Low resting metabolic rate in subjects predisposed to obesity: a role for thyroid status13

Transcription

1 Low resting metabolic rate in subjects predisposed to obesity: a role for thyroid status13 Arne Astrup, Benjamin Buemann, S#{248}renToubro, Claudia Ranneries, and Anne Raben ABSTRACT A low resting metabolic rate (RMR) for a given body composition has been identified as a risk factor for weight gain and obesity, and has also been reported in formerly obese individuals with the genetic predisposition for obesity. The possible role of thyroid hormone in low RMR was studied in a large sample of postobese women. RMR was measured by indirect calorimetry in 28 weight-stable postobese women with a family history of obesity (P0 group) and in a control group of 28 nonobese women closely matched for age, fat mass, and fat-free mass. RMR was 8% lower in the P0 than in the control group [i; 95% CI: 5856 (5520, 6214) compared with 6408 kj/d (6096, 6768 kj/d), P < 0.02], and the group difference remained unchanged after fat-free mass and fat mass were adjusted for (552 kj/d, P < 0.015). The P0 group had lower plasma free triiodothyronine [2.4 (1.9, 3.0) compared with 3.4 pmol/l (2.9, 3.9 pmolil), P < 0.011, whereas plasma androstenedione only tended to be lower in the P0 than in the control group. Adjustment for differences in androstenedione did not reduce the difference in RMR, whereas adjustment for differences in plasma free triiodothyronine eliminated the group difference (96 kj/d, P = 0.59). The present study shows that RMR for a given body composition is lower among postobese than among matched never-obese control subjects. Statistically, the lower plasma free triiodothyronine concentrations of the postobese subjects could explain their lower RMRs, but it remains to be established whether these findings are causally related. Am J Clin Nutr 1996;63: KEY WORDS Androgens, obesity, energy metabolism, fat-free mass, weight gain, fatness, resting metabolic rate, risk factors for obesity INTRODUCTION Obesity is a highly prevalent condition, not only in industrialized countries but also in many developing countries. The prevalence of obesity has increased dramatically and its complications are among the leading causes of premature mortality, such as coronary heart disease, stroke, non-insulin-dependent diabetes mellitus, and certain cancers (1). Obesity develops as an interaction between a genetic predisposition and certain environmental factors, such as high-fat diets and a low habitual energy expenditure (EE) caused by low levels of physical activity. There is accumulating evidence for a low habitual EE as an important risk factor for weight gain and obesity (2). Daily EE can be divided into three major components: the resting metabolic rate (RMR), the thermic effect of food, and the energy cost of physical activity. RMR is the major determinant of total EE (3) and is mainly determined by body size and composition, but even after these determinants are controlled for there remains considerable interindividual variation in RMR, which is partly genetically determined (4, 5). A low relative RMR, expressed in relation to fat-free mass (FFM), has been found to be a risk factor for subsequent weight gain. In a prospective study in Pima Indians, Ravussin et al (2) showed that both a low relative RMR and a low relative 24-h EE were risk factors for body weight gain. After 4 y of follow-up the risk of gaining 10 kg was approximately seven times greater in those subjects with the lowest relative RMR (lower tertile) than in those with the highest RMR (higher tertile). The rate of relative 24-h EE was estimated to be responsible for up to 40% of the weight change. Studies in white children have produced concordant results. In a study of 4-5-y-old children, Griffith et al (6) reported that the RMR and estimated total EE of those with an obese parent were 16% and 22% lower, respectively, than those of children whose parents had never been obese. Another line of evidence stems from studies of formerly obese subjects who normalized their body composition by dietary energy restriction. Some studies have found lower RMR in postobese subjects than in matched control subjects (7-1 1), although most have not found any difference (12-20), and a few have even reported a higher RMR in postobese individuals (7, 21). The higher RMR of postobese subjects reported in some studies can be explained by their enhanced susceptibility to carbohydrate-rich diets, which causes an increase in RMR in postobese but not in never-obese control subjects (7, 21, 22). It is unclear, however, whether RMR for a given body composition, when a normal-carbohydrate diet is consumed, is normal or low in postobese subjects with a genetic predisposition for obesity. The conflicting results in the literature on RMR in postobese subjects could be a result of limited sample sizes in most of the published studies. If low I From the Research Department of Human Nutrition and KVL Centre for Food Research, The Royal Veterinary and Agricultural University, Copenhagen. 