_ 15 5 HIGH CHO DA Y #{19} PROTEIN INTAKE (g/day) 6 8 CHO INTAKE (g/day) Lean; r=.6, n.s. #{19}Obese; r=-.8, n.s. #{19} #{19}#{19} a #{19} a o.#{19} I I FAT INTAKE (g/day) FIG 2. Relationship between intake (x) and oxidation (y) of protein, carbohydrate (CHO), and fat for all subjects on day 7 of the highcarbohydrate feeding period. (-), regression line for all 21 subjects; (-- -), regression line for lean subjects; and (.. ), regression line for.. obese subjects Downloaded from ajcn.nutrition.org at PENNSYLVANIA STATE UNIV PATERNO LIBRARY on February 21, 213

8 NUTRIENT BALANCE IN HUMANS 91 _ 2 Co C) x e _ 5 >.? 3-2. High Fat Day PROTEIN INTAKE (g/day) -) Leanr-. Obese; r=.1,n.s Co u Lean;r=.78,p<.1 CHO INTAKE (g/day). #{19} Obese; r=o.2, n.s. ; FAT INTAKE (g/day) FIG 3. Relationship between intake (x axis) and oxidation (y) of protein, carbohydrate (CHO), and fat for all subjects on day 7 of the highfat feeding period. (-), regression line for all 2 1 subjects; (---), regression line for lean subjects; and (.... ), regression line for obese subjects.. current techniques. Energy expenditure can be subdivided into various components, such as SMR, RMR, EEACT, and TEF. The component that would be expected to differ because of diet condition would be TEF, because the energy costs of utiliing fat and carbohydrate are different. However, the expected difference would be small and would not likely be reliably measured with current methodologies. We found no significant differences in SMR, RMR, EEAC-I., or TEF between diet conditions. We believe that these results provide support for use of the nutrient-balance technique over a relatively short period of time to assess the role of the amount and composition of food intake in affecting body weight. These results in combination with our previous report (1) suggest that measurable changes in substrate oxidation occur within 7 d of altering the composition of the habitual diet. By measuring nutrient balance (intake - oxidation), predictions can be made about longer-term changes in body weight and body composition. Although, it will be necessary to verify these with long-term studies, this technique can contribute to the theoretical and practical basis for conducting such studies. We were also encouraged with the use of food modules that allowed total intake to vary while diet composition remained fixed. Although this technique clearly does not provide true ad libitum intake, the subjects did make use of the modules and there were differences in food intake across the conditions of the study. One objective of the present study was to assess whether the 2 l-wk baseline period would allow us to demonstrate an equality ofthe FQ and RQ. Flatt clearly pointed out that this is a necessary condition of energy balance and stability of body composition. - We did this by assessing usual food intake from diet diaries and then feeding subjects what they reported eating. The hope was that at the end of the week, we could demonstrate ero balance for all nutrients. This was not the case and it may be useful to consider why. First, the theoretical condition of FQ = RQ requires not only that nutrient intake equal nutrient oxidation but also that the amount of energy ingested be exactly enough to meet energy requirements, because overfeeding increases RQ in relation to FQ and underfeeding reduces RQ in relation to FQ. Additionally, the amount of physical activity performed during the RQ measurement must be exactly equal to usual physical activity. This is obviously not the case inside the whole-room calorimeter. Therefore, we conclude that the equality offq and RQ, although a necessary condition for energy and substrate balance, is extremely difficult to demonstrate in human subjects over the short term (2 h). Although equality of FQ and RQ must occur over the long term, it is not clear that this would always be apparent over 2-h periods. It is possible that mdividuals differ in how quickly the equilibrium between the FQ and RQ is achieved after disruption ofdiet and/or exercise, and that such time-course differences may be important in determining susceptibility to dietary obesity. Our results suggest that HF diets are more obesity producing than are HC diets. This is because there was a greater total energy intake on HF than on HC diets and because humans have a lesser ability to increase fat oxidation in response to increased fat intake than to increase carbohydrate oxidation in response to increased carbohydrate intake. Additionally, we found some evidence to suggest that obese humans may be more susceptible to dietary obesity than are lean humans. When the diet contained a high amount of carbohydrate, obese subjects oxidied proportionally more carbohydrate and less fat than did lean subjects. Additionally, on day 7 of the HF