The contribution of body composition, substrates, and hormones to the variability in energy expenditure and substrate utilization in premenopausal women.

0021-972x/92/7402-0279$03.00/0 Journal of Clinical Endocrinology and Metabolism Copyright 0 1992 by The Endocrine Society

Vol. 74, No. 2 Printed

The Contribution of Body Composition, Hormones to the Variability in Energy Substrate Utilization in Premenopausal ARNE JOOP

ASTRUP, MADSEN,

BENJAMIN CHRISTIAN

BUEMANN, NIELS JUEL GLUUD, POUL BENNETT,

CHRISTENSEN, AND BIRGIT

Substrates, Expenditure Women

and and

SVENSTRUP

Research Department of Human Nutrition, The Royal Veterinary and Agricultural Uniuersity, Frederiksberg; Department of Internal Medicine and Endocrinology, Herlev Hospital; Medical Department, Division of Hepatology, Hvidoure Hospital, Institute of Medical Physiology C, Panum Institute, University Copenhagen; Hormone Department, The State Serum Institute, Denmark

ABSTRACT. Twenty-four-hour energy expenditure and substrate use were measured by indirect calorimetry in respiration chambers on a fixed physical program and related to body composition and plasma concentrations of various substrates and thermogenic hormones. Fifty premenopausal women with a wide range of body weight were examined in the follicular menstrual phase under weight stable conditions. Most of the variance in the sleeping energy expenditure (82%) was accounted for by two covariates, lean body mass (75%, P < O.OOOl), and fat mass (7%, P < 0.0001). An additional 6% of the variance in sleeping energy expenditure was accounted for by plasma androstenedione concentration (4%, P = 0.0005) and by free T3 index (2%, P = 0.03). Thus physiological variation among individuals in plasma androstenedione concentration may result in a difference in energy expenditure of 908 kJ/day and the corresponding variation in free Tg index may result in a difference between individuals of 594 kJ/day. Fiftyfour percent of the variation in carbohydrate oxidation

in U.S.A.

of

rates was accounted for by 24-h energy balance, and by plasma concentrations of insulin, nonesterified fatty acids, and estradiol. Waist circumference, plasma nonesterified fatty acids, and estradiol concentrations explained 49% of the variance in 24-h lipid oxidation. An obese subgroup of women (n = 27) had significantly higher 24-h energy expenditure, lipid, and carbohydrate oxidation rates than an age-matched normal weight group (n = 16), but the entire group difference in energy expenditure was explained by differences in body composition. We conclude that physiological variations in plasma androstenedione and Ta concentrations contribute to the interindividual variance in energy expenditure of women, and their role is not different in obese women. A positive energy balance and increased insulin action may be mediators of the higher carbohydrate oxidation in obesity, whereas an increased substrate availability seems to bring about the increased lipid oxidation. (J Clin Endocrinol Metab 74: 279-286,1992)

E

sible for the familial effect and the unexplained variation are unknown. Fat mass may contribute to explain a small proportion of the residual variation of REE (2), but it may also be due to heterogeneity of LBM, i.e. differences in amount of tissues with differing specific EE (heat production per mass unit tissue). Another contributing factor to the unexplained variation of REE may exist due to the interindividual differences in circulating levels of thermogenic hormones or sympathetic activity. The present study was undertaken to measure EE in 50 premenopausal women in respiration chambers on a fixed diet and a fixed physical activity program, and to relate it to differences in body composition and to hormones with known thermogenic properties such as thyroid hormones, catecholamines, androgens, GH, and insulin. Additional aims were to study determinants of substrate use and to compare two subgroups from the same cohort, one obese and one age-matched normal weight control group.

NERGY requirements of different individuals of similar size may vary remarkably, but they are reasonably stable in each subject during weight stability provided physical activity is unchanged. Because resting energy expenditure (REE) generally accounts for twothirds or more of the daily energy expenditure, there has increasingly been focused on the factors determining its amount. Bogardus et al. (1) reported that 82% of the variance of REE could be accounted for by differences in lean body mass (LBM), whereas adding sex and age only increased this figure to 83%. Bogardus et al. (1) reported, however, that family membership accounted for an additional ll%, thus 94% of the total variability of REE was explained. The physiological factors responReceived March 21, 1991. Address all correspondence and requests for reprints to: Arne V. Astrup, Research Department of Human Nutrition, The Royal Veteranarian and Agricultural University, Rolighedsvej 25, 1958 Frederiksberg, Copenhagen, Denmark. 279

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 12 October 2015. at 02:46 For personal use only. No other uses without permission. . All rights reserved.