2 Supported by the Danish Medical Research Council (grant ), The Foundation of P Carl Petersen, and The Danish Research and Development Programme for Food Technology (1991-4). 3 Address reprint requests to A Astrup, Research Department of Human Nutrition, The Royal Veterinary and Agricultural University, Rolighedsvej 30, DK-1958 Frederiksberg. Denmark. Received November 8, Accepted for publication February 12, Am J C/in Nutr l996;63: Printed in USA American Society for Clinical Nutrition 879

2 880 ASTRUP ET AL RMR is a risk factor, it cannot be expected a priori to be found in groups with 6-15 postobese subjects. From studies reported previously and unpublished data we collected data on the RMR of a larger sample of 28 weight-stable, postobese women with a family history of obesity, and of a group of 28 nonobese women matched for age, fat mass, and FFM. To assess the possible role of thyroid hormone and androstenedione in the variability in RMR we analyzed concentrations of free triiodothyronine and androstenedione in plasma samples taken in relation to the RMR measurement and stored frozen until measured. SUBJECTS AND METHODS TABLE 1 Anthropometric data on a postobese and a closely matched never-obese control Twenty-eight healthy postobese women and 28 matched control subjects were studied. They were normal by physical examination and routine biochemical screening. The study comprised, in part, data from six previously published studies on smaller groups of postobese women and pair-matched control subjects (10, Il, 19, 22), and from additional unpublished data from I 6 subjects. In some of these studies assessment of RMR was not the central issue and was not reported previously (10, 22). Included in the joint analysis are studies in which RMR data were available for both postobese and control subjects, and in which both subjects of each matched pair were measured by the same method (either in respiratory chambers or by a ventilated-hood system) and in which blood samples were obtained under identical conditions. Only one measurement of each subject is included. Studies with interventions that may have influenced RMR-such as cold exposure, endurance training, and pharmacologic agents-or with an antecedent diet that was low-fat and high-carbohydrate, which may have caused RMR to increase in the postobese subjects, are not included. Age, body weight, and body composition are given in Table 1. None of the subjects took any medication other than contraceptives, which were taken by seven postobese and eight control subjects. Body weights had been stable (± 3 kg) for a period of between 3 mo and 2 y. All postobese subjects except one had a family history of obesity (at least one obese parent or sibling), whereas none of the control subjects had obese parents. Body weight was measured on a decimal scale (model 707; Seca, Copenhagen). Body composition was estimated by the bioimpedance method using an Animeter (HTS-Engineering Inc. Odense, Denmark). FFM and fat mass were calculated as reported previously (23). group Age (y) Body weight (kg) Fat mass (kg) Fat-free mass (kg) 1i: 95% CIs in between groups. 38(36,41) 38 (35, 41) 63.7 (61.9, 65.5) 62.8 (61.0,64.6) 17.3 (16.0, 18.5) 16.2 (15.0, 17.4) 46.4 (45.2, 47.6) 46.7 (45.5, 47.9) parentheses. There were no significant differences The study protocol was approved by the Municipal Ethical Committee of Frederiksberg and Copenhagen and the study was conducted in accordance with the Helsinki II Declaration. Indirect calorimetry RMR was measured in respiration chambers as part of a 24-h measurement of EE and substrate oxidation rates in 12 pairs of subjects, and by a ventilated-hood system in 16 pairs. All measurements were performed between the 4th and the 14th day after the onset of menstruation in each subject. The two open-circuit respiratory chambers were described in detail previously (23, 24). Typically, the subject stayed in the chamber from 0900 to 0900 the next day, and followed a fixed schedule. RMR was measured on the second morning from 0730 to 0900, with the subjects remaining in bed, relaxed