Thermic effect of food in humans: methods and results from use of a respiratory chamber1 2

Transcription

1 Thermic effect of food in humans: methods and results from use of a respiratory chamber1 2 Pietro A Tataranni, D Enette Larson, Soren Snitker, and Eric Ravussin ABSTRAT During the past two decades, many investigators have measured the thermic effect of food (TEF) in humans and have speculated on its role in the development of obesity. In this study we compared different ways of computing TEF from daily energy expenditure measurements in a respiratory chamber, evaluated the determinants of TEF, and more importantly assessed for the first time the relation between TEF and change in body weight. In 471 subjects, TEF was 1697 ± 857 kj/d ( ± SD), ie, 18 ± 9% of energy intake. In 1 14 subjects studied more than once, intraindividual TEF variability was very high (V 48%). TEF correlated positively with the level of spontaneous physical activity (SPA) and negatively with fasting plasma glucose and insulin concentrations. TEF correlated inversely with age (males only) and body weight, percent body fat, and waist-to-hip ratio (females only). The level of SPA and fasting plasma glucose concentration were the only significant determinants of TEF, explaining 15% of its variance. In 137 subjects in whom body weight was measured 6 mo after TEF measurement (mean follow-up duration of 2.9 ± 1.7 y), a low TEF was not predictive of body weight gain. We conclude that, despite the low reproducibility of TEF from use of a respiratory chamber, data in a large number of subjects suggest that TEF is increased by higher SPAs and that insulin resistance is associated with a low TEF. More important, longitudinal data indicate that the variability in TEF is not associated with changes in body weight. Am J li,z Nutr 1995;61 : KEY WORDS Energy expenditure, thermogenesis, obesity, predictors of weight gain Introduction The thermic effect of food (TEF) is commonly defined as the increase in energy expenditure in response to food intake (1). Since the early reports of a blunted TEF in obese subjects, great effort has been devoted to uncovering the possible role of TEF in the etiology of obesity. This seems to have occurred because the close correlation between resting metabolic rate (RMR) and body size led to the erroneous assumption (2, 3) that RMR is constant for a given body size and therefore cannot be a determinant of body weight change. On the contrary, because of its large variability, TEF has been considered the only measurable component of daily energy expenditure involved in the development of obesity. However, nearly 2 y after Pittet et al (4) reported a low TEF in obese subjects, there is still disagreement regarding the role of TEF in the pathogenesis of obesity (5). TEF is the most difficult to measure and the least reproducible component of daily energy expenditure because of physiological factors (subject s genetic background, age, physical fitness, and sensitivity to insulin), the characteristics of the test meal (size, composition, palatability, and timing), and most important, methodological problems (indirect calorimetry equipment, interfering environmental factors, and duration of the measurement). TEF is most often measured by using yentilated-hood systems (6, 7) or respiratory chambers (8, 9). Ventilated-hood systems are usually used to assess TEF during short-duration tests in response to a single meal. The use of a respiratory chamber to measure TEF has the advantage of reproducing more physiological conditions over a longer penod of time while regular meals are consumed throughout the day. Schutz et al (9) proposed to calculate TEF over a 15-h (15-h TEF) period using energy expenditure data adjusted for the variability of SPA. In the present study we evaluated TEF using a respiratory chamber and computed its value in three ways: 1) as the difference in 24-h energy expenditure (24-h EE) between two measurements, one in the fed state and the other in the fasted state (TEF); 2) as previously proposed by Schutz et al (9) (15-h TEF); and 3) as the 24-h resting energy expenditure above the sleeping metabolic rate (SMR) integrated over a period of 24 h (24-h TEF). The aims of the study were 1) to compare the three ways of