International Journal of Obesity (1999) 23, 830±838 ß 1999 Stockton Press All rights reserved 0307±0565/99 $12.00 http://www.stockton-press.co.uk/ijo Weight loss and changes in energy metabolism in massively obese adolescents

P Tounian 1*, M-L Frelut 2, G Parlier 1, C Abounaufal 1, N Aymard 3, F Veinberg 4, J-L Fontaine 1 and J-P Girardet 1

1Pediatric Gastroenterology and Nutrition Department, and 4Biochemistry Department, Armand-Trousseau Hospital (AP-HP), 26 avenue du Dr A Netter, 75012 Paris; 2French Red Cross, Margency; and 3Pharmacology Laboratory, Ste Anne Hospital, Paris, France

OBJECTIVE: To investigate the energy metabolism modi®cations induced by energy restriction and weight loss in massively obese adolescents. SUBJECTS: Ten massively obese girls (179Æ 31% of ideal body weight; age, 13.3 ± 16.4 y) after 2 ± 5 weeks on a low- energy diet and 4.5 ± 11.5 months later, that is, after a substantial weight loss, and eight controls. MEASUREMENTS: Resting energy expenditure (REE) and carbohydrate-induced thermogenesis (CIT) after a sucrose load (by indirect calorimetry), plasma glucose and insulin before and after the sucrose load. RESULTS: After 2 ± 5 weeks on a low-energy diet, REE (7415Æ 904 kJ=d) was lower than the expected value calculated from the regression equation of REE on fat free mass in controls (P ˆ 0.005). After a 37Æ 17% reduction in excess weight, REE decreased (6405Æ 613 kJ=d) and remained lower than the expected value (P ˆ 0.005). At the early stages of weight loss, the area under the plasma glucose response curve was negatively correlated with CIT (r ˆ 7 0.80, P ˆ 0.01) and was higher in the six obese adolescents with low CIT than in the four with normal CIT (396Æ 52 vs 283Æ 26 mmol.l71.min71, P ˆ 0.01). After substantial weight loss, the area under the plasma insulin response curve decreased by 32% (P ˆ 0.02), and both CIT and the area under the plasma glucose response curve became similar in obese patients with low and normal CIT prior to weight loss. CONCLUSION: These results indicate that in massively obese adolescents, REE for fat-free mass is decreased at the very beginning of the process of losing weight and remains decreased as long as energy restriction and weight reduction carry on. They also indicate that the impaired CIT sometimes observed returns to normal after weight reduction suggesting that it is secondary to a decrease in glucose uptake induced by obesity-associated insulin resistance.

Keywords: resting energy expenditure; thermic effect of food; energy metabolism; obesity; children; insulin resistance

Introduction genesis which is reduced in some patients1,4 ± 6,10 and normal in others.1,7,11 Weight loss in obese patients mainly results from a Obesity results from an excess of energy intake in decrease in energy intake and to a lesser extent from comparison to energy expenditure. An increase in an increase in physical activity. Studies on energy energy intake is probably the main cause of this metabolism changes in obese patients resulting from imbalance in many obese patients. However, reduced weight loss showed con¯icting results. Weight loss energy expenditure may also be a contributory factor, induces a decrease in REE, but whether this decrease particularly in massively obese patients and in chil- is proportional to the loss of FFM5,10 or greater than dren with early-onset obesity. 1±6 could be accounted for by loss of FFM12,13 remains Total daily energy expenditure has three main controversial. components: resting energy expenditure (REE), Similarly, some ®ndings supported the persistence food-induced thermogenesis, that is the energy expen- after weight reduction of reduced thermogenesis in diture due to the costs of digesting, absorbing, proces- response to carbohydrate,4±6 whereas other studies sing and storing the nutrients, and energy expenditure showed that the defect in food-induced thermogenesis related to physical activity. Energy metabolism of seen in obese patients resolved after weight loss.10,14 obese individuals is characterised by an increased Evaluation of changes in energy expenditure during absolute REE but a similar REE adjusted for fat-free weight loss is of great interest since it may provide mass (FFM) indicating that the elevated REE is due to insights into the mechanisms of the energy metabo- their larger FFM,7±9 and a food-induced thermo- lism disorders observed in obese patients, and may help to understand why massively obese individuals often regain an excessive body weight after periods of *Correspondence: Dr P Tounian, Pediatric Gastroenterology and weight loss. Nutrition Department, Armand-Trousseau Hospital, 26 avenue Most available studies have been done in adults,4± du Dr. Arnold Netter, 75012 Paris, France. 