but awake, until the measurements were completed. Subjects had fasted since the evening meal, which was served from 1900 to 2100, and they went to bed at The gas exchange of the subjects was calculated from measurements of oxygen (Magnos 4 G analyzer; Hartman and Braun, Frankfurt, Germany) and carbon dioxide (Ureas 3 G; Hartman and Braun) concentrations at the outlet of the chamber and from air flow measured at the outlet of the chamber. Protein oxidation was estimated from nitrogen excreted in the urine, and RMR was calculated as reported previously (24). The intraindividual CV for repeated measure of RMR by this method is 4-5% (24). For the ventilated-hood RMR measurement, subjects generally arrived at the laboratory between 0600 and 0700, by car, bus, train, or taxi to minimize physical activity. They had fasted from 2200 the evening before. After voiding, the subjects rested in the supine position on a bed and RMR measurements for mm, depending on the protocol, commenced 30 mm after insertion of a catheter into an antecubital or a dorsal hand vein. Blood samples were drawn after the start of the RMR measurement. RMR was measured by an open-circuit computerized ventilated-hood system (25). Briefly, carbon dioxide was measured by a Servomex 1490 infrared analyzer or a Hartmann and Braun Uras 10 P analyzer; oxygen was measured by a Servomex paramagnetic analyzer (model 1 looa; Sussex, United Kingdom). Ventilation of the canopy was measured by a mass flow meter (HFM ; Hastings, Hampton, VA)). RMR was calculated as for the respiratory chambers, assuming a constant protein oxidation during the day. Hormone analyses Venous or arterialized blood was sampled without stasis through the indwelling cannula. The blood was centrifuged for 10 mm at 3000 x g and 4 #{176}C, and the plasma was stored at - 20 #{176}Cuntil analyzed. Free triiodothyronine concentrations were analyzed enzymatically with kits purchased from Serono Diagnostics SA (Coinsins, Switzerland). Androstenedione was analyzed by a radioimmunoassay kit commercially available from Diagnostic System Laboratories (Houston). The precision, expressed as the CV, was 4.3% for intraassay and 6.0% for interassay duplicates with a concentration of 7 nmol/l. No drift in hormone concentrations over the storage time could be found by correlation and covariance analyses of plasma concentrations as a function of storage time. Pairs of postobese and control subjects were analyzed in the same assay. Urinary nitrogen was measured by using a nitrogen analyzer (NA 1500; Carlo Erba Strumentazione, Milan, Italy).

3 LOW RESTING METABOLIC RATE AND OBESITY 881 Statistical analysis Unless otherwise stated all results are given as means and 95% CIs. RMR was adjusted for differences in a covariable by using a linear-regression analysis between RMR (the variable) and the residuals reported. Group comparisons were made by unpaired, two-tailed t tests. These tests and simple-linearregression analyses were performed with STATGRAPHICS software, version 4.2 (Graphic Software Systems Inc, Rockville, MD). Power and sample size were calculated with the SIGMASTAT program (Jandel Scientific, Erkrath, Germany). RESULTS The group of postobese women was closely matched with the subjects of the never-obese control group, and there were no differences between groups with respect to age, body weight, or body composition (Table 1). RMR was 8% lower in postobese than in control subjects [5856 (5520, 6214) compared with 6408 kj/d (6096, 6768 kj/d), P < 0.02], and the group difference remained unchanged after FFM and fat mass (552 kj/d, P < 0.015) were adjusted for. The postobese group had lower plasma free triiodothyronine concentrations than the control group [2.4 (1.9, 3.0) compared with 3.4 pmol/l (2.9, 3.9 pmolll), P < 0.01] (Table 2), although none of the postobese subjects had values below the normal range. There was no significant correlation between the plasma free triiodothyronine concentration and RMR adjusted for differences in FFM and fat mass (r = 0. 