measuring TEF in a respiratory chamber, 2) to assess the determinants of TEF in a large number of subjects, and 3) to prospectively evaluate whether the variability in TEF is related to changes in body weight. Subjects and methods Subjects Since 1985 Pima Indians and whites have been admitted to the metabolic ward of the linical Diabetes and Nutrition I From the linical Diabetes and Nutrition Section, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Phoenix, AZ. 2 Reprints not available. Address correspondence to PA Tataranni, linical Diabetes and Nutrition Section, NIDDK, NIH, 4212 North 16th Street, Room 541, Phoenix, AZ Received October 6, Accepted for publication December 8, A,n J li,z Niar 1995:61: Printed in USA American Society for linical Nutrition 113 Downloaded from on 17 February 218

2 114 TATARANNI El AL Section of the National Institutes of Health in Phoenix, AZ, for prospective studies of obesity including measurements of 24-h EE with a respiratory chamber. These studies were designed to look at metabolic predictors of body weight gain. To assess the reproducibility and the determinants of TEF and to investigate its effect on subsequent weight change, a total of 471 volunteers (291 males, 18 females; 258 Pima Indians, 213 whites) were selected for this analysis. The physical characteristics of the subjects are presented in Table 1. Repeated TEF measurements were available in 1 14 of the 471 subjects. Fasting plasma glucose and insulin concentrations were measured in 38 and in 249 of the 471 subjects, respectively. Body weight and body-composition data at baseline and at follow-up ( 6 mo after TEF measurement) were available for 137 of the 471 subjects. Their physical characteristics at baseline are presented in Table 2. Three methods of measuring the TEF were used in 18 of the 471 subjects (8 males, 1 females; 16 Pima Indians, 2 whites; 26 ± 5 y, 99.7 ± 3.2 kg, 38 ± 8% body fat), who were studied twice in the respiratory chamber, first in the fed state and then in the fasted state. These subjects had a wide range of body weights ( kg) and fatness (22-51% body fat). The interval between the two measurements was < 1 y. Except for obesity, all subjects were healthy and not taking medication. On admission, all subjects received a weightmaintenance diet consisting of 5% carbohydrate, 3% fat, and 2% protein. The protocol was approved by the Institutional Review Board of the National Institute of Diabetes and Digestive and Kidney Diseases, the Indian Health Service, and the Tribal ouncil of the Gila River ommunity. Written, informed consent was obtained. Methods Body composition was estimated by hydrodensitometry with simultaneous determination of lung residual volume and percent body fat calculated from the equation of Sin (1). Body fat distribution was estimated as the ratio of waist-to-thigh circumferences (W:T). ircumferences of the waist (at the level of the umbilicus) and thigh (at the gluteal fold) were measured while subjects were supine and standing, respectively. Plasma glucose concentrations were measured by the glucose oxidase method with a Beckman glucose analyzer (Fullerton, A). Plasma insulin concentrations were determined by using a modification by Herbert et al (11) of the radioimmunoassay method of Berson and Yalow (12). Twenty-four-hour EE was measured in a respiratory chamber as previously described (8). Briefly, the subject entered the respiratory chamber at 8 after 3 d on a weight-maintenance diet, and remained inside until 7 the next day. No vigorous exercise was allowed in the chamber and the SPA was estimated by a radar system. Meals supplied from the metabolic kitchen provided 8% of the energy requirements on the ward to account for the decrease in physical activity in the chamber (13). The 18 subjects studied in the fasted state did not eat any food during the day spent in the respiratory chamber; only water and noncaffeinated diet sodas were provided. SMR was calculated as the metabolic rate measured between 23 and 5 when the 15-mm activity by radar was < 1.5%. Basal metabolic rate (BMR) was measured in the chamber over 21 mm by using a ventilated plastic hood placed over the subject s head at 7, 1 1 h after an evening snack. alibration procedures, precision, accuracy, response time, and variability of the respiratory chamber were previously published (8). alculation of the thermic effect of food TEF was calculated by using three different approaches: 1) ivfef: In the 18 subjects studied twice, TEF was calculated as the difference in 24-h EE between the day in the fed state and the day in the fasted state. For both days, energy expenditure was adjusted for the level of SPA. Briefly, for each subject the energy expenditure due to the SPA was calculated as the product of the mean SPA measured by radar and the slope of the regression line between energy expenditure and activity from 8 to 23. The 24-h EE due to the SPA was then subtracted from total 24-h EE to obtain a resting 24-h EE. 2) 15-h TEF: In all subjects the TEF was calculated as proposed by SchuLz et al (9). Briefly, the mean RMR from 8 to 23 (15 h), which includes the TEF, was obtained from the intercept of the linear regression between energy expenditure and activity measured by radar. The difference between the RMR and BMR represents the TEF of the meals when computed over 15 h, an estimate of the duration of the TEF (Figure 1). TABLE 1 Physical characteristics of 471 subjects studied for 24 h in the respiratory chamber Males (n = 291) Females (n = 18) All (n = 471) Age (y) 3 ± 9 (18-63) 32 ± 11 (1865)2 31 ± 1 (18-65) Weight (kg) 94.8 ± 27.1 ( ) 9.5 ± 27.4 ( ) 93.1 ± 27.2 ( ) Height (cm) ± 6.6 ( ) ± 6.7 ( ) 169. ± 8.7 ( ) Body fat (%) 29 ± 12 (3-57) 42 ± 1 (1-62) 34 ± 13 (3-62) FFM (kg) 64.4 ± 1.4 ( ) 5. ± 9. ( ) 58.9 ± 12.1 ( ) FM (kg) 3.3 ± 19.1 ( ) 4.5 ± 2.1 ( ) 34.2 ± 2.1 ( ) W:T 1.62 ±.18 ( ) 1.54 ±.2 ( ) 1.59 ±.19 ( ) Fasting glucose (mmol/l)5 5. ±.5 ( ) 5.1 ±.4 (4.26.9)2 5. ±.5 ( ) Fasting insulin (pmolil) ± ( ) 222. ± 13.2 ( ) ± ( ) I SD; range in parentheses. FFM, fat-free mass; FM, fat mass; W:T, waist-to-thigh ratio. 2.3 Significantly different from males: 2 p < o.os, p <.1. 4 Measured in 46 subjects (286 males, 174 females). 5 Measured in 38 subjects (191 males, 117 females). 6 Measured in 249 subjects (155 males, 94 females). Downloaded from on 17 February 218

3 ThERMI EFFET OF FOOD AND OBESITY 115 TABLE 2 Physical characteristics at baseline of the 137 subjects with follow-up (mean follow-up 2.9 ± 1.7 y) measurements of body weight Males (n = 92) Females (n 45) All (n = 137) Age (y) 27 ± 6 (18-42) 27 ± 5 (18-41) 27 ± 6 (18-42) Weight (kg) 96.9 ± 24.9 ( ) 98.8 ± 26.9 ( ) 97.5 ± 25.5 ( ) Height (cm) ± 5.8 ( ) ± 5. (14817) ± 7.9 ( ) Body fat (%) 32 ± 1 (8-51) 46 ± 7 (2561)2 37 ± 11 (8-61) FFM (kg) 63.6 ± 9.2 ( ) 51.3 ± 8.8 ( ) ± 1.7 ( ) FM (kg) 33.3 ± 17.5 ( ) 47.5 ± 19.4 ( ) ± 19.3 ( ) W:T 1.66 ±.17 ( ) 1.67 ±.16( ) 1.66 ±.16 ( ).1 ± SD; range in parentheses. Mean body weight changed by 3.6 ± 9.2% of initial body weight at a rate of 1.4 ± 3.9% per year of follow-up. FFM, fat-free mass; FM, fat mass; W:T, waist-to-thigh ratio. 2 Significantly different from males, P <.1. 3) 24-h TEF: In all subjects the TEF was also calculated as the increase of energy expenditure above the SMR corrected for the SPA as described for LTEF (Figure 2). LTEF, 15-h TEF, and 24-h TEF were expressed as an absolute value (kj/d) and as a percentage of energy intake. Statistical Energy (kj/min) analysis All values are expressed as mean ± SD unless otherwise indicated. Statistical analyses were performed by using the procedures of the SAS Institute Inc (ary, N). Plasma insulin concentrations were log transformed to approximate normal distribution. Differences between variables were tested by using Student s t test. P values were considered significant when <.5. For the 1 14 subjects studied at least twice, the withinsubject variability in TEF was expressed as the V. omparisons between TEF and physical characteristics were assessed by Pearson product-moment correlations and by multiple-linear-regression analysis. In those subjects with a follow-up measurement 6 mo after the TEF measurement, the percent Expenditure of body weight and fat mass change (relative to initial body weight/fat mass) and the rate of body weight and fat mass change (percent of body weight/fat mass change per year of follow-up) were calculated and these values were tested against the TEF by Pearson product-moment correlation. The impact of 24-h TEF on subsequent weight change (arbitrarily defined as a weight gain of 1% and 2% of initial body weight), independent of the duration of follow-up, was also examined by proportional hazard analysis (14). Results In 18 subjects studied in both the fed and the fasted states, the energy intake on the fed day was 9748 ± 186 kj/d. LTEF, calculated as the difference between 24-h EE in the fed state and 24-h EE in the fasted state, was 1262 ± 635 kj/d (range E Ui L. Ui I... I - I. -. I.. I -. _ I IS t t t t Time (hours) B L D S Activity by radar (%) FIGURE 1. Relation between energy expenditure and activity in one subject. Each point represents a 15-mm period from 8 to 23. Resting energy expenditure can be calculated as the y intercept of the regression line, and the thermic effect of food (IEF) is calculated as the difference between resting energy expenditure and basal metabolic rate expressed over 15 h (15-h TEF). Modified from Ravussin et al (8). FIGURE 2. Energy expenditure over 24 h in one subject. The upper solid line represents the total energy expenditure averaged over 3-mm periods. The dotted line represents the resting energy expenditure after adjustment for the level of spontaneous physical activity. For graphic presentation only, a 15-point cubic moving average was computed on resting 24-h EE values (solid curved line superimposed on the dotted line). The area under the dotted line and under the solid curved line has the same numerical value. The bottom solid straight line represents the value of the sleeping metabolic rate. The thermic effect of food (TEF) is calculated over 24 h (24-h TEF) as the increase of resting energy expenditure above sleeping metabolic rate (dashed area). Letters represent meal start time: B, breakfast; L, lunch; D, dinner; S, snack. Downloaded from on 17 February 218

4 116 TATARANNI El AL kj/d), ie, 12 ± 5% of energy intake (range 3-21%). Figure 3 shows the time course of 24-h EE adjusted for the SPA. When energy expenditure was adjusted for the SPA, LTEF was 1212 ± 736 kj/d (range -414 to 2512 kj/d), ie, 12 ± 7% of energy intake (range -6% to 22%). Fifteen-hour TEF, calculated on the fed day, was 564 ± 66 kj/d (range -677 to 181 kj/d), ie, 5 ± 7% of energy intake (range -1% to 17%), and was significantly lower than LTEF (P <.1). Twentyfour-hour TEF was ± 786 kj/d (range -171 to 2725 kj/d), ie, 1 1 ± 7% of the energy intake (range -2% to 26%), and was not significantly different from LTEF. LTEF correlated with 15-h TEF (r =.53, P <.5) and 24-h TEF (r =.7, P <.1). There was no correlation between LTEF, 15-h TEF, and 24-h TEF and body weight, percent body fat, and W:T. In the 1 14 subjects with repeated measurements, the intraindividual variability expressed as V was 125% for 15-h TEF (ie, 631 kj/d) and 48% for 24-h TEF (ie, 25 kj/d). Only 24-h TEF values are presented in the rest of the Results section. In 471 subjects, 24-h TEF was 1697 ± 857 kj/d (range -464 to 4786 kj/d), ie, 18 ± 9% (range -5% to 54%) ofenergy intake and there was no difference between sexes and races. The relations between 24-h TEF and age, body weight, percent body fat, W:T, fasting glucose, fasting insulin, and the SPA are presented in Table 3. Twenty-four-hour TEF correlated positively with the SPA and negatively with fasting plasma glucose. In all subjects the 24-h TEF also correlated negatively with fasting plasma insulin concentrations. However, when the sexes were considered separately, this correlation was significant only in females. In males only, a weak negative correlation between 24-h TEF and age was found. In females, 24-h TEF correlated negatively with body weight, percent body fat, waist circumference, and W:T. By multiple-regression analysis in E U I I I I I I t t t t Time (hours) B L D S FIGURE 3. Energy expenditure measured on the day in the fed state (upper line) and the day in the fasted state (lower line). For both days energy expenditure was adjusted for the level of spontaneous physical activity (SPA) measured by radar. Each line represents the mean ± SE (shaded areas) of the 18 individual 24-h resting metabolic rates smoothed by a 15-point cubic moving average. The thermic effect of food was computed as the difference between energy expenditure on the day in the fed state and energy expenditure on the day in the fasted day (TEF). Letters represent meal start time: B, breakfast; L, lunch; D, dinner; S, snack. Only water and noncaffeinated diet sodas were consumed during the day in the fasted state. TABLE 3 Pearson correlation coefficients for the correlation between the thermic effect of food (TEF) computed as 24-h resting energy expenditure above sleeping metabolic rate (24-h TEF) and physiological variables 24-h TEF Males Females All (n = 291) (n = 18) (n = 471) Age Weight Percent body fat W:r Waist l -.4 Fasting glucose Fasting insulin SPA W:T, waist-to-thigh ratio: SPA, spontaneous physical activity. 2 p <.5. 3 P <.1. 4 Measured in 46 subjects (286 males, 174 females). 5 Measured in 38 subjects ( 191 males, 1 17 females). 6 Measured in 249 subjects ( 155 males, 94 females). 38 subjects, only the SPA and plasma fasting glucose concentrations were significant, independent determinants of 24-h TEF, but explained only 15% of the interindividual variability (,2 o.is, p <.1). In the 137 subjects with follow-up measurements (mean follow-up 2.9 ± 1.7 y), mean body weight changed by 3.6 ± 9.2% of initial body weight (range -25.5% to 3.1%) with a rate of change of 1.4 ± 3.9% per year of follow-up. Twentyfour-hour TEF did not correlate with changes in body weight (Figure 4), percent body fat, or fat mass. By proportional hazard analysis, 24-h TEF was not a predictor of body weight gain, arbitrarily defined as 1% and 2% of initial body weight (P -.7 and P -.6, respectively). Discussion The TEF measured in a respiratory chamber and calculated as previously proposed by Schutz et al (9) underestimated the TEF of food calculated as the difference in 24-h EE when subjects were fed or fasted. On the other hand, the TEF calculated as the 24-h resting energy expenditure above the SMR, was similar to the difference in 24-h EE when subjects were fed or fasted. The TEF correlated positively with the level of SPA -a and negatively with fasting plasma glucose and insulin concentrations. Also, negative correlations with age (males only) and body weight, percent body fat, and waist-to-hip ratio (females only) were observed. The level of SPA and fasting plasma glucose explained 15% of variance in the TEF. ontrary to what is often suggested, a low TEF did not predict body weight gain. Despite an intensive research effort over the past two decades to relate human obesity to a blunted TEF, very little attention has been devoted to methodological appraisal of this measurement. In the majority of studies, the TEF was measured by using a ventilated-hood system. Recently, Segal et al (7) reported that the TEF determined by ventilated-hood systems has an intraindividual V of 5.7%, which is much less than the 3% variation reported by Weststrate (6). The TEF has also Downloaded from on 17 February 218

5 4 3 I I I -1-2 #{149}:#{149}#{149}_51 :.- #{149} S.11s,!_ I! #{149} #{149} ; #{149};.:, - #{149} - I #{149} #{149} ThERMI EFFET OF FOOD AND OBESITY 117 r=o.1o P=O r-o.11, P=O #{149} #{149} 5 #{149} #{149}.#{149}.: #{149}- #{149} #{149} #{149}_:._-#{149}#{149}, #{149} #{149} #{149} #{149}.-1:.1i...- #{149},u..;. I u.s #{149} #{149} #{149} (6). TEF (S Enirgy Intake) FIGURE 4#{149} Relation between the thermic effect of food (TEF) and percent body weight change (weight change/initial weight) and rate of body weight change (percent weight change/years of follow-up) in 137 subjects. been measured by using respiratory chambers (8, 9) and is by far the least reproducible of all the components of 24-h EE. Ravussin et al (8) reported a within-subject (n = 12) variability of 43% when computing the TEF as proposed by Schutz et al (9). The reasons for the poor reliability of this method are likely to be related to the large variability of the terms used in the computation of TEF, ie, BMR and the intercept of the regression line between energy expenditure and SPA. Both values can be subject to errors: eg, the patient can move or fall asleep during the BMR measurement and the intercept value can be over- or underestimated as the result of a few extreme points influencing the slope of the relationship. Indeed, in our analysis 15-h TEF varied considerably within an individual from day to day. Also, 15-h TEF underestimated LVI EF, which theoretically represents the best way of measuring TEF over 24 h, because it is the actual difference between the 24-h EE in the fed state and the 24-h EE in the fasted state. The 24-h TEF was calculated by using a more reproducible baseline than BMR, ie, the SMR (8), and a slightly different adjustment for the level of SPA. The 24-h TEF tended to be more reproducible than 15-h TEF, but, more importantly, correlated better with LTEF than with 15-h TEF and was of the same magnitude as LTEF. This is because the 24-h TEF includes the energy cost of arousal, which might explain at least in part the difference observed between 15-h TEF and 24-h TEF. The rest of the discussion, therefore, refers to 24-h TEF only. The reasons for the low reproducibility of TEF remain to be explained. The validity of the measurement of physical activity by the radar system itself is questionable. For example, when the subject is standing or lying (motionless in both cases), the radar will detect no movements; however, the energy expenditure is higher in the standing position than in the lying position. Also, the radar detects only the amount of time the patient is in motion, but does not provide information on the intensity of the activity. These problems can account for that part of intraindividual variability of TEF that is due to the method. Biological day-to-day variation in the postprandial processing of nutrients, due to differences in food intake or physical activity, can also be a source of intraindividual vanation of the TEF. In this study, all subjects had been on a weight-maintenance diet for 3 d before the TEF measurement, and were not allowed to drink any alcohol or to perform strenuous physical activity while on the metabolic ward. Because the TEF was calculated over a 24-h period, circadian variation of the TEF (15) cannot be considered as an additional source of variability. The phase of the menstrual cycle was not controlled in this study. This is, however, unlikely to explain some of the variability in females, because the phase of the menstrual cycle was previously shown not to affect the TEF Although our data indicate that the use of a respiratory chamber is not ideal for measurement of the TEF, the available 24-h results in a large number of subjects allowed us to address some important questions regarding the physiological determinants of TEF and the possible role of a low TEF in the development of obesity. In fact, given the intraindividual vanability observed in our population, it is possible to calculate that sample sizes of 161, 25, and 7 subjects, respectively, would be sufficient to detect group differences of 1%, 25%, or 5% in the TEF with a power 9%. This implies that if large differences in the TEF are important predictors of body weight changes, the number of subjects involved in this study would be large enough to detect such an effect. The role of biological variables such as age, body weight, percent body fat, body fat distribution, fasting plasma glucose and insulin, and the SPA as determinants of intenindividual variability in the TEF was explored. In males only, 24-h TEF correlated inversely with age. This relation was observed previously (16) and is thought to be related to a decreased sensitivity ofthe sympathetic nervous system with age (17). As have some investigators (18-2), but not all (21), we found a negative correlation between the TEF and body weight, percent body fat, and upper body obesity in females only. Vansant et al (22) reported a positive relation between upper-body obesity and the TEF (22). However, in a very elegant study, Brundin et al (23) showed that the TEF is a function of the heat leakage across the abdominal wall, which is inversely related to the thickness of the abdominal adipose tissue layer (23). It is therefore not surprising that in our analysis, subjects with a larger waist circumference, ie, with a thicker abdominal adipose layer, had a lower TEF. Downloaded from on 17 February 218

6 118 TATARANNI El AL The finding of a relation between the TEF and the level of SPA in both sexes confirms other reports on thc effect of exercise on the TEF (24, 25). However, Segal and Pi-Sunyer (26) suggested that the magnitude of this effect in terms of actual energy is minimal, and therefore is of little importance for energy balance. As previously reported (1, 27-29), the TEF was inversely related to fasting plasma glucose and insulin concentrations, both indirect indexes of insulin resistance. A reduced rate of glucose storage as glycogen in the skeletal muscle (3), and an impaired activation of the sympathetic nervous system (31, 32) and of the sodium-potassium pump (33) have been indicated as possible mechanisms by which insulin resistance may reduce the TEF. Together with the level of SPA, fasting plasma glucose was the only other significant determinant of TEF and could explain only 15% of the variance in the TEF. The remaining unexplained variance may be accounted for by biological variables that were not considered in this analysis (ie, hormones involved in the digestive process and activation of the sympathetic nervous system) and methodological errors. The relation between the TEF and obesity has been intensively studied. Several cross-sectional studies (34-36), but not all (37-39), have indicated a blunted TEF in obese people. This led to the hypothesis that a defective thermogenesis could contribute to the maintenance of obesity or could cause obesity itself. Obese subjects have a higher energy expenditure than lean subjects. Weight gain is associated with an increase in BMR and in the cost of physical activity, which easily offset the small energy deficit related to a blunted thenmogenesis. Therefore a reduced 1FF is unlikely to be an important determinant of energy balance in obese subjects. Studies in postobese subjects have also given contradictory results about the importance of the TEF in the etiology of obesity (4-42). Despite the large quantity of data on the relation between TEF and obesity, the role of the TEF on subsequent weight gain has never been explored in a prospective manner. In our study the TEF was not related to changes in body weight, percent body fat, or fat mass and did not predict subsequent weight gain. Accordingly, a low TEF cannot be considered a causal factor of obesity, but more likely a consequence (23). We conclude that the measurement of TEF in a respiratory chamber is not ideal because of its poor reproducibility. However, data in hundreds of subjects confirm that the TEF is increased by higher levels of physical activity and is lower in people with insulin resistance. More important, prospective data indicate that a blunted TEF in obese people is more likely to be a consequence rather than a cause of obesity. U References 1. Jequier E, Schutz Y. Energy expenditure in obesity and diabetes. Diabetes Metab Rev 1988;4: Ravussin E, Bogardus. Relationship of genetics, age, and physical fitness to daily energy expenditure and fuel utilization. Am J lin Nutr 1989;49: Ravussin E, Lillioja 5, Knowler W, et al. Reduced rate of energy expenditure as a risk factor for body-weight gain. N EngI J Med 1988;318: Pittet P, happuis P. Acheson K, de Techtermann F, J#{233}quierE. Thermic effect of glucose in obese subjects studied by direct and indirect calorimetry. Br J Nutr l976;35: Ravussin E, Swinburn BA. Energy metabolism. In: Stunkard AJ, Wadden TA, eds. Obesity: theory and therapy. New York: Raven Press, 1993: Weststrate JA. Resting metabolic rate and diet-induced thermogenesis: a methodological reappraisal. Am J lin Nutr 1993:58: Segal KR, hun A, oronel P. ruz-noori A, Santos R. Reliability of the measurement of postprandial thermogenesis in men of three levels of body fatness. Metabolism 1992:7: Ravussin E, Lillioja 5, Anderson TE, hristin L, Bogardus. Determinants of 24-hour energy expenditure in man. J lin Invest 1986:78: Schutz Y, Bessard T, Jequier E. Diet-induced thermogenesis measured over a whole day in obese and nonobese women. Am J lin Nutr 1984:4: Sin WE. Body composition from fluid density: analysis of methods. In: Brozek J, Herschel A, eds. Techniques for measuring body composition. Washington, D: National Academy of Sciences, National Research ouncil, 1961: Herbert V, Lan K, Gottlieb W, Bleicher SI. oated charcoal immunoassay of insulin. J lin Endocrinol Metab 1965:25: Berson SA, Yalow RS. Quantitative aspects of reaction between insulin and insulin binding antibody. J lin Invest 196;38: Abbott WGH, Howard By, hristin L, et al. Short-term energy balance: relationship with protein, carbohydrate, and fat balances. Am J Physiol 1988;255:E ox DR. Regression model and life-tables. J R Stat Soc 1972:34: Romon M, Edme JL, Boulenguez, Lescroart JL, Frimat P. ircadian variation ofdiet-induced thermogenesis. Am J lin Nutr 1993;57: Th#{246}rneA, Wahren J. Diminished meal-induced thermogenesis in elderly men. lin Physiol 199;1: Schwartz RS, Jaeger LF, Veith R. The thermic effect of feeding on older men: the importance of the sympathetic nervous system. Metabolism 199;39: Segal KR, Gutin B, Albu J, Pi-Sunyer FX. Thermic effect of food and exercise in lean and obese men of similar lean body mass. Am J Physiol 1987;252:E Segal KR, Gutin B, Asa MN, Pi-Sunyer FX. Thermic effect of food at rest, during exercise, and after exercise in lean and obese men of similar body weight. J lin Invest 1985:76: Segal KR, Lacayanga I, Dunaif A, Gutin B, Pi-Sunyer FX. Impact of body fat mass and percent fat on metabolic rate and thermogenesis in men. Am J Physiol 1989;256:E D Alessio DA, Kavie E, Mozzoli MA, et al. Thermic effect of food in lean and obese men. J lin Invest 1988:81: Vansant GA, Van Gaal LF, De Lceuw IH. Decreased diet-induced thermogenesis in gluteal-femoral obesity. J Am oil Nutr 1989:8: Brundin T, Thdrne A, Wharen J. Heat leakage across the abdominal wall and meal-induced thermogenesis in normal-weight and obese subjects. Metabolism 1992;41 : Bogardus, Thuillez P. Ravussin E, Vasquez B, Narimiga N, Azhar S. Effect of muscle glycogen depletion on in vivo insulin action in man. J lin Invest 1983:72: Von Euler US. Sympathoadrenal activity in physical exercise. Med Sci Sports 1974;6: Segal KR, Pi-Sunyer FX. Exercise, resting metabolic rate, and thermogenesis. Diabetes Metab Rev 1986:2: Bogardus, Lillioja 5, Mott D, Zawadzki J, Young A, Abbott W. Evidence for reduced thermic effect of insulin and glucose infusions in Pima Indians. J lin Invest 1985:75: Golay A, Schutz Y, Meyer HU, et al. Glucose-induced thermogenesis in nondiabetic and diabetic obese subjects. Diabetes. l982;31: Ravussin E, Bogardus, Schwartz RS, et al. Thermic effect of infused glucose and insulin in man. Decreased response with increased insulin resistance in obesity and non insulin-dependent diabetes mellitus. J lin Invest 1983;72: Downloaded from on 17 February 218

7 THERMI EFFET OF FOOD AND OBESITY Felher JP, Meyer HU, urchod B, et al. Glucose storage and oxidation in different degrees of human obesity measured by continuous indirect calorimetry. Diabetologia 1981 ;2: hristin L, Nacht A, Vernet, Ravussin E, Jequier E, Acheson Ki. Insulin, its role in the thermic effect of glucose. J lin Invest 1986; 77: Rowe JW, Young JB, Minaker KL, Stevens AL, Pallotta J, Landsberg L. Effect of insulin and glucose infusions on sympathetic nervous system activity in normal man. Diabetes 1981:3: Mott DM, lark RL, Andrews JW, Foley JE. Insulin-resistant Na pump activity in adipocytes from obese humans. Am J Physiol 1985: 249:El Bessard T, Schutz Y, Jequier E. Energy expenditure and postprandial thermogenesis in obese women before and after weight loss. Am J lin Nutr I983;38: Katzeff HL, Danforth E. Decreased thermic effect of a mixed meal during overnutrition in human obesity. Am J lin Nutr 1989:5: Swaminathan R, King RFGJ, Holmfield J, Siwek RA, Baker M, Wales JK. Thermic effect of feeding carbohydrate, fat, protein and mixed meal in lean and obese subjects. Am J lin Nutr 1985;42:l77-8l. 37. Felig P, unningham J, Levitt M, Hendler R, Nadel E. Energy expenditure in obesity in fasting and postprandial state. Am J Physiol 1983;244:E Nair KS, Halliday D, Garrow is. Thermic response to isoenergetic protein, carbohydrate or fat meals in lean and obese subjects. lin Sd 1983;65: Welle SL, ampbell RG. Normal thermic effect of glucose in obese women. Am J lin Nutr 1983:37: Schutz Y, Golay A, Felber JP, iequier E. Decreased glucose-induced thermogenesis after weight loss in obese subjects: a predisposing factor for relapse of obesity? Am I lin Nutr 1984:39: Tataranni PA, Mingrone G, Greco AV, et al. Glucose-induced thermogenesis in postobese women who have undergone biliopancreatic diversion. Am J lin Nutr 1994:6: Th#{246}rneA, Nhslund I, Wahren J. Meal-induced thermogenesis in previously obese patients. lin Physiol 199:1: Downloaded from on 17 February 218