6,12 ± 14 and energy metabolism studies in obese chil- E-mail: [email protected] 9,10,15 Received 21 September 1998; revised 25 February 1999; dren during weight loss are few in number. accepted 18 March 1999 Changes in energy metabolism due to weight loss Energy metabolism in obese adolescents P Tounian et al 831 may be different in obese children, since children France, for a weight-losing program whose two com- have to gain height even during weight reduction. ponents were a controlled diet of low-energy meals The objective of this study was to investigate REE prepared by a dietitian, and a physical activity pro- and carbohydrate-induced thermogenesis (CIT) in gram. The experiment at the beginning of weight massively obese girls at the beginning of and after a reduction was performed after 16 ± 37 d at the substantial weight reduction, in order to determine the centre, and the one after weight reduction 4.5 ± 11.5 nature of the changes in these two components of months later. Both experiments were performed in the energy expenditure resulting from energy restriction same conditions since the patients were on a low- and weight reduction. energy diet (6471Æ 190 kJ=d or 112Æ 6 kJ.kg FFM71.d71 at the beginning of weight reduction, and 6764Æ 412 kJ=d or 131Æ 21 kJ.kg FFM71.d71 after weight reduction) and had already lost 2.3 to Methods 5 kg when the ®rst experiment was performed. The control girls were on a free diet. On each study day, the following parameters were Subjects measured in all subjects: body weight; height; and Ten girls with massive obesity (179Æ 31% of ideal biceps, triceps, subscapular, and suprailiac skinfolds body weight for height) were studied after 2 ± 5 weeks using a Holtain Skinfold Caliper. Relative body on a low-energy diet and 4.5 ± 11.5 months later, that weight (that is body weight=ideal body weight for is after a substantial weight loss. Based on ®ndings height) was calculated using the reference values from a medical history and physical examination, all determined in French children by Sempe and ten patients were considered to be in good health apart PeÂdron.16 FFM and body fat were calculated from from their overweight. The age range was 13.3 ± skinfold thicknesses using the Durnin and Rahaman 16.4 y at study initiation. All ten patients were post- formula.17 menarcheal. Duration of obesity ranged from 4 ± 12 y All experiments were performed in the morning (meanÆ s.d., 9.4Æ 2.6 y). The body weight curves after an overnight fast. The patients rested comforta- showed a steady increase up to the time of admission bly in a recumbent position with their head enclosed to the Nutritional Therapy Center (Margency, France), _ in an air-tight canopy. Oxygen consumption (VO2) indicating that the patients were still in the dynamic _ and carbon dioxide production (VCO2) were continu- phase of obesity. The physical characteristics of the ously measured using an open-circuit indirect calori- patients at the early stages of and after substantial meter (MMC Horizon-Beckman gas analyzer, Sensor weight loss are shown in Table 1. Medics Corp., Anaheim, CA) as previously Results in the obese patients were compared with described.1 On the day of each study, a 24- h urine those in eight healthy age-matched postmenarcheal sample was obtained from each girl for determination girls with normal anthropometric parameters (Table of total nitrogen excretion using the Kjeldahl method. 1). Energy expenditure and the net carbohydrate oxida- Written informed consent was obtained from the _ _ tion rate were calculated from VO2, VCO2 and total girls and their parents before the study. The study nitrogen excretion.18,19 Nonprotein respiratory quoti- protocol was approved by the Paris-Cochin institu- ent was calculated as the ratio of nonprotein tional review board. _ _ 18 VCO2=nonprotein VO2. After an adaptation period of 30 min, REE was measured over 30 min, and each subject then received Experimental protocol an oral sucrose load of 3 g=kg ideal body weight All ten obese girls were admitted to the French Red diluted in the juice of two lemons to be ingested in Cross Nutritional Therapy Centre in Margency, less than 5 min through a ¯exible piece of tygon tubing. Energy expenditure and the net carbohydrate Table 1 Physical characteristics in obese girls at the beginning oxidation rate were measured during 180 min after the of (after 2 to 5 weeks on a low-energy diet) and after substantial weight loss, and in controls (meansÆ s.d.) sucrose load. CIT was calculated as the area under the curve of the energy expenditure rise over the REE Obese girls Obese girls during the 3 h following the sucrose load. During the at the beginning after weight Control of weight loss (n ˆ10) loss (n ˆ10) girls (n ˆ8) indirect calorimetry measurements, the subjects rested quietly while watching television. Age (y) 14.8Æ 1.1 15.4Æ 1.1a 14.1Æ 1.4 Weight (kg) 93.9Æ 7.4 77.0Æ 6.7a 46.4Æ 7.9 Blood samples were collected just before and at 30- Height (m) 1.64Æ 0.06 1.65Æ 0.06b 1.57Æ 0.09 min intervals after the sucrose load, in the obese BMI (kg=m2) 35.1Æ 4.3 28.2Æ 2.7a 18.6Æ 1.4 subjects only, because of ethical considerations. To % IBW 178.5Æ 31.2 141.3Æ 18.0a 99.9Æ 5.4 FFM (kg) 57.3Æ 4.5 51.9Æ 4.4a 34.5Æ 6.5 avoid stress, blood samples were collected via an Body fat (%) 39.0Æ 2.8 32.5Æ 3.4a 25.8Æ 1.7 intravenous catheter inserted into an antecubital vein about one hour before the beginning of the calorime- BMI, body mass index; IBW, ideal body weight for height; FFM, fat-free mass.a,bSigni®cantly different from beginning of weight try study and kept patent with physiological saline. loss, aP ˆ 0.005, bP ˆ 0.03 (Wilcoxon test). Plasma glucose and insulin concentrations were Energy metabolism in obese adolescents P Tounian et al 832 determined at each study timepoint. Plasma adrena- observed FFM using the regression equation obtained line and noradrenaline concentrations were deter- from the controls. The measured REE was lower