17, P = 0.13), but plasma free triiodothyronine was positively associated with adjusted RMR in an analysis of covariance (Table 3). The plasma androstenedione concentration tended to be lower in the postobese than in the control group, but the difference was not significant. There was no correlation between adjusted RMR and plasma androstenedione concentrations (r = , P = 0.33), but plasma androstenedione was a significant covariate in an analysis of covariance (P < 0.02). Adjustment for differences in androstenedione did not reduce the group difference in RMR (477 kj/d, P < 0.04), whereas adjustment for differences in plasma free triiodothyronine eliminated the group difference (96 kj/d, P = 0.59). DISCUSSION As far as we know the present study reports results on the largest group of postobese individuals examined as yet. The results show that a low RMR for a given body composition is a trait found among postobese women with a family history of obesity compared with a closely matched control group. The postobese women expended 540 kj/d less than their never- TABLE 2 Plasma concentrations of free triiodothyronine and androstenedione in postobese women and matched control subjects Free triiodothyronine (pmolll) 2.4 (1.9, 3.0)2 3.4 (2.9, 3.9) Androstenedione (nmol/l) 4.8 (2.8, 6.9) 7.6 (5.5, 9.7) I CIs in parentheses. 2 Significantly different from control group, P < TABLE 3 Resting metabolic rate (RMR) and residuals after adjustment for various determinants in formerly obese women and matched control subjects RMR (kj/d) RMR (unadjusted) 5856 (5520, 6216) 6408 (6096, 6768) 0.02 Residuals FFM -240( ) 312(0,624) 0.02 FFM, fat mass -240 ( ) 312 (0. 624) FFM, fat mass, androstenedione -72 ( ) 408 ( ) 0.04 FFM. fat mass. androstenedione. triiodothyronine 96 (- 192, 384) 192 ( ) 0.59 I The residuals are the deviations from predicted values based on the relation between RMR and the covariate given by simple-linear-regression analysis or by analysis of covariance. FFM. fat-free mass. obese counterparts. It is known that the thermic effect of food in postobese subjects is either normal (19) or reduced (26). So, assuming a similar level of physical activity, they must consume 540 kj/d less energy than their normal counterparts to avoid weight gain. According to a prediction model used to estimate the theoretical effect of a lower energy expenditure on long-term weight gain (27), a difference of 540 kj/d can explain a weight gain of 10 kg. However, when obese, our postobese female subjects weighed kg more than they weighed at the time of the study, which indicates that their lower RMRs explained < 50% of their previous weight gain. This agrees with the finding that other factors, such as an increased susceptibility to overeating on a high-fat diet and low levels of physical activity, are equally important for the development of obesity. Since the early reports by Lean and James (7), Geissler et al (8), and Shah et al (9) that postobese subjects had lower RMRs and 24-h EEs than control subjects, several studies have reassessed this issue, but the results have been conflicting. In Lean and James study, RMR was 10% lower, after adjustment for FFM, in the postobese than in the control group (7). In the study by Geissler et al (8) 24-h EE was 1 5% lower in postobese than in control subjects (8), but we estimated this difference roughly to be only 5% if the lower FFM of the postobese subjects was taken into account by dividing by the size of the FFM. Subsequent studies have found RMR or sleeping EE to be 3-15% lower in postobese than in matched control subjects, but, except for two reports (10, 1 1), the differences in most studies have not been significant (12-20). The assessment of RMR in postobese individuals seems to be confounded by the phenomenon that the consumption of a high-carbohydrate, low-fat diet may increase RMR in postobese subjects so that it equals, or even exceeds, that of control subjects (7, 21, 22). This effect of carbohydrate on RMR was also observed in vegetarians (28). Therefore, the present study excluded measurements preceded by a diet with > 55% of energy as carbohydrate. When reviewing the reports that did not find significantly lower RMRs in postobese than in control subjects, it is striking that all studies show a lower RMR in postobese subjects if possible differences in the size of FFM between the groups were taken into account, although the differences were not

4 882 ASTRUP ET AL significant. Also, most of the studies cited examined groups of only six to nine subjects, and only five of the studies with negative results involved > 10 postobese subjects (12, 16, 18-20). We predict that most of the studies lacked the necessary statistical power to detect differences in RMR of the order of 5%, so it is likely that the lower RMR of postobese mdividuals is real. The small sample size in most of these studies creates the risk that an actual difference may be overlooked (Type II error). To detect a 5% mean difference in RMR at the 5% significance level with 90% power requires two groups of 25 subjects each, assuming an SD of 5% of mean RMR adjusted for