280

ASTRUP

Materials

and Methods

Subjects Fifty premenopausal women (15-49 yr) with a wide range of body weights (49-116 kg) were recruited to participate in the study. To look at the effects of obesity in this cohort, data of a subgroup of obese women [body mass index (BMI) > 30 kg/ m’] were compared with those of an age-matched group of normal weight women (BMI < 25 kg/m*). All subjects but the normal weight ones had a full medical history and physical examination, with routine hematology and biochemistry screens and urine analysis. At the time of the study none of the subjects had evidence of significant disease apart for some of them being obese. None had noninsulin-dependent diabetes mellitus or other endocrine diseases. The physical characteristics and anthropometry of the entire group and of the two subgroups are given in Table 1. Experimental design The study was approved by the Municipal Ethical Committee of Frederiksberg and Copenhagen and all subjects gave informed consent. All measurements of 24-h EE in the respiration chamber were carried out in the follicular phase of the menstrual cycle. Body weight was measured on a decimal scale (Seca model 707, Copenhagen, Denmark), and body fat topography was assessed in the standing position by measurement of the waist-hip circumference ratio (WHR). Body composition was estimated by the bioimpedance method by an Animeter (HTS-Engineering Inc., Odense, Denmark). LBM and fat-mass were calculated by using the equations by Heitmann (3). Diet For at least 4 days before the stay in the respiration chamber the subjects were instructed by our dietitian in a weightmaintenance, conventional diet that supplied 55% of the energy as carbohydrate, 30% as fat, and 15% as protein. During the 24 h spent in the respiratory chamber, the subjects were fed a similar diet with an energy content computed from equations giving the relation between LBM and 24-h EE based on a previous study (4), but corrected for a slight change in the physical activity programme. Returned food was weighed and corrections were made in the calculation of energy and nutrient intakes. EE Twenty-four-hour EE was measured in two open-circuit indirect calorimetry chambers, which have previously been described in details (4). EE and substrate utilizations were calculated as described by Livesey and Elia (5). EE was adjusted for differences in LBM or in both LBM and body fat mass by the normalization method described in details by Ravussin and Bogardus (6). The acute energy balance during the 24 h in the respiration chamber was computed as energy intake - energy expenditure. To ensure optimal adherence to the protocol and for ethical reasons the subjects were kept under 24-h surveillance by medical students. The subjects fasted overnight (>9 h) and abstained from smoking before arriving at the depart-

ET AL.

JCE & M a1992 Vol74*No2

ment at 0800 h. After voiding, body weight, bioimpedance, and circumferences were recorded, and a fasting blood sample was drawn from an antecubital vein when the subject was resting sitting in an armchair. A breakfast meal was served at 0930 h. and the respiratory measurement started at 1100 h and ended 24 h later. A fixed physical activity program was followed including three bicycling sessions of 10 min each (75 watt), and meals were served at 0930, 12:30, 1800 h. The daytime period went from 1100-2300 and from 0900-1100 h the next day. The sleeping EE (SEE) was measured from 2300-0800 h, and after voiding basal EE was measured under standardized conditions from 0800-0900 h. The within-subject coefficient of variation for repeated 24-h measurements were 2.5% for EE and 2.7% for RQ (4). Laboratory analyses Blood was sampled without stasis in ice-cooled syringes and centrifuged at 4 C. Glucose, nonesterified fatty acids (NEFAs), glycerol, triglyceride, immunoreactive insulin, and C-peptide were measured in plasma as previously described (7). For determination of catecholamines blood was collected in tubes containing reduced glutathione and ethylene-glycolbis(aminoethyl-ether)tetra-acetate. The tubes were centrifuged immediately and the plasma was stored at -80 C. Until determination of catecholamines by a radioenzymatic method (8). GH (hGH RIA 100, Pharmacia, Uppsala, Sweden) and cortisol (Farmos Diagnostica, Turku, Finland) were analyzed by RIA kits. Androgens, estradiol, and sex hormone-binding globulin (SHBG) were analyzed as described by Lykkesfeldt et al. (9). All plasma samples were coded and analyzed in a random order to avoid any systematic error attributable to the order of analysis. Coefficients of variation for measurements were as follows: NEFAs, 3.5%; glucose, 1%; glycerol, 5%; insulin, 9%; C-peptide, 11%; catecholamines, 8%; SHBG, 7.5%; androgens, ~14%; GH, 5%; cortisol, 3%; and thyroid hormones, 5%. Statistical analysis Age, body weight, WHR, LBM, and body fat of the two groups were compared by using an unpaired, two-sided t test. All other group comparisons were made by a multifactor analysis of variance including age as covariate. These tests and simple linear and multiple stepwise regression analyses were performed with Statgraphics software (Graphic Software Systems Inc., Rockville, MD). Adjusted correlation coefficients derived from the multiple stepwise regression analysis are denoted as ra. Unless otherwise stated, all results are expressed as means f SEM.

Results Physical

characteristics

and plasma variables

The physical characteristics of the entire group and of the two subgroups are displayed in Table 1. Plasma levels of glucose, insulin, C-peptide, and free T, index were elevated in the obese compared with the normal weight group, whereas GH, cortisol, and norepinephrine were lower in the obese than in the normal weight group


RELATIONSHIP

OF ENERGY

EXPENDITURE

TABLE 1. Physical characteristics of total group of premenopausal women and of two age-matched subgroups of obese and normal weight women Variable Age (yr) Body weight (kg) WHR LBM Body fat (kg)

Total group (n = 50)

Normal weight (n = 16)

Obese (n = 27)

31.0 f 1.3 (15-49) 84.1 f 2.4 (49-116) 0.849 + 0.015

25.1 + 0.8

28.9 + 1.6

61.0 f 1.8

93.2 + 2.4

(0.68-1.04) 49.8 + 0.8 (35-62) 34.5 + 1.6 (12-54)

0.75

* 0.01

0.86

f 0.02

43.1

+ 1.1

52.4

+ 1.0

17.4 + 0.9 (12-24)

40.8 f 1.5 (32-54)

Values are means + SEM (range). The classification of the two groups was based upon BMI: BMI < 25 kg/m2 = normal weight, BMI > 30 kg/m2 = obesity. All differences between normal weight and obese group were highly significant (P < O.OOl), except for age.

(Table 2). Although the age effect on free Ts index was not significant [Analysis of variance (ANOVA), P = 0.081, an inverse correlation was found between free Ts index and age (r = -0.41, P = 0.004). Dehydrotestosterone, androstenedione (AD), testosterone, and SHBG were lower in obese than in normal weight women (Table 3). Determinants

of EE

TO BODY

COMPOSITION

281

hormones were apparantly positively associated with SEE/basal EE (BEE) (insulin, free Tq index, free testosterone, and dehydroepiandrosterone (DHEA), whereas others were inversely associated (GH, cortisol, dehydrotestosterone, and SHBG). Most of their effect on EE, however, was found to be exerted through covariation with body size (either body weight, LBM, or fat mass). Body weight explained 77% and LBM 75% of the variation in SEE (P < 0.05). Stepwise multiple regression analysis showed, however, that body composition explained 82% of the variation of SEE, as LBM accounted for 75% (P < 0.0001) and body fat mass for an additional 7% (P < 0.0001). The concentration of AD further accounted for 4% (P = 0.0005), and the free Ts index for another 2% (P = 0.03). This means that these four covariates accounted for 88% of the variance in sleeping EE (r = 0.94, P < 0.0001). The residual standard deviation of SEE after adjusting for the four covariates was 380 kJ/day or 4.2% of the mean value. After entering the stepwise multiple regression analysis all the hormones that were significantly correlated to SEE or BEE in the simple analysis (Table 4) lost their apparant effect. By contrast, both AD and free T3 index were significantly correlated to SEE after adjustment for body composition. The prediction equation for SEE had no intercept with the abscissa (P = 0.66). Full equation: SEE (kJ/h) = 3.8 LBM (kg) + 2.1 body fat (kg) + 3.9 AD (nmol/L) + 16.5 free T3 index. Based on the ranges of plasma AD concentration and of free Ts index given in Table 3, it can be estimated that variation in AD

When the concentrations hormones were correlated

of the potential thermogenic with EE (Table 4), several

TABLE 2. Plasma concentrations weight and obese women

of glucose and hormones in total group of premenopausal women, and of two age-matched subgroups of normal

Variable

Total group (n = 50) 4.81 + 0.06

Normal weight (n = 16) 4.47 f 0.09

Glucose (mmol/L) (3.7-6.0) Insulin 89 f 6 (14-215) bmol/L)” C-Peptide 590*40 (250-1905) (pmol/L) GH 1.6 + 0.4 (0.1-12.8) WL) Cortisol 333k 26 (nmol/L) (99-1215) Epinephrine 93 + 11 (O-490) bmol/L) Norepinephrine 1.77 + 0.11 (nmol/L) (0.11-3.66) Free T, index 93 f 2 (arb. U) (55-141) Free TB index 1.81 f 0.04 (arb. U) (1.27-2.77) Estradiol 221 f 19 (77-770) (pmol/L) Values are means & SEM (range). Differences between a covariable. NS denotes nonsignificant P values > 0.10. a The conversion factor for insulin concentration from

65 414+

Obese (n = 27) 4.97 + 0.09

Obese vs. normal weight

99 2 9

0.009

NS

0.005

NS

0.0001

NS

0.018

NS

+9 30

714+

4.2

f 1.0

0.3

433

A 60

281+

93 + 16 1.78 + 0.13 85

+ 4

1.89 f 0.09 209+ 30

87

71

f 0.1 31

+ 16

1.62 f 0.11

0.003

NS

P 0.10. Variable

TABLE 4. Correlation matrix between EE and fasting plasma concentrations of various hormones and metabolites in the follicular phase of premenopausal women (n = 50) Variable Insulin GH Cortisol Epinephrine Norepinephrine Free TI index Free T, index Dehydrotestosterone AD Testosterone Free testosterone DHEA-S SHBG o P c 0.0001. b P < 0.01. c P < 0.05.

SEE (4 0.55” -0.41* -0.34’ -0.09 -0.09 -0.04 0.30’ -0.36* -0.12 -0.09 0.28’ 0.25 -0.35*

BEE (r) 0.47b -0.3gb -0.37b -0.14 -0.08 0.01 0.25 -0.25 -0.15 -0.13 0.20 0.38* -0.30’

among individuals may result in a difference in SEE of 908 kJ/day, and the corresponding variation among our individuals in free Ts index may result in a difference between individuals in SEE of 594 kJ/day. When the stepwise regression analyses were performed with 24-h EE or BEE as dependent variables, less of the total variability could be explained by the covariates, and the contribution of free T3 index did not reach statistical significance. In addition, plasma concentration of sulfated DHEA (DHEAS) replaced AD as a significant determinant of EE. LBM accounted for 75% (P < O.OOOl), fat mass for 3% (P = 0.003), and DHEAS for 3% (P = 0.02), so together they explained 81% (P < 0.0001) of the variation in 24-h EE (residual SD = 569 kJ/d). Only 68% (P < 0.0001) of the variance in basal EE was explained by LBM (60%, P = 0.001) and by

Age as covariate NS NS NS NS NS NS ANOVA with age as

DHEAS concentration (8%, P = 0.002), giving a residual SD of 743 kJ/day. It should be noted that AD and DHEAS were positively correlated (r = 0.34, P = 0.02). Determinants

of substrate oxidations

By simple linear correlation analysis, 24-h carbohydrate oxidation was best correlated with fasting plasma insulin concentration (r = 0.61, P < 0.0001, Fig. 1). The 24-h carbohydrate oxidation also correlated with acute energy balance [(24-h energy intake - 24-h energy expenditure); r = 0.55, P < O.OOl]. The 24-h carbohydrate oxidation correlated negatively with plasma NEFA concentration (r = -0.48, P < 0.001). In multiple linear regression analysis a total of 54% of the variance (P < 0.0001) in 24-h carbohydrate oxidation was explained by plasma insulin concentration (37%, r’a = 0.61) together with plasma NEFA concentration (8% ra = -0.36), acute energy balance (5%, ra = 0.32) and plasma estradiol concentration (4%, ra = 0.29). Twenty four-h lipid oxidation had a positive correlation with both body fat mass (r = 0.54, P < O.OOOOl), WHR (r = 0.46, P < 0.0001) and with waist circumference (r = 0.59, P < 0.00001, Fig. 1). No significant correlations were found between 24-h lipid oxidation and acute energy balance or plasma concentrations of NEFA, triglyceride, insulin, or estradiol. In the multiple stepwise regression analysis, however, waist circumference (35%, ra = 0.59), plasma NEFA concentration (8%, r, = 0.35), and plasma estradiol concentration (6%, ra = -0.32) together explained 49% of the variance in 24-h lipid oxidation. LBM and fat mass did not correlate significantly with 24-h carbohydrate oxidation or with lipid oxidation in the multiple regression analysis. Twenty four-h protein oxidation was correlated solely to LBM (r = 0.62, P = 0.0001).


RELATIONSHIP 24-hWdQdbLbn 100

OF ENERGY

w

EXPENDITURE

Variable EE (kJ/day) E&LBM>

40 . .* 1

COMPOSITION

283

TABLE 5. Twenty four-h EE and substrate uses before and after adjustments for differences in body composition

. .

1

TO BODY

EE(LBM+FM)

Lipid oxidation k/day) Carbohydrate oxidation (g/day) Protein oxidation k/day) Protein oxida-

r-oB@Pcaaml

20

Normal weight (n = 16) 8422 8769 9402 83.1

& 198 + 158 2 165 -c 6.3

Obese (n = 27) 10097 9664 9449 109.9

+ + f +

208 94 84 3.5

P value

caffeine mixture on energy expenditure and body composition in obese women.

Resting energy expenditure and body composition in rural Sarawaki adults.

Energy intake, energy expenditure, and body composition of poor rural Philippine women throughout the first 6 mo of lactation.

The relative contribution of androgens and insulin in determining abdominal body fat distribution in premenopausal women.

Lower resting and total energy expenditure in postmenopausal compared with premenopausal women matched for abdominal obesity.

Re-examination of the contribution of substrates to energy expenditure during high-intensity intermittent exercise in endurance athletes.

Critical review of beige adipocyte thermogenic activation and contribution to whole-body energy expenditure.

Oxygen consumption, energy expenditure, and substrate utilization after cardiac surgery in children.

Genotype by energy expenditure interaction and body composition traits: The Portuguese Healthy Family Study.

Energy expenditure and body composition changes after an isocaloric ketogenic diet in overweight and obese men.

Sexual activity, endogenous reproductive hormones and ovulation in premenopausal women.

Influence of dietary protein level on body composition and energy expenditure in calorically restricted overweight cats.

Body composition and resting energy expenditure in long term spinal cord injury.

Changes in body composition and energy expenditure after six weeks' growth hormone treatment.

Relationship between submaximal oxygen uptake, detailed body composition, and resting energy expenditure in overweight subjects.

Exercise, energy expenditure, and body composition in people with spinal cord injury.

Body composition and energy expenditure in adolescents with cerebral palsy or myelodysplasia.

The effects of amphetamine on body weight and energy expenditure.

Effects of rapid weight loss and regain on body composition and energy expenditure.

Effect of dietary fatty acid composition on substrate utilization and body weight maintenance in humans.

Gut hormones such as amylin and GLP-1 in the control of eating and energy expenditure.

Effects of oral testosterone undecanoate on growth, body composition, strength and energy expenditure of adolescent boys.

Lean-body-mass composition and resting energy expenditure before and after long-term overfeeding.

Correction: Energy Expenditure and Substrate Oxidation in Response to Side-Alternating Whole Body Vibration across Three Commonly-Used Vibration Frequencies.