than mined before and 60 min after the sucrose load. the expected REE in every case (Wilcoxon test, Glycated hemoglobin in blood (HbA1c) was deter- P ˆ 0.005, Figure 1). The expected REE calculated mined before the sucrose load. from the predictive equation for REE from FFM and FM remained higher than the measured FEE (Wil- coxon test, P ˆ 0.02, Table 2). Analytic procedures After weight loss, mean absolute REE in the obese Blood samples were collected into iced tubes and patients decreased by 14% and was no longer sig- immediately centrifuged at 4C, and plasma was  ni®cantly different from that in controls (Table 2). stored at 780 C until analysis. Plasma glucose was REE decreased by 64Æ 38 kJ=d for each kg of weight determined by the glucose oxidase method. Plasma lost. However, the measured REE after weight loss insulin was measured by radioimmunoassay using remained lower than the expected REE calculated polyclonal antibodies (INSI-PR, CIS bio international, either from the FFM after weight loss (Wilcoxon Gif-sur-Yvette, France). Plasma adrenaline and nor- test, P ˆ 0.005, Figure 1) or the FFM and FM after adrenaline were measured using high-performance liquid chromatography with electrochemical detec- tion.20 Another blood sample was collected before the sucrose load for determination of HbA1c by high performance liquid chromatography.

Statistical analyses All results were expressed as means Æ standard deviations (s.d.). A Mann-Whitney test was used for between-group comparisons of means. Comparisons of measured and expected values, and values at the early stages of and after substantial weight loss in each patient were done using a Wilcoxon test. Corre- lations between variables were calculated using simple and multiple regression analysis. Figure 1 Measured resting energy expenditure vs fat free mass in obese girls at the beginning of (after 2 to 5 weeks on a low- energy diet (d)) and after substantial (u) weight loss. The full line is the extrapolated regression line of resting energy expenditure on fat free mass obtained in controls, it represents the expected Results values. The dotted lines are the regression lines of resting energy expenditure on fat free mass obtained in obese girls at the beginning of (- - -) and after (- Á - Á ) weight loss. Measured Weight reduction resting energy expenditures in obese girls were lower than Table 1 shows the changes that occurred in the obese expected values at the beginning of (P ˆ 0.005) as well as after girls following the weight reduction. After a mean of (P ˆ 0.005) weight loss. seven months of dietary treatment, mean weight loss Table 2 Resting energy expenditure (REE) and carbohyrdrate- was 37% of the excess weight for height, which left a induced thermogenesis (CIT) in obese girls at the beginning of mean excess weight of 41%. Mean absolute weight (after 2 to 5 weeks on a low-energy diet) and after substantial loss was 17 kg, composed of 68% fat and 32% fat free weight loss, and in controls (meansÆ s.d.) mass. Obese girls Obese girls Control The obese patients continued to gain height despite at the beginning of after weight girls their signi®cant decrease in weight. weight loss (n ˆ10) loss (n ˆ10) (n ˆ8) Measured REE (kJ=d) 7415Æ 904c 6405Æ 613d 6389Æ 797 e e ExpectedFFM 8599Æ 436 8084Æ 427 Resting energy expenditure REE (kJ=d)a f e In controls, REE was signi®cantly correlated with ExpectedFFM & FM 8154Æ 432 7884Æ 437 REE (kJ=d)b FFM. The predictive equation for REE from FFM CIT (% ingested 2.9Æ 1.7 3.1Æ 1.4 3.2Æ 1.4 was REE (kJ=d) ˆ 97 FFM (kg) ‡ 3041 (r ˆ 0.78; energy) P ˆ 0.02). When entering fat mass (FM) in the CIT (kJ=3h) 74Æ 39 86Æ 47 78Æ 38 model along with FFM using multiple regression a ExpectedFFM REE is calculated by applying the observed fat free analysis, the predictive equation for REE from FFM mass of obese girls to the REE on the fat free mass regression b and FM was REE (kJ=d) ˆ 103 FFM (kg) 7 24 FM equation obtained from the controls. ExpectedFFM & FM REE is calculated by applying the observed fat free mass and fat mass (kg) ‡ 3136 (R ˆ 0.79; P ˆ 0.09). of obese girls to the predictive equation for REE from fat free After 2 to 5 weeks on a low-energy diet, mean mass and fat mass obtained from the controls. cSigni®cantly absolute REE (kJ=d) was 16% higher in the obese different from control girls, P ˆ 0.026 (Mann-Whitney test). dSigni®cantly different from beginning of weight loss, P ˆ 0.005 girls than in the controls (Table 2). For each obese (Wilcoxon test). e,fSigni®cantly different from measured REE, girl, the expected REE was calculated from the eP ˆ 0.005, fP ˆ 0.02 (Wilcoxon test). Energy metabolism in obese adolescents P Tounian et al 833 weight loss (Wilcoxon test, P ˆ 0.005, Table 2). The REE over FFM ratio was unchanged after weight loss (129Æ 11 kJ.kg FFM71.d71 after 2 to 5 weeks on a low-energy diet vs 124Æ 11 kJ.kg FFM71.d71 after weight loss). The fasting nonprotein respiratory quotient in obese girls after 2 to 5 weeks on a low-energy diet was not signi®cantly different from that in controls (0.82Æ 0.05 vs 0.81Æ 0.02) and was not signi®cantly modi®ed by weight loss (0.84Æ 0.05). In obese girls, the mean nonprotein respiratory quotient during the three hours after sucrose ingestion was not signi®- cantly different after 2 to 5 weeks on a low-energy diet (0.96Æ 0.06) and after weight loss (0.99Æ 0.09).

Carbohydrate-induced thermogenesis After 2 to 5 weeks on a low-energy diet, mean CIT expressed as a percentage of ingested energy or as an absolute value (kJ=3 h) was not signi®cantly different between obese and control subjects (Table 2). Sub- stantial between-subject variations in CIT were seen, however, and the obese girls fell into two groups according to whether their pre-weight loss CIT was decreased (n ˆ 4) or normal (n ˆ 6), that is, < or  than CIT 7 1 s.d. in controls. Age, body weight, excess weight for height, FFM, and duration of obesity were not signi®cantly different between these two groups. After weight loss, mean CIT was not signi®cantly modi®ed and remained similar to that in controls (Table 2). CIT became similar in the obese patients with low and normal pre-weight-loss CIT (Table 3). Figure 2 Plasma insulin, plasma glucose and net carbohydrate Moreover, in the group with low CIT before weight (CHO) oxidation rate response curves after the sucrose load loss, CIT after weight loss was similar to that in (given at time 0 min) in obese girls at the beginning of (u±u) controls. and after (r±r) weight loss. Means Æ s.e.m. * Indicates that a signi®cant difference (P < 0.02) was found between mean values.

Biological parameters and net carbohydrate oxidation rate 61Æ 22 pmol=l, P ˆ 0.005), the area under the insulin HbA1c was not signi®cantly modi®ed by weight loss response curve after the sucrose load (102Æ 34 vs (4.8Æ 0.4 at the beginning of weight loss vs 70Æ 33 nmol.l71.min71, P ˆ 0.03), and the peak 5.0Æ 0.8% after weight loss). plasma insulin concentration after the sucrose load The plasma insulin, glycaemia, and net carbohy- (1033Æ 409 vs 689Æ 330 pmol=l, P ˆ 0.02) were sig- drate oxidation rate responses to the sucrose load are ni®cantly lower at the end than at the beginning of the shown in Figure 2. Fasting plasma insulin (135Æ 66 vs weight loss period. However, weight loss was not

Table 3 Metabolic characteristics of obese girls with low pre-weight-loss carbohydrate-induced thermogenesis (CIT), i.e., < mean CIT 7 1 s.d. in controls (Low CIT) or normal pre-weight loss CIT, i.e.,  mean CIT 7 1 s.d. in controls (Normal CIT) (meansÆ s.d.)

At the beginning of weight loss After weight loss

Low CIT (n ˆ4) Normal CIT (n ˆ6) Low CIT (n ˆ4) Normal CIT (n ˆ6)

CIT (% ingested energy) 1.2Æ 0.2e 4.0Æ 1.1 3.4Æ 1.8 2.9Æ 1.3 Insulin areaa (nmol.l71.min71) 105Æ 37 100Æ 35 82Æ 37 59Æ 29 Glucose areab (mmol.l71.min71) 396Æ 52e 283Æ 26 260Æ 82 304Æ 65 CHO Oxidationc (mmol=kg FFM per 3 h) 4.1Æ 1.0 4.1Æ 0.9 4.8Æ 1.9 4.1Æ 1,1 Noradrenaline (T 60=baseline)d 0.90Æ 0.12 1.00Æ 0.15 1.00Æ 0.09 1.05Æ 0.15 Adrenaline (T 60=baseline)d 1.03Æ 0.67 0.62Æ 0.43 0.95Æ 0.20 0.85Æ 0.53 aInsulin area, area under the plasma insulin response curve after the sucrose load. bGlucose area, area under the plasma glucose response curve after the sucrose load. cCHO oxidation, net carbohyrdrate oxidation rate after the sucrose load. dT60=baseline, ratio of plasma concentrations at baseline and 60 min after the sucrose load. eSigni®cantly different from group with normal CIT, P ˆ 0.01 (Mann-Whitney test). Energy metabolism in obese adolescents P Tounian et al 834 associated with signi®cant modi®cations in fasting matched with a theoretical control having the same plasma glucose (4.5Æ 0.3 vs 4.4Æ 0.4 mmol=l), the FFM. Using this approach, we found that after 2 to 5 area under the plasma glucose response curve after the weeks on a low-energy diet our obese girls had low sucrose load (333Æ 70 vs 284Æ 72 mmol.l71.min71), REE values as compared to the controls, and that the peak plasma glucose concentration after the these low values persisted after substantial weight sucrose load (7.1Æ 0.5 vs 7.3Æ 0.8 mmol/l), or the loss. The same results were found when REE was cumulative net carbohydrate oxidation rate during the normalized for fat mass together with FFM using three hours after the sucrose load expressed as the multiple regression analysis. absolute value (232Æ 41 vs 231Æ 81 mmol per 3 h) or Leibel et al 12 showed that the process of losing normalized for FFM (4.1Æ 0.9 vs 4.4Æ 1.5 mmol=kg weight in obese adults on a low-energy diet resulted in FFM per 3 h). Neither at the beginning of weight loss a decrease in REE to a level below that predicted for (noradrenaline: 2.12Æ 0.51 nmol=l (basal) and the decrease in body weight. As a consequence, our 1.99Æ 0.34 nmol=l (after sucrose); adrenaline results are probably due to the fact that obese girls 0.23Æ 0.21 nmol=l (basal) and 0.13Æ 0.11 nmol=l were on a low-energy diet and had lost weight at the (after sucrose)) nor after weight loss (noradrenaline time of both REE measurements. This indicates that 1.96Æ 0.41 nmol=l (basal) and 2.06Æ 0.50 nmol=l the process of losing weight in massively obese (after sucrose); adrenaline:0.21Æ 0.20 nmol=l (basal) adolescents on a low-energy diet is associated with a and 0.25Æ 0.30 nmol=l after sucrose)) did the plasma decrease in REE for FFM at the very beginning of catecholamine levels show any signi®cant changes in weight reduction, and that REE for FFM remains response to the sucrose load. All plasma catechola- decreased as long as energy restriction and weight mine values after weight loss were similar to those at reduction carry on. the corresponding timepoints before weight loss. Energy restriction rather than weight loss per se is CIT was negatively correlated with the area under likely to be the major cause of decrease in REE for the plasma glucose response curve after the sucrose FFM. Indeed, it was shown that, in children9 as well load at the beginning of weight loss (r ˆ 7 0.80, as in adults,22 REE adjusted for FFM measured in P ˆ 0.01) but not after weight loss (r ˆ 7 0.43). No obese subjects after weight loss was not signi®cantly correlations were found between CIT and the area different from that obtained either before weight loss under the insulin response curve after the sucrose or in nonobese controls when REE measurements load, the cumulative net carbohydrate oxidation rate were assessed a few days after the end of the hypo- after the sucrose load, or the changes in plasma caloric diet. Leibel et al 12 also showed that the catecholamines from baseline concentrations. process of losing weight on a hypocaloric diet resulted The obese patients with low CIT prior to weight in a larger reduction in REE than a stabilized weight loss had a signi®cantly higher area under the plasma loss. Furthermore, it was shown in animals that energy glucose response curve than those with normal CIT restriction induces a reduction of weight and oxygen prior to weight loss, whereas the area under the insulin consumption, that is metabolic rate, of gut and liver in response curve, the cumulative net carbohydrate oxi- less than one week while total body weight loss is still dation rate after the sucrose load, and the plasma moderate.23,24 Because of the high metabolic rate of catecholamine changes were similar in the two groups these organs and their major contribution to the (Table 3). After weight loss, CIT and the area under whole-body energy expenditure, the decrease in the plasma glucose response curve became identical in REE for FFM observed in obese girls at the very the two groups, and the other parameters remained beginning of the low-energy diet might be partly similar in the two groups (Table 3). explained by reductions of organ size and metabolic rate rather than the 2.3 to 5 kg weight loss. After weight reduction, absolute REE in obese girls Discussion decreased to the values seen in the controls despite persistence of a 41% excess weight for height and higher FFM values than controls (52 vs 34 kg). This As compared to the controls, the obese girls had a suggests that at a certain point of energy restriction 16% higher absolute REE value. However, to com- and weight reduction, obese children have to either pare individuals with different body compositions, the reduce their energy intake to levels lower than those in metabolic rate should be normalized for FFM. We age-matched, lean girls or increase their physical used a regression-based approach, that is, linear activity to continue to lose weight. These constraints regression of REE on FFM, instead of the ratio probably explain the high failure rate of weight method, that is, the REE over FFM ratio. The ratio reduction therapy for massively obese children.25 method can produce spurious results because the We cannot exclude that the difference between relationship between REE and FFM does not regress measured REE and expected REE for FFM is partly through the zero intercept and therefore the effect of due to an overestimation of FFM in obese children. the normalizing variable is not adequately removed Body fat is probably underestimated in massively from the dependent variable.21 The regression-based obese subjects when assessed by skinfold thicknesses approach used in our study allows each patient to be measurements and using the Durnin and Rahaman Energy metabolism in obese adolescents P Tounian et al 835 formula established from nonobese girls.17 This with the area under the plasma glucose response curve method may therefore artefactually overestimate before but not after weight loss. Since the area under FFM indirectly obtained from body weight minus fat the plasma glucose response curve indirectly re¯ects mass. However, obese girls after weight reduction had glucose uptake by tissues, these results suggest that absolute REE values similar to that in controls despite the impaired CIT observed in our children with persistence of an excess body weight (77 vs 46 kg), massive obesity was secondary to decreased glucose and therefore probably an excess FFM. This clearly utilisation due to insulin resistance. con®rms that REE per unit of FFM was decreased in Ingested sucrose is hydrolysed into fructose and obese girls after weight reduction, independently of a glucose. Fructose is ef®ciently removed from the possible overestimation of FFM. portal circulation by liver cells and is mainly stored Another criticism may be the fact that the predictive as liver glycogen or converted to lactate in the liver, equation of REE developed using nonobese adoles- the lactate subsequently being oxidised in extrahepatic cents was not appropriate for obese adolescents with tissues.27 These metabolic processes are non-insulin- greatly larger FFM and fat mass and may therefore dependent and do not increase plasma glucose con- lead to an overestimation of expected REE. However, centrations.27,28 Glucose is mainly oxidised in periph- when expected REE is calculated from the regression eral tissues and stored as glycogen in muscles and in equation of REE on FFM obtained by Bandini et al 7 in the liver.29 These metabolic processes are insulin- 16 obese girls of the same age (15.2Æ 1.8 y) and FFM dependent and result in an elevation in plasma glucose (52.6Æ 7.8 kg) but not on a low-energy diet, we concentrations.29 These considerations suggest that observe that measured REE in our obese girls remains the increase in the area under the plasma glucose lower than expected, both after 2 to 5 weeks on a low- response curve found in our obese girls with low pre- energy diet (8471Æ 656 kJ=d, P ˆ 0.007) and after weight-loss CIT was probably due to a decrease in substantial weight reduction (7697Æ 642 kJ=d, glucose oxidation or storage rates. Since the net P ˆ 0.005). These data con®rm that energy restriction carbohydrate oxidation rate was similar between the reduces the REE in obese children when compared two groups, it can reasonably be assumed that glucose with obese children on a free diet. storage was decreased in the group with low pre- Roughly equal numbers of studies have con®rmed weight-loss CIT. We therefore suggest that insulin and refuted a defect in food-induced thermogenesis in resistance-induced impairment of glucose storage, obese adults,4,5,11 as well as in children1,7,8,10,26 CIT was which is an energy-requiring process,30 was probably similar in our obese girls and in our controls, con®rming the mechanism underlying the CIT decrease observed our previous report.1 However, considerable interindi- in some of our massively obese girls. vidual variability in CIT values was seen in the obese The post-sucrose load measurement period of three patients, who fell into two groups, one with low and the hours only and the small size of our patient sample are other with normal CIT values. This variability may limitations to our study. The duration of total thermo- explain the discrepancies between the results of studies genesis after sucrose ingestion is longer than three of food-induced thermogenesis in obesity. hours; this probably explains the fact that the CIT The pathophysiological mechanism underlying the values found in our subjects were lower than the defect in food-induced thermogenesis observed in theoretical values.31 However,a few studies in obese some obese patients is still debated. Again, discrepan- adults have identi®ed a decreased rate of glucose cies exist between studies that found a persistent storage due to insulin resistance as the likely cause decrease in food-induced thermogenesis after weight for the decreased glucose-induced thermogenesis in reduction consistent with a primary metabolic disor- these patients.14,32 Moreover, in obese adults glucose der 4±6 and those in which food-induced thermogen- intolerance, that is, hyperglycaemia, after an oral esis returned to normal after weight reduction glucose load results primarily from decreased glucose suggesting an obesity-induced de®ciency in food- storage.33 In obese children, a relation between induced thermogenesis.10,14 Either glucose6,14 or a decreased food-induced thermogenesis and insulin test meal with carbohydrate, fat and protein4,5,10 resistance has been also speculated but not clearly were used in these studies to evaluate the post-pran- demonstrated.8,10 Our results are consistent with these dial energy expenditure. earlier data and therefore lend credence to the hypoth- To gain insight into the mechanism of the CIT esis that decreased glucose storage induced by insulin decrease observed in obese adolescents, we compared resistance is responsible for the CIT decrease our obese girls with low CIT to those with normal observed in some massively obese children. CIT. At the beginning of weight loss, obese girls with Glucose-induced thermogenesis is divided into an low CIT had a signi®cantly higher area under the obligatory and a facultative component.34 The obliga- plasma glucose response curve but a similar area tory component corresponds to the energy required for under the plasma insulin response curve, as compared glucose absorption and storage. The facultative com- to obese girls with normal CIT. After weight loss, CIT ponent is mediated by the sympathetic nervous system and the area under the plasma glucose response curve via catecholamines acting on b-adrenergic receptors. became similar in the two groups. Moreover, in the Studies in obese adults designed to look for a decrease overall obese group, CIT was negatively correlated in the plasma catecholamine response to food Energy metabolism in obese adolescents P Tounian et al 836 ingestion have yielded con¯icting results.5,6,35 7 Bandini LG, Schoeller DA, Dietz WH. Energy expenditure Whether the facultative component of food-induced in obese and nonobese adolescents. Pediatr Res 1990; 27: 198 ± 203. thermogenesis is blunted in obese subjects remains 8 Molnar D, Varga P, Rubecz I, Hamar A, Mestyan J. Food- therefore controversial, as does the possible contribu- induced thermogenesis in obese children. Eur J Pediatr 1985; tion to obesity of a reduction in sympathetic nervous 144: 27 ± 31. system activity.36 Ours is the ®rst study of its kind in 9 Maffeis C, Schutz Y, Pinelli L. Effect of weight loss on resting obese children. The CIT decrease in our patients energy expenditure in obese prepubertal children. Int J Obes 1992; 16: 41 ± 47. involved the obligatory component, and since 10 Maffeis C, Schutz Y, Pinelli L. Postprandial thermogenesis in plasma catecholamine concentrations were not mod- obese children before and after weight reduction. Eur J Clin i®ed by weight loss and were similar in groups of Nutr 1992; 46: 577 ± 583. obese girls with low and normal pre-weight-loss CIT, 11 Welle SL, Campbell RG. Normal thermic effect of glucose in our data failed to support an abnormality in the obese women. Am J Clin Nutr 1983; 37: 87 ± 92. 12 Leibel RL, Rosenbaum M, Hirsch J. Changes in energy facultative component. expenditure resulting from altered body weight. N Engl J In summary, the objective of this study was to Med 1995; 332: 621 ± 628. determine changes in energy expenditure induced by 13 Heshka S, Yang MU, Wang J, Burt P, Pi-Sunyer FX. Weight energy restriction and weight reduction in massively loss and change in resting metabolic rate. Am J Clin Nutr obese adolescents. The low resting energy expenditure 1990; 52: 981 ± 986. 14 Ravussin E, Bogardus C, Schwartz RS, Robbins DC, Wolfe values for fat free mass found early at the beginning of RR, Horton ES, Danforth E, Sims EAH. Thermic effect the process of weight loss suggest a metabolic resis- of infused glucose and insulin in man. Decreased response tance to the maintenance of a reduced body weight with increased insulin resistance in obesity and non- and may explain the poor long-term ef®cacy of weight insulin-dependent diabetes mellitus. J Clin Invest 1983; 72: reduction therapy. The decreased carbohydrate- 893 ± 902. 15 Zwiauer KF, Mueller T, Widhalm K. Resting metabolic rate in induced thermogenesis observed in some of our obese children before, during and after weight loss. Int J Obes obese children at the beginning but not after weight 1992; 16: 11 ± 16. reduction was probably secondary to decreased glu- 16 Sempe M, PeÂdron G, Roy-Pernot MP (eds). Auxologie. MeÂth- cose utilisation induced by obesity-associated insulin odes et seÂquences. Theraplix: 1979. resistance. 17 Durnin JV, Rahaman MM. The assessment of the amount of fat in the human body from measurements of skinfold thick- ness. Br J Nutr 1967; 2: 681 ± 689. Acknowledgements 18 Jequier E. Measurement of energy expenditure in clinical We thank Yves Le Bouc and his staff (Endocrinology nutritional assessment. J Parent Ent Nutr 1987; 11 (Suppl): Laboratory, Armand-Trousseau Hospital) for plasma 865 ± 895. glucose and insulin measurements, and Dominique 19 Ben-Porat M, Sideman S, Bursztein S. Energy metabolism rate Costagliola (INSERM SC4, Saint-Antoine School of equation for fasting and postabsorptive subject. Am J Physiol 1983; 244: R764 ± R769. Medicine, Paris) for her help with the statistical 20 Elworthy PM, Hitchcock ER. Estimation of plasma catecho- analysis. We thank also Annie Leyris (Pharmacology lamines by high-performance liquid chromatography with Laboratory, Ste Anne Hospital) for her technical electrochemical detection in patients with subarachnoid hae- assistance, and the dieticians FreÂdeÂrique Jhean morrhage. J Chromatogr 1986; 380: 33 ± 41. (French Red Cross) and FrancËoise Sarrio (Pediatric 21 Poehlman ET, Toth MJ. Mathematical ratios lead to spurious conclusions regarding age- and sex-related differences Gastroenterology and Nutrition Department). in resting metabolic rate. Am J Clin Nutr 1995; 61: 482 ± 485. References 22 Larson DE, Ferraro RT, Robertson DS, Ravussin E. Energy 1 Tounian P, Girardet JP, Carlier L, Frelut ML, Veinberg F, metabolism in weight-stable postobese individuals. Am J Clin Fontaine JL. Resting energy expenditure and food-induced Nutr 1995; 62: 735 ± 739. thermogenesis in obese children. J Pediatr Gastroenterol Nutr 23 Koong LJ, Ferrell CL, Nienaber JA. Assessment of interrela- 1993; 16: 451 ± 457. tionship among levels of intake and production, organ size and 2 Roberts SB, Savage J, Coward WA, Chew B, Lucas A. Energy fasting heat production in growing animals. J Nutr 1985; 115: expenditure and intake in infants born to lean and overweight 1383 ± 1390. mothers. N Engl J Med 1988; 318: 461 ± 486. 24 Ortigues I, Durand D. Adaptation of energy metabolism to 3 Ravussin E, Lillioja S, Knowler WC, Christin L, Freymond D, undernutrition in ewes. Contribution of portal-drained viscera, Abbott WGH, Boyce V, Howard BV, Bogardus C. Reduced liver and hindquarters. Br J Nutr 1995; 73: 209 ± 226. rate of energy expenditure as a risk factor for body-weight 25 Girardet JP, Tounian P, Lebars MA, Boreux A. ObeÂsite de gain. N Engl J Med 1988; 318: 467 ± 472. l'enfant: inteÂreÃt des indicateurs cliniques d'eÂvaluation. Ann 4 Bessard, T, Schutz Y, JeÂquier E. Energy expenditure and Pediatr 1993; 40: 297 ± 303. postprandial thermogenesis in obese women before and after 26 Epstein LH, Wing RR, Cluss P, Fernstrom MW, Penner B, weight loss. Am J Clin Nutr 1983; 38: 680 ± 693. Perkins KA, Nudelman S, Marks B, Valoski A. Resting 5 Nelson KM, Weinsier RL, James LD, Darnell B, Hunter G, metabolic rate in lean and obese children: relationship to Long CL. Effect of weight reduction on resting energy child and parent weight and precent-overweight change. Am expenditure, substrate utilization, and the thermic effect of J Clin Nutr 1989; 49: 331 ± 336. food in moderately obese women. Am J Clin Nutr 1992; 55: 27 Tounian P, Schneiter P, Henry S, JeÂquier E, Tappy L. Effects 24 ± 33. of infused fructose on endogenous glucose production, gluco- 6 Astrup A, Andersen T, Christensen NJ, BuÈlow J, Madsen J, neogenesis, and glycogen metabolism. Am J Physiol 1994; Breum L, Quaade F. Impaired glucose-induced thermogenesis 267: E710 ± E717. and arterial norepinephrine response persist after weight reduc- 28 Chen M, Whistler RL. Metabolism of D-fructose. Adv Carbo- tion in obese humans. Am J Clin Nutr 1990; 51: 331 ± 337. hydr Chem Biochem 1977; 34: 285 ± 343. Energy metabolism in obese adolescents P Tounian et al 837 29 Kelley D, Mitrakou A, Marsh H, Schwenk F, Benn J, Sonnen- 33 Felber JP, Haesler E, JeÂquier E. Metabolic origin of insulin berg G, Archangeli M, Aoki T, Sorensen J, Berger M, Sonksen resistance in obesity with and without type 2 (non- P, Gerich J. Skeletal muscle glycolysis, oxidation, and storage insulin-dependent) diabetes mellitus. Diabetologia 1993; 36: of an oral glucose load. J Clin Invest 1988; 81: 1563 ± 1571. 1221 ± 1229. 30 Flatt JP. The biochemistry of energy expenditure. In: Bray GA 34 Acheson KJ, Ravussin E, Wahren J, JeÂquier E. Thermic effect (ed). Recent advances in obesity research. Newman Publish- of glucose in man: obligatory and facultative thermogenesis. J ing: London 1978, pp 211 ± 228. Clin Invest 1984; 74: 1572 ± 1580. 31 Tappy L, Jequier E. Fructose and dietary thermogenesis. Am J 35 Schwartz RS, Halter JB, Bierman EL. Reduced thermic effect Clin Nutr 1993; 58: 766S ± 770S. of feeding in obesity: role of norepinephrine. Metabolism 32 Segal KR, Albu J, Chun A, Edano A, Legaspi B, Pi-Sunyer 1983; 32: 114 ± 117. FX. Independent effects of obesity and insulin resistance on 36 Welle S. Sympathetic nervous system response to intake. Am J postprandial thermogenesis in men. J Clin Invest 1992; 89: Clin Nutr 1995; 62 (Suppl): 1118S ± 1122S. 824 ± 833.