Research Department of Human Nutrition, The Royal Veterinary and Agricultural University, 1958 Frederiksberg, Denmark

Transcription

1 International Journal of Obesity (1998) 22, 869±877 ß 1998 Stockton Press All rights reserved 0307±0565/98 $ Substrate oxidation and thyroid hormone response to the introduction of a high fat diet in formerly obese women B Buemann, S Toubro and A Astrup Research Department of Human Nutrition, The Royal Veterinary and Agricultural University, 1958 Frederiksberg, Denmark OBJECTIVE: To investigate the adaptation in substrate utilization to a sudden change in dietary composition from a medium fat to a high fat diet, during a three day period in formerly obese and never obese women. METHODS: Energy expenditure (EE) and substrate oxidation rates were measured in eight healthy formerly obese women and eight never obese controls, during four consecutive days in a respiration chamber. The rst day and the day prior to the experiment, the subjects consumed a diet with 30 energy-% fat, whereas the diet had 55 energy-% fat on the subsequent three days. RESULTS: The rate of adjustment of oxidative substrate partitioning expressed as 24 h non-protein respiratory quotient (RQnp) was similar in the two groups. RQnp on each of the days was also similar between the two groups, after accounting for a group difference in energy balance, caused by a non-signi cantly lower EE in the formerly obese women. However, the formerly obese subjects, demonstrated a greater suppression of postprandial fat oxidation after supper, which was unrelated to energy balance. Furthermore, the formerly obese subjects, in contrast to the controls, exhibited a reduction in plasma triiodothyronine/thyroxine ratio (T 3 /T 4 ) following the high fat diet. A positive correlation between T 3 /T 4 and EE was found in the 16 subjects. CONCLUSIONS: The formerly obese subjects did not show a slower adaptation rate of substrate utilization when challenged with a high fat diet, but exhibited an enhanced suppression of fat oxidation and a lower T 3 /T 4 ratio after supper, when fed a high fat diet. Keywords: diet composition; macronutrient balance; obesity; substrate utilization; whole body calorimetry Introduction Numerous studies have investigated possible metabolic abnormalities predisposing to obesity. In particular, a poor capacity to oxidize fat has been suggested to be associated with the development of obesity 1. However, two 24 h whole-body calorimetry studies, 2,3 where the participants were exposed to diets with a varying macronutrient composition, could only demonstrate a reduced lipid utilization in formerly obese subjects when they were fed a high fat diet. Energy balance is a strong determining factor for the partitioning between lipid and carbohydrate oxidation, since a surplus or de cit in energy balance is accounted for by a surplus or de cit in lipid balance, while carbohydrate oxidation on a 24 h basis relates Abbreviations: ANOVA ˆ analysis of variance. BMI ˆ body mass index. EE ˆ energy expenditure. HDL ˆ high density lipoprotein, NEFA ˆ non-esteri ed fatty acids. RQnp ˆ non-protein respiratory quotient. SPA ˆ spontaneous physical activity. T 3 ˆ triiodothyronine. T 4 ˆ thyroxine. Correspondence: Benjamin Buemann, Research Department of Human Nutrition, The Royal Veterinary and Agricultural University, Rolighedsvej 30, 1958 Frederiksberg, Denmark. Received 21 July 1997; revised 10 April 1998; accepted 20 April 1998 directly to carbohydrate intake. 4 Furthermore, carryover effects from the diet consumed during the days prior to the experiment, may in uence the substrate partitioning. Differences in dietary habits or energy restriction may, therefore, confound the comparison between study groups of short term substrate utilization. Accurate control of energy balance and macronutrient intake, preceding and during the experiment is, therefore, crucial when studying 24 h substrate utilization rates. Macronutrient manipulation of energy balance has been reported to induce a rapid shift in substrate oxidations. 5±8 However, the immediate adjustment of macronutrient utilization reported in these studies, was in some cases incomplete. In particular, the introduction of a medium fat diet, has been found to cause a transient positive carbohydrate balance, and a negative fat balance, in spite of an energy balance close to zero. 5±8 Provided that energy balance is maintained, these acute substrate imbalances must gradually vanish, as the capacity of the con ned glycogen stores to act as a substrate buffer is exceeded. Consequently, a study design where indirect calorimetry is performed after several days when an energy balance has been achieved through a diet with a constant macronutrient composition may, in fact, fail to disclose possible differences in the capacity to oxidize lipid.

2 870 A more pertinent approach may, therefore, be to study the dynamic response of macronutrient metabolism during a tightly controlled acute change in diet composition. Differences in hormonal or enzymatic determinants of substrate partitioning may thus be revealed, by different rates of increase in lipid oxidation after an acute change from a low to a high fat diet. To provide a comprehensive evaluation of the dynamics in the adjustment of substrate utilization to dietary alterations, indirect calorimetry studies are required where energy and macronutrient balances are assessed throughout a period of several days. In the present study, whole-body indirect calorimetry was conducted for a period of four consecutive days, to clarify whether individuals with a predisposition to obesity demonstrate a delayed adjustment of substrate oxidation rates or an abnormal hormonal response when exposed to a sudden rise in the fat content of an isocaloric diet. Hour by hour pro les of non-protein respiratory quotient (RQnp) were compared to disclose possible differences in the response to daily activities. Table 1 Subject characteristics. Mean s.e.m. Range indicated in parentheses FO (n ˆ 8) Controls (n ˆ 8) P Age (y) (27± 47) (30± 43) Weight (kg) (53.2± 68.4) (54.8± 68.1) BMI (kg/m 2 ) (19.8± 24.3) (19.6± 24.0) BMI max (kg/m 2 ) < (28.7± 38.2) (20.1± 24.9) Pre-study weight changes (g/d) a (7120±19) (7183±91) Fat free mass (kg) (41.5± 55.7) (43.4± 48.5) Fat mass (kg) (11.2± 18.8) (11.4± 19.9) FO ˆ formerly obese; BMI ˆ body mass index; BMI max ˆ maximum BMI based on measured height and self reported maximum weight. a Pre-study weight changes are calculated as body weight on day 1 minus body weight measured at the screening examination divided by the number of days between the two events ( d, range 6± 27 d). The other gures are based on measurements made immediately before commencing chamber experiments on day 1. Methods Subjects Eight formerly obese women and eight never-obese matched controls, were accommodated in a respiration chamber (described in detail in Ref. 9 for four consecutive days. The majority of the formerly obese subjects were recruited from a group of women who had participated in dietary slimming programmes at our department. All subjects had been weight stable ( 2 kg) for a three month period, documented by questionnaire or by reports from the department's dietitian or the subject's general practitioner. The rate of weight changes (g/d) between the screening examination and the rst day of the metabolic measurements, did not differ signi cantly between the groups (Table 1). None of the subjects had undergone any surgical procedures to achieve weight loss. All subjects were premenopausal and healthy, without any metabolic or eating disorders. No subjects were pregnant, breast feeding or taking any medication regularly. Three of the formerly obese subjects and four of the controls were smokers. The characteristics of the subjects are presented in Table 1. Diet On the rst day in the chamber (day 1) a medium fat diet was given containing 30% energy as fat and 55% energy as carbohydrate. For the remaining three days of the experiment, the diet was changed to a high fat diet containing 55% energy as fat and 30% energy as carbohydrate. The day prior to day 1, the subjects consumed, at home, a diet prepared by the metabolic kitchen, with a composition similar to the medium fat diet of day 1. From an algorithm based on results from previous measurements of energy expenditure (EE) in lean subjects and using a similar protocol, performed at the department, 24 h energy requirements of each subject were calculated: EE kj ˆ 153:9 fat free mass kg 1166: The diets were prepared from ordinary food items used in conventional Danish meals. A 10 MJ standard diet was calculated for each of the four experimental days and the day preceding the experiment. Energy content, macronutrient composition and bre content, of the diets were calculated by DANKOST dietary assessment software (Danish National Food Agency). The 24 h diets were prepared with differences in energy level of 0.5 MJ, and each subject was appointed to the level closest to her individually calculated energy requirement. Adiabatic bomb calorimetry (IKA, Janke & Kunkel GmbH, Staufen, Germany) was performed on freeze dried samples from the 10 MJ standard diets given on days 0, 1, 2 and 4, to determine their gross energy. We were not able to estimate the energy content of the day 3 diet, because one of the food items was missed in the collection for bomb calorimetry. For the calculation of energy and substrate balances, gross energy of the diet on day 3 was taken to be the mean between the gross energy of the day 2 and day 4 diets. Three meals were served on each day of the stay in the respiration chamber and virtually all the food was eaten. 1 The macronutrient composition of each meal was the same as the macronutrient composition of the diet for the entire day. Breakfast, lunch and supper constituted 25%, 25% and 50% of the total energy intake, respectively. Fibre content of a 10 MJ diet ranged between 21.8 g on day 4 and 23.3 g on day 3.

3 Protocol Indirect calorimetry was performed for a 23 h period on each of the four days. The measurements were discontinued each morning between 09.00±10.00 h, to allow cleaning of the chambers, morning toilet, and scheduled measurements and blood samples (see below). The same schedule was followed on each of the four experiment days. Breakfast, lunch and supper were served at h, h and h, respectively. The subjects were in bed between 23.00± h. Five periods of light exercise during daytime (a bicycling session at 75 watts for 10 min at h, h and h, and short walking sessions at h and h) were included in the programme. The subjects were otherwise free to follow their own daily routine, but they were not allowed to take additional exercise. A radio, a television, a video and a telephone were available in the chamber and could be used freely during daytime and evening. All the experiments commenced on Monday morning, but the subjects were accommodated in the chamber from Sunday night. No attempt was made to conduct the experiments while the women were in the same phase of the menstrual cycle, however, ve of the subjects in both the study groups were in the luteal phase (calculated from the time of the onset of last menstruation) on the rst study day. Smokers were allowed to smoke Prince Light cigarettes and the number of cigarettes was recorded. The oxygen consumption and carbon dioxide production of Prince Light cigarette combustion has previously been measured and could therefore be accounted for when calculating the subject's oxidation. Smoking was allowed between 09.00±23.00 h. The experiments were under constant surveillance by a technician or a medical student. Spontaneous physical activity (SPA) was assessed by a microwave radar (Sisor Mini Radar, Static Input System SA, Lausanne, Switzerland) and heart rate was measured by a telemetry system (Dialogue 2000, Danica Electronics, Copenhagen, Denmark). SPA and heart rate were continuously stored on a hard disk parallel to the gas measurements. Fat free mass (FFM) was assessed on each morning of the experiment by body weight and tetrapolary bioimpedance (ANIMETER, H.T.S. Engineering, Odense, Denmark) using the algorithm given by Heitmann. 10 Fasting blood samples were drawn from an antecubital vein in the morning of day 2, before the rst meal of the high fat diet, and in the morning after the 3 d period on the high fat diet. Plasma glucose and lactate concentrations were analysed by standard enzymatic methods. 11 Non-esteri ed fatty acids (NEFA) was determined as described by Laurell and Tibbling 12 and glycerol was assessed following trichloracetic acid precipitation, according to Chernick. 13 Plasma triglyceride was determined as described by Giegel et al. 14 Plasma insulin and cortisol were assayed using radio immunoassay (RIA) kits (Novo Nordisk, Bagsvaerd, Danmark and Farmos Diagnostica, Turku, Finland, respectively), (intra-assay CV ˆ 9% and 4.85% for insulin and cortisol, respectively). For determination of catecholamines, blood was collected in tubes containing reduced glutathione and ethylene glycol-bis(b-aminoethyl ether)-n,n,n 0,N 0 -tetraacetic acid. The tubes were centrifuged immediately and plasma was stored at 780 C until analysis by a radioenzymatic method, 15 (intra-assay CV ˆ 6% and 8% for noradrenaline and adrenaline, respectively). Free thyroxine (T 4 ), and free triiodothyronine (T 3 ), (intra-assay CV 5.3 and 4.75% for free T 4 and free T 3, respectively) were measured enzymatically by an immunoassay kit (Serozym, SERENO Diagnostics SA, Switzerland). For technical reasons, different kits were used for analysis of thyroid hormones from the two dietary regimens. Comparisons between days of absolute values of free T 3 and T 4 between diets, are therefore not possible. However, a possible inter-assay difference should not invalidate the comparisons between the two groups of diet induced changes in thyroid hormone values, as the interassay difference can be expected to impose the same bias on both groups. Calculations and statistics EE and non-protein respiratory quotient (RQnp) were calculated from O 2 uptake, CO 2 production and urinary nitrogen excretion, assuming that 6.25 g protein is oxidized for each g nitrogen excreted in the urine and using the heat equivalent factors given by Brouwer, 16 to calculate EE. To calculate 24 h EE and substrate oxidations, the missing hour was accounted for by extrapolation, assuming that the values for this hour equalled average daytime metabolism (i.e., from to h). Urine was collected for a daytime interval (10.00±23.00 h) and a night-time interval (23.00±09.00 h). Constant protein oxidation rates were assumed throughout these periods. In addition to total 24 h EE and RQnp, hour by hour RQnp were also calculated. Differences between the groups in metabolic data and blood plasma concentrations, were tested by simple ANOVA or repeated measures ANOVA, when applicable. Twenty-three-hour RQnp was adjusted to zero energy balance along the regression line of the simple regression of RQnp on energy balance. RQnp was compared to non protein foo quotient (FQnp) in order to evaluate the adaptation in oxidative macronutrient partitioning to the change in dietary composition. The de nition of FQnp was based on the ratios between carbohydrate and fat in the non-protein part of the diet and calculated as (energy-%np carbohydrate energy-%np fat 0.71)=100. Simple or multiple linear regression analyses were deployed to clarify associations. Statistical analyses were performed with STATGRAPHICS (Graphic Software Systems, Rockville, MD) and SPSS 7.5 for Windows (SPSS Inc, Chicago, IL). 871

4 872 Table 2 Energy expenditure (EE) and balance for each day in the two study groups. Mean s.e.m. Day 1 a 2 b 3 b 4 b EE (kj) Controls FO P Intake of metabolizable Controls energy (kj) FO Energy balance (kj) Controls P FO P FO=formerly obese. a Medium fat diet; b high fat diet. Energy balances are calculated as metabolizable energy minus respective oxidation. P values refer to differences between study groups in energy expenditure and to differences from zero in the energy balance gures tested by one way ANOVA and paired t-test, respectively. The slightly lower calculated metabolizable energy of the high fat diet was due to the nding of a lower gross energy content when the diet was analysed by adiabatic bomb calorimetry. Results Effect of diet on EE and RQnp The formerly obese subjects tended to have a lower 24 h EE, compared to the never-obese group on each of the four days (Table 2), but this difference did not reach signi cance even when the four days were analysed pooled by repeated measures ANOVA (P ˆ 0.12). Furthermore, sleeping EE also failed to demonstrate a signi cant difference between the two groups (P ˆ 0.11 across the four nights, by repeated measures ANOVA). EE declined by 183 kj from day 1 to day 4 in the total group (P < 0.01). When the groups were analysed separately, this decline was only signi cant in the formerly obese group (248 kj, P ˆ 0.02). However, no day x group interaction was found in 24 h or night EE. Energy balances were negative on each day for the control group. In the formerely obese women, the differences between 24 h EE and metabolizable energy, did not reach signi cance because they had a lower average EE (Table 2). The negative energy balance was attributable to an apparent overestimation of the metabolizable energy by the dietary assessment software on which we based the diet preparations. Energy balances did not change across days in any of the two groups (Table 2). On day 1, 23 h RQnp did not correlate with the rate of body weight changes (g/d) from the screening examination to day 1 (R 2 < 0.01, P ˆ 0.89). Twenty three-hour RQnp correlated strongly with energy balance of the actual day on all four days (R 2 > 0.35, P < 0.02). On day 1, 23 h RQnp adjusted to zero energy balance, was signi cantly lower than FQnp in both groups. Oxidative autoregulation did not respond completely to the shift in dietary macronutrient composition on day 2, as adjusted 23 h RQnp was signi cantly higher than FQnp in both groups on that day. However, no signi cant difference between adjusted 23 h RQnp and FQnp was found on day 4 (Figure 1). Figure 1 Twenty-three-hour non-protein respiratory quotient (RQnp), measured in a respiration chamber, of the whole experimental group during consumption of a medium fat diet (day 1) and after the change to a high fat diet (day 2, day 3 and day 4). Solid line: unadjusted data. Broken line: after adjustment to zero energy balance. Dotted line: non-protein food quotient (FQnp) of the consumed diet. *P < 0.05, for difference between RQnp and FQnp. Error bars indicate s.e.m. A repeated measures ANOVA across all four days, showed no differences in RQnp between the groups when energy balance was included as a co-variable. This analysis showed a signi cant effect of day but no groupday interaction on RQnp, suggesting that the dietary shift similarly affected oxidative substrate partitioning in the two groups. Twenty three-hour pro les of RQnp for each group (Figure 2) showed a greater increase induced by the largest meal (supper at h) in the formerly obese subjects, than in the controls, during the high fat diet. Comparing the post-prandial increase in RQnp, de ned as the difference between average RQnp

5 (between 18.00±21.00 h) and baseline RQnp (between 17.00±18.00 h) by t-tests showed a signi cantly greater increase in the formerly obese subjects than the controls, for all the three days on the high fat diet, but not on the medium fat diet (Table 3). The difference between the groups remained signi cant on day 2 and day 4, after including energy balance of the actual day as a covariate. Blood analyses, heart rate and spontaneous physical activity No correlations were found between FFM and the last T 3 =T 4 or 23 h EE measurement (R 2 ˆ 0.02, P ˆ 0.95 and R 2 ˆ 0.04, P ˆ 0.47, respectively). A decrease in free T 3 =T 4 from day 2 to day 5, was found in the formerly obese women (P < 0.005), whereas no such change was observed in the control group. The changes differed between the two groups (P < 0.05) and a daygroup interaction was detected for free T 3 / T 4 ratio and free T 3 Fasting plasma adrenaline and noradrenaline concentrations on day 5 and FFM, did not correlate with the 3 d EE on the high fat diet (R 2 < 0.05). Three-day EE on the high fat diet, correlated positively with free T 3 =T 4 ratio on day 5 in the whole group (Figure 3) although it did not correlate with the respective free T 3 or T 4 concentrations. T 3 =T 4 ratio on day 5, also correlated with sleeping EE during the high fat diet (R 2 ˆ 0.38, P ˆ 0.01). A relationship between free T 3 =T 4 and EE was further supported by the nding of a positive association for the whole group in day 1 to day 4 changes in sleeping EE and day 2 to day 5 changes in free T 3 =T 4 (R 2 ˆ 0.36, P ˆ 0.02). Free T 3 and free T 3 =T 4 were lower in the formerly obese women, compared to control subjects, after the 3 d high fat diet period, whereas no signi cant differences were seen after the 2 d medium fat diet (Table 4). An analysis of covariance, demonstrated that day 5 free T 3 =T 4 remained as a determinant of day 4 EE (P ˆ 0.02), after the group effect was accounted for as a class variable. Stepwise regression analysis including energy balance, group membership and smoking, was performed to test whether relationships existed between the fasting hormones measured on day 5 and day 4 RQnp. Amongst insulin, adrenaline, noradrenaline and free T 3 =T 4 insulin was the only hormone which demonstrated a signi cant relationship with day 4 RQnp. Subsequent simple regression analysis demonstrated a negative relationship between day 5 plasma insulin level and day 4 RQnp, adjusted for differences in energy balance on that day (R 2 ˆ 0.30, P ˆ 0.03). Day 4 RQnp showed a positive association with energy balance in this analysis, whereas group membership had no effect (P ˆ 0.42). Simple regression analyses showed a negative correlation between day 4 RQnp and day 5 plasma free fatty acid concentration (R 2 ˆ 0.43, P ˆ 0.01), whereas the trend of a negative relationship between day 4 RQnp and day 5 plasma triglyceride level, did not reach statistical signi cance (R 2 ˆ 0.13, P ˆ 0.09). Day 5 plasma free fatty acid concentration, was closely negatively correlated with the 3 d energy balance (R 2 ˆ 0.84, P < ). Fasting plasma glucose concentration, declined from the medium to the high fat diet, without any signi cant difference between the two groups (Table 5). Plasma insulin concentrations were not signi cantly affected by the change in dietary composition (Table 4). Plasma triglyceride concentration declined signi cantly from the medium fat to the high fat diet and tended to be lower in the formerly obese group on both day 2 and day 5 (Table 5). In contrast, the total plasma cholesterol level increased and high density lipoprotein (HDL) to total cholesterol ratio decreased slightly on the high fat diet. The formerly obese subjects had a lower plasma HDL to total plasma cholesterol ratio. 873 Table 3 Baseline and post-supper non-protein respiratory quotient (RQnp) in formerly obese (FO) and control subjects. Mean s.e.m. Baseline Post-supper Increase Day 1 Controls FO P ˆ 0.36 P* ˆ 0.87 Day 2 Controls FO P ˆ 0.01 P* ˆ Day 3 Controls FO P ˆ 0.03 P* ˆ 0.12 Day 4 Controls FO P ˆ 0.04 P* ˆ 0.02 Supper commenced at h. The hour between 17.00± h was applied as baseline RQnp period. Postsupper RQnp was de ned as average RQnp for the 3 h period between 18.00±21.00 h. P values were obtained by comparing the post-prandial increase in RQnp between the groups by two-sided t-test for independent samples. P* values obtained by comparing the post-prandial increase between the groups by one-way ANOVA with inclusion of energy balance of the actual day as covariate.

6 874 Figure 2 Twenty-three-hour pro les of non-protein respiratory quotient (RQnp) for each day in the two study groups, measured in a respiration chamber. Squares: formerly obese. Dots: controls. Error bars indicate s.e.m. Over the four days, average heart rate was lower in formerly obese subjects (formerly obese vs controls daytime: vs min 71, P < 0.05; night-time: vs min 71, P ˆ 0.051). No difference in average heart rate was found between the medium and high fat diets, and no day- group interactions were present in average daytime or sleeping heart rates. No differences were found between the two groups in SPA (formerly obese vs controls daytime day 1: vs , P ˆ 0.72; day 4: vs , P ˆ 0.60). Figure 3 Association between day 4 energy expenditure (EE) MJ/24 h, while consuming the high fat diet and fasting free triiodothyronine (T 3 ) to free thyroxine (T 4 ) concentration ratio in plasma measured the morning after the experiment (day 5). R 2 ˆ 0.34, P ˆ Squares: formerly obese subjects. Dots: controls. Discussion After accounting for the group difference in energy balance, caused by the trend to a lower EE in the

7 Table 4 Fasting plasma hormone concentrations in the two study groups measured at the termination of the medium fat diet (morning day 2) and high fat diet (morning day 5). Mean s.e.m. 875 Day FO (n) Controls (n) P P group P day P groupday Insulin (pmol/l) (7) (8) (8) 87 9 (8) Noradrenaline (nmol/l) (8) (8) 0.16 Adrenaline (pmol/l) (8) (8) 0.23 Free T 4 (pmol/l) (7) (8) (8) (8) Free T 3 (pmol/l) (6) (8) (8) (8) Free T 3 =free T (6) (8) (8) (8) Cortisol (nmol/l) (8) (8) 0.09 FO ˆ formerly obese; T 4 ˆ thyroxine; T 3 ˆ triidothyronine. Some of the plasma values were not obtainable for various technical reasons. P value: differences between groups within each day level tested by unpaired t-test. P group, P day and P groupday are group and day effects and groupday interactions tested by repeated measures ANOVA, respectively. P day values for thyroid hormone variables are not shown because different kits were used to analyse free T 3 and T 4 on the samples taken on the two days. Table 5 Fasting plasma concentrations of metabolites in the two study groups, measured at the termination of the medium fat diet (morning day 2) and high fat diet (morning day 5). Mean s.e.m. Day FO (n) Controls (n) P P group P day P groupday Glucose (mmol/l) (7) (8) (8) (8) Lactate (mmol/l) (7) (8) (8) (8) NEFA (mmol/l) (6) (7) (6) (7) Glycerol (mmol/l) (6) 74 8 (7) (7) 83 6 (7) Triglyceride (mmol/l) (7) (8) (8) (8) < Total cholesterol (mmol/l) (7) (8) (8) (8) HDL cholesterol (mmol/l) (7) (8) (8) (8) HDL-/total-cholesterol (7) (8) (8) (8) NEFA ˆ non-esteri ed fatty acids; HDL ˆ high density lipoprotein. Some of the plasma values were not obtainable for various technical reasons. P value: differences between groups within each day level tested by unpaired t-test. P group, P day and P groupday are group and day effects and groupday interactions tested by repeated measures ANOVA, respectively. formerly obese subjects compared to the control group, similar RQnp and response of RQnp to the shift in dietary macronutrient composition, were found in the two groups. The slight decline in fasting blood glucose concentration after the high fat diet period, probably re ects a reduced hepatic glycogen level and this may be one of the factors mediating the increased relative fat oxidation. In contrast to the similar study performed by Schrauwen et al 8 the present study failed to detect an increase in plasma NEFA concentration with the high fat diet. The increase in RQnp due to the largest meal was higher in the formerly obese subjects when consuming the high fat diet. This is in agreement with a previous report from our department. 24 An exaggerated postprandial insulin action or some deviations in the metabolism of the absorbed fat, could therefore be speculated to be associated with a predisposition to weight gain. Hence, although the lower relative postprandial fat oxidation in the formerly obese group did not, in this study, result in a signi cantly higher 23 h RQnp, it may be hypothesized that factors in uencing postprandial metabolism, contribute to the establishment of a positive fat balance, when an unrestricted diet is consumed. Lower plasma levels of free T 3 and free T 3 =T 4 in the formerly obese women, were observed contemporary with a positive association between free T 3 =T 4 and 24 h EE. This is in agreement with another study based on 28 formerly obese and 28 never obese individuals, that found a lower plasma free T 3 level in the formerly obese subjects, which statistically explained a lower 24 h EE in this group. 17 However, no signi cant group difference in 24 h EE was found in the present study, possibly due to the small number of subjects. Results from studies of larger groups, suggest that thyroid hormone levels within the normal range, may play a role as a determinant of metabolic rate. 18,19 Moreover,

8 876 that a signi cant day x group interaction in free T 3 /T 4 was found in the present study, implies that thyroid hormones may respond differently to a high fat diet in some individuals prone to weight gain (the formerly obese, in contrast to the control subjects, demonstrated a decrease from day 2 to day 5 in free T 3 /T 4 ). Dietary energy restriction, decreases the rate of conversion of T 4 to T 3 20 and plasma concentration of T 3. 20±22 It could therefore be argued that the lower T 3 /T 4 in the formerly obese women, could re ect that these subjects had restricted their energy intake prior to the study. We consider this unlikely, as we followed most of these subjects during the previous two months before the experiment, to verify that they were not dieting. Moreover, lower free T 3 and free T 3 /T 4 ratio were most pronounced and only signi cant after the high fat diet, where energy balance had been less negative in the formerly obese subjects than in the controls. In line with previous studies demonstrating that the adaptation in oxidative substrate partitioning requires some days, 5±8 RQnp adjusted to zero energy balance was signi cantly higher than FQnp on the rst and second day after the introduction of the high fat diet, but it declined to a level identical to FQnp after three days. A study by Schrauwen et al, 8 which imposed a similar acute reduction in FQ through a controlled diet, reported that more than three days were needed for the oxidation to adjust to the new substrate composition. This may be explained by the fact that energy balance was better in Schrauwen's study and that glycogen stores may therefore have been reduced at a slower rate than in the present study. The response in plasma triglyceride to sudden alterations in dietary carbohydrate content, is well documented and explained by an impact on hepatic triglyceride production. 23 The lower fasting plasma triglyceride concentration on the high fat diet, compared to the medium fat diet found in the present study, is therefore in agreement with previous observations. Conclusions The previous nding of a more pronounced postprandial reduction in fat oxidation in formerly obese subjects 24 was supported by the present study. However, earlier observations indicating a lower 24 h fat oxidation in formerly obese subjects, independent of energy balance, were not con rmed in this study, in which diet composition was controlled throughout a ve-day period. Furthermore, the study con rms an association between physiological levels of plasma thyroid hormones and EE. It also implies that thyroid hormones may be involved in the lower metabolic rate in formerly obese subjects observed in some studies. Footnote 1) One of the formerly obese subjects left 85 g cherries at supper on day 1. Acknowledgments We thank the staff of the metabolic unit of the Research Department of Human Nutrition, John Lind, Tina Cuthbertson, Inge Timmermann, Bente Knap, Bente Mathiasen, Lis Kristoffersen, Lene Kromann-Larsen, Inger-Lise Grùnfeldt and Charlotte Kostecki. This study was supported by the Danish Medical Research Council, Grant no and References 1 Larson DE, Ferraro RT, Robetson DS, Ravussin E. Energy metabolism in weight-stable postobese individuals. Am J Clin Nutr 1995; 62: 735± Astrup A, Buemann B, Christensen NJ, Toubro S. Failure to increase lipid oxidation in response to increasing dietary fat content in formerly obese women. Am J Physiol 1994; 266: E592±E Lean MEJ, James WPT. Metabolic effects of isoenergetic nutrient exchange over 24 hours in relation to obesity i women. Int J Obes 1988; 12: 15±27. 4 Abbott WGH, Howard BV, Christin L, Freymond D, Lillioja S, Boyce VL, Anderson TE, Bogardus C, Ravussin E. Shortterm energy balance: relationship with protein, carbohydrate, and fat balances. Am J Physiol 1988; 255: E332±E Hill JO, Peters JC, Reed GW, Schlundt DG, Sharp T, Greene HL. Nutrient balance in humans: effects of diet composition. Am J Clin Nutr 1991; 54: 10±17. 6 Stubbs JR, Harbron CG, Murgatroyd PR, Prentice AM. Covert manipulation of dietary fat and energy density: effect on substrate ux and food intake in men eating ad libitum. Am J Clin Nutr 1995; 62: 316± Verboeket-van de Venne WPHG, Westerterp KR, ten Hoor F. Substrate utilization in man: Effects of dietary fat and carbohydrate. Metabolism 1994; 43: 152 ± Schrauwen P, Lichtenbelt WDM, Saris WHM, Westerterp KR. Changes in fat oxidation in response to a high-fat diet. Am J Clin Nutr 1997; 66: 276± Astrup A, Thorbek G, Lind J, Isaksson B. Prediction of 24-h energy expenditure and its components from physical characteristics and body composition in normal-weight humans. Am J Clin Nutr 1990; 52: 777± Heitman BL. Prediction of body water and fat in adult Danes from measurements of electrical impedance. A validation study. Int J Obes 1990; 14: 789± Bergmayer HV. Academic Press: New York, Methods of Enzymatic Analysis Laurell S, Tibbling G. Colorimetric microdetermination of free fatty acids in plasma. Clin Chim Acta 1967; 16: 57 ± Chernick SS. Determination of glycerol in acylglycerols. In Lowenstein JM (ed). Methods of Enzymology. vol. 14, Academic Press: New York, 1969, pp 627± Giegel JL, Ham AB, Clema W. Manual and semi-automated procedures for measurements of triglyceride in serum. Clin Chem 1975; 21: 1575 (Abstr) 15 Christensen NJ, Vestergaard P, Sùrensen T, Rafaelsen OJ. Cerebrospinal uid adrenaline and noradrenaline in depressed patients. Acta Psychiatr Scand 1980; 61: 178± Brouwer E. Report of sub-committee on constant factors. In Baxter KL (ed). Energy Metabolism. (Proceedings of the 3rd Symposium on Energy Metabolism, EAAP Publ 77). Academic Press: London, 1965, pp 441± 443.

9 17 Astrup A, Buemann B, Toubro S, Ranneries C, Raben A. Low resting metabolic rate in subjects predisposed to obesity: A role for thyroid status. Am J Clin Nutr 1996; 63: 879± Astrup A, Buemann B, Christensen NJ, Madsen J, Gluud C, Bennet P, Svenstrup B. The contribution of body composition, substrates, and hormones to the variability in energy expenditure and substrate utilization in premenopausal women. J Clin Endocrinol Metab 1992; 74: 279 ± Toubro S, Sùrensen TIA, Rùnn B, Christensen NJ, Astrup A. Twenty-four-hour energy expenditure: The role of body composition, thyroid status, sympathetic activity and family membership. J Clin Endocrinol Metab 1996; 81: 2670 ± Eisenstein Z, Hagg S, Vagenakis AG, Fang SL, Ransil B, Burger A, Balsam A, Braverman LE, Ingbar SH. Comments: Effect of starvation on the production and peripheral metabolism of 3,3 0,5-triiodothyronine in euthyriod obese subjects. J Clin Endocrinol Metab 1978; 47: 889± Cavallo E, Armellini F, Zamboni M, Vicentini R, Milani MP, Bosello O. Resting metabolic rate, body composition and thyroid hormones. Short term effects of very low Horm Metab Res 1990; 22: 632± Jung RT, Shetty PS, James WPT. The effect of refeeding after semistarvation on catecholamine and thyroid metabolism. Int J Obes 1980; 4: 95± Ullmann D, Connor WE, Hatcher LF, Connor SL, Flavell DP. Will a high-carbohydrate, low-fat diet lower plasma lipids and lipoproteins without producing hypertriglyceridemia? Arteriosclerosis Thrombosis 1991; 11: 1059 ± Raben A, Andersen HB, Christensen NJ, Madsen J, Holst JJ, Astrup A. Evidence for an abnormal postprandial response to a high-fat meal in women predisposed to obesity. Am J Physiol 1994; 267: E549±E