diets, lean subjects demonstrated a significant positive relationship between fat intake and oxidation whereas obese subjects did not. These differences are interesting and could contribute to resistance to weight gain on an HF diet in lean subjects. We thank the staff of the Vanderbilt Clinical Research Center, particularly Patricia Heller, Donna Rice, for assistance in conducting this study. References 1. Flatt J. Importance of nutrient balance in body weight regulation. Diabetes 1988;: Flatt JP. The difference in the storage capacities for carbohydrate and for fat, and its implications in the regulation of body weight. Ann NY Acad Sci 1987;99:1-23. Downloaded from ajcn.nutrition.org at PENNSYLVANIA STATE UNIV PATERNO LIBRARY on February 21, 213

9 92 THOMAS ET AL 3. Sims EAH, Danforth E Jr. Expenditure and storage of energy in man. J Clin Invest l987;79: Hill JO, Dorton J, Sykes MN, DiGirolamo M. Duration of dietary obesity in rats influences resistance to obesity reversal. lnt J Obes l989;13: Levin BE, Triscari J, Hogan 5, Sullivan AC. Resistance to dietinduced obesity: food intake, pancreatic sympathetic tone, and insulin. Am J Physiol 1987;252:R Chang 5, Graham B, Yakubu F, Lin D, Peters JC, Hill JO. Metabolic differences between obesity-prone and obesity-resistant rats. Am J Physiol 199;259:Rl Watt BK, Merrill AL. Agricultural handbook no. 8. Washington, DC: US Government Printing Office, Goldman RF, Buskirk ER. Body volume measurement by underwater weighing description ofa method. In: Broek J, ed. Techniques for measuring body composition. Washington, DC: National Academy ofsciences, 196 1: Lohman TG, Slaughter MH, Boileau RA, Bunt J, Lussier L. Bone mineral measurements and their relation to body density in children, youth and adults. Hum Biol 198;56: Hill JO, Peters JC, Reed GW, Schlundt DO, Sharp T, Greene HL. Nutrient balance in man: effects of diet composition. Am J Clin Nutr l991;5:lo Jequier E, Acheson KJ, Schut Y. Assessment ofenergy expenditure and fuel utiliation in man. Annu Rev Nutr l987;7: Schut Y, Ravussin E, Diethelm R, Jequier E. Spontaneous physical activity measured by radar in obese and control subjects studied in a respiration chamber. Int J Obes 1982;6: Tappy L, Owen OE, Boden. Effect of hyperinsulinemia on urea pool sie and substrate oxidation rates. Diabetes 1988;37:l Livesey G, Elia M. Estimation of energy expenditure, net carbohydrate utiliation, and net fat oxidation and synthesis by indirect calorimetry: evaluation of errors with special reference to the detailed composition of fuels. Am J Clin Nutr 1988;7: Yeon JY, Youn JH, Bergman RN. Modified protocols improve insulin sensitivity estimation using the minimal model. Am J Physiol 1987;253:E Morgan CR, Larneri. Immunoassay ofinsulin: two antibody system: plasma insulin levels of normal, subdiabetic and diabetic rats. Diabetes 1988;12:l Winer BJ. Statistical principles in experimental design. New York: McGraw-Hill, Krieger DR, Landsberg L. Role of hormones in the etiology and pathogenesis of obesity. In: Frankle RT, Yang MU, eds. Obesity and weight control. Rockville, MD: Aspen Publications, McGarry JD, Foster DW. Regulation ofhepatic oxidation and ketone body production. Annu Rev Biochem l98;9: Zieler K, Rabinowit D. Effect ofvery small concentrations of insulin on forearm metabolism. Persistence ofits actions on potassium and free fatty acids without its effect on glucose. J Clin Invest 196;3: Flatt JP, Ravussin E, Acheson KJ, Jequier E. Effects of dietary fat on postprandial substrate oxidation and on carbohydrate and fat balances. J Clin Invest 1985;76:l Bennett C, Reed GW, Peters JC, Abumrad NN, Sun M, Hill JO. The short-term effects ofdietary fat ingestion on energy expenditure and nutrient balance. Am J Clin Nutr (in press). 23. Schut Y, Hart JP, Jequier E. Failure ofdietary fat intake to promote fat oxidation: a factor favoring the development of obesity. Am J Clin Nutr l989;5:37-l. 2. Acheson KJ, Schut Y, Bessard T, Anantharaman K, Raft JP, Jequier E. Glycogen storage capacity and de novo lipogenesis during massive carbohydrate overfeeding in man. Am J Oin Nutr l988;8: Tremblay A, Plourde G, Despres JP, Bouchard C. Impact of dietary fat content and fat oxidation on energy intake in humans. Am J Clin Nutr 1989;9: Lissner L, Levitsky DA, Strupp BJ, Kalkwarf HJ, Roe DA. Dietary fat and the regulation of energy intake in human subjects. Am J Clin Nutr l987;6: Kendall A, Levitsky DA, Strupp RI, Lissner L. Weight loss on a low-fat diet: consequence of the imprecision of the control of food intake in humans. Am J Clin Nutr 199 1;53:l Abbott WGH, Howard BV, Ruotolo G, Ravussin E. Energy expenditure in humans: effects of dietary fat and carbohydrate. Am J Physiol l99;258:e Downloaded from ajcn.nutrition.org at PENNSYLVANIA STATE UNIV PATERNO LIBRARY on February 21, 213