FFM. Recently, Goran et al (29) found that RMR was 6% or 210 kj/d lower in children having only one obese parent (P < 0.005), but was not lower in children having two obese parents (29). Although the mechanisms responsible for this phenomenon are unknown, in this context it is interesting that most of the postobese subjects in the present study had only one obese parent. Taken together with the prospective studies from the Phoenix group showing that subjects with a low RMR are more likely to gain weight (see Introduction), it is conceivable that the lower RMR of postobese individuals is a phenotypic expression that precedes their development of obesity. We assessed the role of plasma androstenedione concentrations in the low RMR of the postobese subjects because we (23) and others (30) found that this androgen hormone is a possible determinant of basal metabolic rate and 24-h EE after body size, body composition, and free triiodothyronine are adjusted for. In the present study, plasma androstenedione was a significant covariate of RMR, and adjustment of RMR for differences in plasma androstenedione did not reduce the difference between postobese and control subjects. Consequently, the results do not support any role of this androgen hormone in the cause of the low RMR of the postobese group. We showed previously that free triiodothyronine is a determinant of sleeping EE (23) and of 24-h EE (3 1) adjusted for differences in body size and composition. The present finding of lower concentrations of free triiodothyronine in the postobese than in the control subjects suggests that this thermogenic hormone could contribute to the lower RMR of the postobese subjects. Moreover, the finding that adjustment of RMR for differences in free triiodothyronine eliminated the group difference in RMR supports the notion of a causal relation. This suggests the possibility that both the low triiodothyronine and RMR of the postobese subjects could be due to their being in a hypoenergetic state at the time of examination. However, this contrasts with the fact that they had been weightstable for several months before the measurements. Alternatively, the lower triiodothyronine concentration could be due to the habitual consumption by the postobese subjects of a diet with a lower carbohydrate content than that of the control subjects. That the postobese subjects-probably to avoid weight regain-tend to select a diet lower in fat and higher in carbohydrate than that of control subjects (32) undermines this hypothesis. Whether the low triiodothyronine is causally related to the low RMR of the postobese subjects remains a question for further study. Although the low triiodothyronine concentration may play a role in the low RMR of some postobese individuals, the polygenic nature of obesity makes is unlikely that the low RMR, at least in whites, is caused by a single gene defect. Rather, the cause of the lower RMR should be sought in several defects in regulatory systems, eg, the sympathetic nervous system (33), 3-adrenergic receptors (34), and sodium ATP-ase (35). In conclusion, the present compilation of data on RMR in 28 postobese women and 28 closely matched control subjects shows an 8% lower RMR in the postobese subjects, which may be responsible for a weight gain of 10 kg. The postobese women also had a lower free triiodothyronine concentration, which statistically, could fully explain their lower RMR. This association may be spurious, and evidence of a causal relation awaits further study. A We thank the staff of the metabolic unit of the Research Department of Human Nutrition, John Lind, Tina Cuthbertson, Inge Timmermann, Bente Knap, and Charlotte Nielsen. REFERENCES 1. Kuczmarski Ri, Regal KM. Campbell SM, Johnson, CL. Increasing prevalence of overweight among US adults. The National Health and Nutrition Examination Surveys, 1960 to JAMA 1994:272: Ravussin E, Lillioja 5, Knowler WC, et al. Reduced rate of energy expenditure as a risk factor for body-weight gain. N EngI J Med l988;3 I8: Ravussin E, Bogardus C. Relationship of genetics. age, and physical fitness to daily energy expenditure and fuel utilization. Am J Clin Nutr 1989;49: Bogardus C, Lillioja 5, Ravussin E, et al. Familial dependence of resting metabolic rate. N EngI J Med 1986:315: Bouchard C, Tremblay A. Nadeau A. et al. Genetic effect in resting and exercise metabolic rates. Metabolism 1989:38: Griffith M, Payne PR, Stunkard AJ, Rivers JPW, Cox M. Metabolic rate and physical development in children at risk of obesity. Lancet 1990:336: Lean MEJ, James WVF. Metabolic effects of isoenergetic nutrient exchange over 24 hours in relation to obesity in women. Int J Obes 1988; 12: Geissler CA, Miller DS, Shah M. The daily metabolic rate of the post-obese and the lean. Am J Clin Nutr l987;45:9l Shah M, Miller DS, Geissler CA. Lower metabolic rates of post-obese versus lean women: thermogenesis, basal metabolic rate and genetics. Eur J Clin Nutr l988;42: Buemann B, Astrup A, Christensen NJ, Madsen J. Effect of moderate cold exposure on 24-hour energy expenditure. Similar response in post-obese and non-obese women. Am J Physiol 1992:263:El Toubro 5, Western P. Billow J, et al. Insulin sensitivity in postobese women. Clin Sci 1994;84: Weinsier RL, Nelson KM. Hensrud DD, Darnell BE, Hunter GR, Schutz Y. Metabolic predictors of obesity. Contribution of resting energy expenditure, thermic effect of food, and fuel utilization to four-year weight gain of post-obese and never-obese women. J CIin Invest 1995:95: Goldberg GR, Black AE, Prentice AM, Coward WA. No evidence of lower energy expenditure in post-obese women. Proc Nutr Soc 1991; 50: lo9a(abstr). 14. McNeill 0, Bukkens SGF, Morrison DC, Smith JS. Energy intake and energy expenditure in post-obese women and weight-matched controls. Proc Nutr Soc 1990;49:l4A(abstr). 15. Bukkens SGF, McNeill G, Smith JS, Morrison DC. Postprandial thermogenesis in post-obese women and weight-matched controls. tnt J Obes l99l;15: Nelson KM, Weinsier RL, James LD, Darnell B, Hunter G, Long CL. Effect of weight reduction on resting energy expenditure, substrate utilization, and the thermic effect of food in moderately obese women. Am J Clin Nutr 1992:55:

5 LOW RESTING METABOLIC RATE AND OBESITY de Peuter R, Withers RL, Brinkman M, Tomas FM, Clark DC. No differences in rates of energy expenditure between post-obese women. Int J Obes l992;16: Amatruda JM, Statt MC, Welle SL. Total and resting energy expenditure in obese women reduced to ideal body weight. J Clin Invest l993;92: 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 l994;267:e Larson DE, Ferraro RT, Robertson DS, Ravussin E. Energy metabolism in weight-stable postobese individuals. Am J Clin Nutr l995;62: Astrup A, Buemann B, Christensen NJ, Madsen J. 24-h energy expenditure and sympathetic activity in postobese women consuming a high-carbohydrate diet. Am J Physiol l992;262:e 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 l994;266:e Astrup A, Buemann B, Christensen NJ, et al. 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: Toubro S. Christensen NJ, Astrup A. Reproducibility of 24-h energy expenditure, substrate utilization and spontaneous physical activity in obesity measured in a respiration chamber. tnt J Obes 1995:19: Fellows 1W, MacDonald I. An automated method for the measurement of oxygen consumption and carbon dioxide excretion in man. Clin Phys Physiol Meas l985;6: Astrup A, Andersen T, Christensen NJ, et al. Impaired glucoseinduced thermogenesis and arterial norepinephrine response persist after weight reduction in obese humans. Am J Clin Nutr l990;5l: Weinsier RL, Bracc D, Schutz Y. Predicted effects of small decreases in energy expenditure on weight gain in adult women. Int J Obes 1993; 17: Toth MJ, Poehlman ET. Sympathetic nervous system activity and resting metabolic rate in vegetarians. Metabolism 1994:43: Goran MI, Carpenter WH, McGloin A, Johnson R, Hardin JM, Weinsier RL. Energy expenditure in children of lean and obese parents. Am J Physiol l995;268:e9l Svendsen OL, Hassager C, Christiansen C. Impact of regional and total body composition and hormones on resting energy expenditure in overweight postmenopausal women. Metabolism 1993:42: Toubro 5, S#{248}rensen TIA, R#{248}nnB, Christensen NJ, Astrup A. 24-h energy expenditure: the role of body composition, thyroid status, sympathetic activity and family membership. J Endocrinol Metab (in press). 32. Hansen H, Raben A, Astrup A. The habitual diet of post-obese is lower in fat and higher in carbohydrate compared to matched controls. Int J Obes l995;19:82(abstr P145). 33. Spraul M, Ravussin E, Fontvieille AM, Rising R, Larson DE, Anderson EA. Reduced sympathetic nervous activity. A potential mechanism predisposing to body weight gain. J Clin Invest 1993:92: Walston J, Silver K, Bogardus C, et al. Time of onset of non-insulindependent diabetes mellitus and genetic variation in the f33-adrenergicreceptor gene. N Engl J Med l995;333: D#{233}riaz 0, Dionne F, P#{233}russe L, et al. DNA variation in the genes of the Na,K-adenosine triphosphatase and its relation with resting metabolic rate, respiratory quotient, and body fat. J Clin Invest 1994:93: