Circulating Insulin Concentrations, Smoking, and Alcohol Intake Are Important Independent Predictors of Leptin in Young Healthy Men

Transcription

1 Circulating Insulin Concentrations, Smoking, and Alcohol Intake Are Important Independent Predictors of Leptin in Young Healthy Men Christos S. Mantzoros, * Antonios D. Liolios, t Nicholas A. Tritos.i Virginia G. Kaklamani, * Dimitrios E. Doulgerakis,t Ioannis Griveas,t Alan C. Moses, * Jeffrey S. Flier* Abstract MANTZOROS, CHRISTOS S., ANTONIOS D. LIOLIOS, NICHOLAS A. TRITOS, VIRGINIA G. KAKLAMANI, DIMITRIOS E. DOULGERAKIS, IOANNIS GRIVEAS, ALAN C. MOSES, JEFFREY S. FLIER. Circulating insulin concentrations, smoking, and alcohol intake are important independent predictors of leptin in young healthy men. Obes Res. 1998;6: Objective: Leptin, an adipocyte-secreted hormone, has been shown to signal the status of energy stores to the brain, regulate energy homeostasis, and mediate the neuroendocrine response to food deprivation. Obesity is associated with increased leptin levels, and several hormones, including insulin and glucocorticoids, have been associated with leptin levels and expression in rodents. Although obesity has been strongly associated with increased leptin in humans, a significant percentage of leptin's variability remains unexplained. The role of endogenous hormones, demographic factors, or certain life-style factors in explaining the residual variability of leptin levels has not yet been clarified. We performed this cross-sectional study to document the relative importance of obesity, lifestyle factor, and endogenous hormones in determining serum leptin levels. Research Methods and Procedures: We measured serum concentrations of insulin, cortisol, testosterone, growth hormone, and dehydroepiandrosterone sulfate; ascertained anthropometric, demographic, and lifestyle characteristics; and studied these variables in relationship to serum leptin concentrations in a sample of young healthy men. Results: Obesity and alcohol intake were independently and positively associated with circulating leptin concentrations. Submitted for publication December Accepted for publication January 16, From the *Division of Endocrinology, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA; [Medical Corps, Greek Army; and :j:joslindiabetes Center, Boston, MA Reprint request to J.S. Flier, Division of Endocrinology. RN 324. Beth Israel Deaconess Medical Center. Harvard Medical School. 330 Brookline Avenue. Boston, MA Copyright 1998 NAASO. Additionally, cigarette smoking was negatively and independently associated with leptin concentrations. Finally, serum insulin concentration was an independent hormonal determinant of circulating leptin concentrations, whereas serum testosterone was negatively associated with leptin only by bivariate analysis. Discussion: We conclude that, in addition to obesity, cigarette smoking, alcohol intake, and serum insulin levels are associated with leptin levels in a population of healthy young men. Key words: leptin, obesity, insulin, smoking, alcohol, cortisol, growth hormone Introduction Leptin, the product of the ob gene (I), is an adipocytesecreted protein, the serum levels of which in the fed state reflect the amount of energy stored in adipose tissue of rodents and humans (2-5). In rodents, leptin acts in the hypothalamus to decrease food intake and increase energy expenditure (6-9), and genetic defects that impair either leptin production or action cause severe obesity (1,4). In addition to its role in regulating energy homeostasis, leptin also appears to play an important role in regulating the neuroendocrine axis (10-12) in rodents. Although leptin levels in the fed state correlate strongly with adiposity, individuals with the same level of adiposity may vary markedly in leptin levels (4,5). Thus, an understanding of the factors that influence leptin levels is highly desirable. Such factors could be endogenous or related to environmental influences. One important environmental factor that is capable of influencing the regulation of energy balance in humans is cigarette smoking (13). Smoking increases endogenous catecholamine levels (14,15), stimulates energy expenditure (16,17) and thermogenesis (13,18), and decreases body weight (19,20). Although smoking might influence energy balance, thermogenesis, and body OBESITY RESEARCH Vol. 6 NO.3 May

2 weight by increasing circulating leptin concentrations, it is also possible that smoking, via activation of the adrenergic system (14,15), could suppress leptin levels and decrease body weight via a mechanism that operates independent of leptin. A possible relationship between cigarette smoking and the leptin axis has not yet been explored in humans. Furthermore, insulin (4,21-23), glucocorticoids (4,24-26), and adrenergic agonists (4,26-29) are additional factors that can regulate leptin secretion both in vitro and in vivo in animals. However, the experimental evidence for the role of insulin in regulating leptin mrna expression or circulating levels in humans remains controversial (4,23,30,31), and studies on the role of endogenous glucocorticoids have revealed conflicting results (4,32-34). Finally, we are aware of no previous studies that have jointly evaluated anthropometric, demographic, hormonal, and lifestyle factors, such as smoking or alcohol intake, as predictors of circulating leptin concentrations in humans. For those reasons, we have studied young healthy men to explore whether lifestyle factors or circulating hormonal levels are independent determinants of circulating leptin concentrations in humans. Subjects and Methods Participants in this study were 82 consecutively enrolled young men aged 18 to 28 who consented to having a single blood sample drawn during a routine physical examination between 9:00 and II :00 AM (35). Subjects were healthy and were taking no medication. The study was conducted according to the principles expressed in the "Declaration of Helsinki." The participants responded to a questionnaire administered by an interviewer that covered basic demographic and lifestyle variables. Specifically, the participants were asked to indicate their age in completed years, area of residence, recorded as a trichotomous variable (0: village or town<lo,ooo residents, I: city 10,000 to 100,000 residents, 2: major urban center> 100,000 residents), and educational level in completed years of schooling. In addition, subjects were asked to indicate whether they were smokers and, if so, the daily number of cigarettes smoked. Information was also provided on the amount of coffee drunk in cups per week and usual alcohol intake in glasses per week. Greek coffee was the type consumed by nearly all subjects, whereas the amount of alcohol consumed in the standard measures of most drinks is about the same for wine, beer, and spirits (36). Thus, coffee intake was measured in cups per week, and alcohol consumption was measured in drinks per week as previously described (37,38). Regular participation in sports (yes/no) and the number of hours of regular weekly exercise were recorded, but no attempt was made to weight it according to intensity (37,38). Height was measured using the "extended neck" technique, and measurements were rounded to the closest centimeter. Weight was measured using the same scales for all subjects, and mea- surements were also rounded to the closest III 0 of a kilogram. Quetelet's index (body mass index, BMI) was calculated as weight/height'. Blood samples for the hormone determinations were centrifuged immediately, and serum was frozen at -34 C until determination. Serum hormone concentrations were determined by commercially available radioimmunoassay kits (Linco Research for leptin, ICN ImmuChem Coated Tubes' for insulin and cortisol, ICN ImmuChem Double Antibody'" for dehydroepiandrosterone sulfate [DHEAS] and growth hormone [GH], and DPC Coat-A-Count for testosterone). The sensitivity of the assays was as follows: 0.31 ng/ml for GH, 0.5 ng/ml for leptin, 0.15 p.g/dl, for cortisol, 10 ng/ml for DHEAS, 4 ng/dl for testosterone, and 2.75 flu/ml for insulin. The intra-assay coefficients of variation in the range of values measured were 8% to 8.8% for DHEAS, 4.22% to 5.39% for insulin, 5.8% to 7% for cortisol, 4.1% to 5.8% for GH, for testosterone and 4.4% to 5.7% for leptin (39,40). Statistical analysis was performed by modeling levels of leptin as a function of demographic, lifestyle, and endocrine variables. Because leptin levels were not normally distributed, we used logarithmic transformation of leptin concentrations to normalize the distribution for statistical analysis, reexponentiating for data representation in the text (41). Results Baseline Data Table I shows representative values of the demographic, lifestyle, and endocrine variables. The mean age of the participants was 22 years (standard deviation, 3 years), and 60% (n = 49) were current smokers. Leptin concentrations were skewed to the right, with a mean of 7.12 ng/ml, a median of 4.89 ng/ml, and a standard deviation of 6.84 ng/ml. Insulin concentrations were also skewed to the right with a mean of flu/ml, a median of flu/ml, and a standard deviation of u.uzrnl.. Simple Linear Regression Analysis The simple regression coefficients of leptin concentrations regressed on each of the demographic, anthropometric, lifestyle (upper part), and endocrine variables (lower part) are seen in Table 1. Weight (p<0.0001), height (p = 0.04), and alcohol intake (p = 0.05) are the anthropometric and lifestyle factors that are positively associated with circulating leptin levels by simple regression analysis. Additionally, serum insulin (p = 0.0 I) and testosterone (p = 0.05) are the only hormones associated with serum leptin by simple regression analysis. Exploratory Multiple LinearRegression Because the studied variables are interrelated, we performed multiple regression analysis to control for confounding factors. In an initial exploratory multiple regression 180 OBESITY RESEARCH Vol. 6 No.3 May 1998

3 Table 1. Simple and multiple regression coefficients of leptin regressed on anthropometric, lifestyle, demographic (upper), and endocrine variables (lower) Bivariate Bivariate regression Pearson Multiple Partial Mean coefficients correlation regression correlation Variables (standard (standard coefficient coefficients coefficient Two-tailed (units) deviation) error) (r) (95% CI) (r) p value Anthropometric and lifestyle variables Weight * < (kg) (14.1) (0.006) (0.026, 0.056) Height * (em) (6.7) (0.013) (-0.042, 0.006) Age (years) (3) (0.036) (-0.129,0.011) Education (years) (3) (0.032) (-0.018,0.104) Smoking (pack/day) (0.68) (0.133) (-0.445, 0.037) Alcohol * (drinks/week) (17) (0.005) (0.006, 0.034) Coffee (cups/week) (9) (0.01) (-0.020,0.016) Exercise (hours/week) (7.6) (0.011) (-0.026,0.012) Area of residence (0.126) (-0.404, 0.068) Endocrine variables Insulin * (I-1IU/mL) (15.10) (0.007) (0.006, 0.040) DHEAS (ng/ml) (270.65) (0.0003) (-0.001, 0.001) Cortis (l-1g/dl) (7.60) (0.012) (-0.022, 0.026) GH (ng/ml) (2.75) (0.033) (-0.053,0.079) testosterone * (ng/dl) (156.92) (0.0005) (-0.001,0.001) *p<o.05 by simple ingression analysis. analysis model, we regressed leptin separately on all of the demographic and lifestyle variables (Table 1, upper) and on all of the endocrine variables (Table 1, lower). Insulin (p = 0.008) was an independent hormonal determinant of leptin levels. Body weight (p< ) and alcohol intake (p = 0.007) were also independently and significantly associated with serum leptin levels, whereas the associations between smoking or age and leptin levels achieved marginal statistical significance (p<0.1). Thus, this analysis indicated that the above variables could be the important predictors of leptin levels to be entered in the final multiple regression analysis model. Additionally, we found that there was no colinearity problem in the statistical analysis. The highest value of the Pearson correlation coefficient between any two of the variables studied was for weight with height. OBESITY RESEARCH Vol. 6 NO.3 May

4 Multiple Linear Regression Analysis Model To build the multiple regression analysis model, we used stepwise forward selection (41). The model was evaluated at each step, and the process continued until adjusted R 2 was increased most, i.e., none of the remaining variables explained any amount of additional variability, as reflected by adjusted R 2 (41). The final multiple regression analysis model is presented in Table 2, where leptin is regressed on insulin, age, body weight, height, alcohol, and smoking. In fact, these were the variables that seemed to be the important predictors of leptin concentrations on the basis of the models presented in Table I. The corresponding adjusted R 2 was higher in this model than in the above exploratory models. Insulin concentrations (p = 0.04), body weight (p<0.000l), smoking (p = 0.03), and alcohol intake (p =0.03) are significant independent predictors of serum leptin concentrations in this model. The values of the Pearson correlation coefficients between leptin, on one hand, and insulin concentrations, body weight, smoking, and alcohol intake, on the other, were 0.20, 0.63, -0.20, and 0.21, respectively. Importantly, these associations were particularly robust. In a similar model, regressing leptin concentrations on the same variables and BMI instead of weight generated essentially identical results; insulin concentrations (p = 0.035), BMI (p<0.000i), smoking (p = 0.037), and alcohol intake (p = 0.027) remained the only significant predictors of leptin concentrations. An increase in the mean insulin concentrations by 7.33 flu/ml (1/2 SD) corresponds to an increase of circulating leptin concentrations by 1.22 ng/ml, and conceivably more, depending on the extent of biologically generated variation and consequent misclassification. Additionally, an increase of body weight by 3.0 kg, an increase of alcohol intake by one drink per day, and a decrease in smoking by 0.50 pack per day correspond to a similar increase of circulating leptin by approximately 1.11 ng/ml. By contrast, concentrations of cortisol, GH, testosterone, and DHEAS within the normal range do not appear to playa role of comparable significance. Body weight or BMI considered alone explained 35% of leptin variability as reflected by R 2 Additionally, insulin levels, smoking, and alcohol consumption explained 49% of leptin variability when considered separately from weight or BMI. However, when all of the variables included in the final multiple regression analysis model (Table 2) were considered together, the explained leptin variability increased further to 65%. Thus, the leptin variability explained jointly by obesity, insulin, smoking, and alcohol consumption is significantly increased. However, it appears that leptin variability does not increase in an additive manner, probably because of significant interactions between the study variables. Finally, the residual leptin variability (35%) that remains unaccounted for after modeling the study variables implies that, in addition to imprecision of measurements, other hormonal or environmental factors could also play an important role. Table 2. Simple and multiple regression coefficients of leptin regressed on insulin, alcohol intake, smoking, age, height, and body weight Bivariate Bivariate regression Pearson Partial Mean coefficients correlation Multivariate correlation Variable (standard (standard coefficient coefficients coefficient Two-tailed (units) deviation) error) (r) (95% CI) (r) p value Insulin (flid/ml) (15.10) (0.007) (0.001,0.025) Alcohol (drinks/week) (17) (0.005) (0.002, 0.028) Age (years) (3) (0.036) (-0.076,0.042) Smoking (pack/day) (0.68) (0.133) (-0.434, ) Height (em) (6.7) (0.013) (-0.036,0.014) Weight < (kg) (14.1) (0.006) (0.024, 0.052) 182 OBESITY RESEARCH Vol. 6 NO.3 May 1998

5 Discussion Leptin levels in the fed state correlate strongly with the magnitude of adipose stores in both humans and rodents (4,5). However, individuals with the same level of adiposity may vary markedly in leptin levels (4,5). Thus, it seems that other factors that influence leptin levels could explain a significant degree of leptin variability. Such factors could be endogenous or related to environmental influences. A previous study has shown that in addition to obesity, age and gender could influence leptin variability (30). Another important environmental factor that is capable of influencing the regulation of energy balance (16,17) and body weight (19,20) in humans, and could potentially influence leptin levels, is cigarette smoking (13). Furthermore, hormones such as insulin (4,21-23), glucocorticoids (4,24-26,42), and adrenergic agonists (4,26-29) are additional factors that can regulate leptin secretion in vitro and in animal studies, although their role in humans has not been elucidated. This study has addressed the possible interrelationship between leptin on one hand and hormonal, demographic, and lifestyle factors on the other. Given the important actions of leptin, hormonal factors, and cigarette smoking on energy balance, their interactions in humans are a matter of clinical significance. At the time of this study, there was equivocal evidence for the role of insulin and cortisol and virtually no evidence concerning the potential role of circulating GH, testosterone, and DHEAS in determining circulating leptin concentrations in normal men. Finally, no previous study had evaluated the potential role of demographic or lifestyle factors or had performed joint evaluation of the above endocrine variables as predictors of circulating leptin concentrations in normal men after controlling for potential confounding variables. Our study suggests that circulating insulin concentrations within the normal range influence leptin circulating levels in healthy men. The independent association of circulattng leptin with insulin concentrations is consistent with experimental evidence that insulin increases leptin gene expression and circulating levels in rodents (4,21,22) and is in agreement with recent data demonstrating increased leptin secretion by human adipocytes in vitro (23) and increased serum leptin levels in vivo (23), in response to chronic hyperinsulinemia. Furthermore, these data are also consistent with recent evidence suggesting that leptin correlates with pancreatic insulin secretion independent of percent body fat in healthy humans, but not in subjects with impaired glucose tolerance (lgt) or non-insulin-dependent diabetes (NIDDM) (43--45). Thus, because insulin-resistant states, i.e., IGT and NIDDM, are not associated with increased leptin levels (30,43--45), these data may suggest that endogenous hyperinsulinemia is associated with increased circulating leptin only in the absence of insulin resistance. Finally, given the previously shown lack of any acute effect of insulin on leptin production (23), our data are consistent with the notion that insulin-stimulated leptin secretion could represent the result of a long-term trophic effect of insulin on adipocytes, rather than one of the classic metabolic actions of insulin, such as inhibition of lipolysis (4). Another important finding of this study is the lack of an independent correlation between leptin and serum testosterone. Leptin influences the reproductive system in rodents (10,11,12), and sex steroids have been proposed as one of the potential explanations for the observed differences in leptin levels between men and women (46). However, although a previous cross-sectional study demonstrated a negative correlation between testosterone and leptin levels in healthy men, this association became nonsignificant after adjusting for the effect ofbmi (32). The negative association by bivariate analysis and the lack of any independent correlation between serum leptin and androgen levels in our study are in agreement with a previous study (32) and suggest that either other factors, such as estrogens or progesterone, or an interaction between high estrogen and low androgen levels may be responsible for the observed sexual dimorphism in leptin levels (46). Future studies are needed to elucidate this issue. Finally, previous studies have revealed inconsistent results with respect to the association of circulating leptin and cortisol levels. More specifically, positive (34), negative (33), and nonsignificant (32) associations have been previously reported. Our study confirms in another population of normal men the lack of an independent correlation between circulating leptin and cortisol levels. This study also demonstrates a positive and independent association of circulating leptin concentrations with alcohol intake (4). Alcohol intake has been associated with a high daily energy intake (47). Furthermore, alcohol intake decreases lipid oxidation, favors energy storage as fat, and promotes weight gain (48). These observations are consistent with the significant positive association between alcohol intake and leptin levels in this study. However, the fact that this association persists after controlling for obesity indicates that alcohol may exert an additional direct effect on the adipocyte. Alcohol has been shown to reduce lipolysis and circulating free fatty acids (FFA) in humans (47--49). Thus, inhibition of lipolysis may underlie the independent positive association between alcohol intake and circulating leptin concentrations in this study, although other mechanisms, metabolic or trophic, on the adipocyte cannot be excluded (4,49). These data are consistent with recent studies that demonstrate that alcohol increases resting energy expenditure (48-50). Cigarette smoking decreases body weight in humans, and cessation of smoking has been associated with weight gain (19,20). This metabolic effect of smoking is due to stimulation of energy expenditure (16,17) and increased thermogenesis (13,18). In addition, cigarette smoking has been shown to increase plasma catecholamine and FFA concentrations (14,15). Circulating catecholamines or activa- OBESITY RESEARCH Vol. 6 NO.3 May

6 tion of adrenergic receptors increased lipolysis in humans (14,15) and have been shown to decrease leptin levels in vitro and in animal studies (26-29,51). Finally, a recent cross-sectional study revealed that smokers had lower leptin levels than nonsmokers, after adjusting for BMI (52). Given the independent, negative association between smoking and leptin concentrations shown in this study, as well as the previously mentioned one (52), it can be concluded that the effect of smoking on energy balance and body weight occurs independent of leptin, most likely via increased sympathetic nervous system activity, which might itself be responsible for the fall in leptin levels. Our cross-sectional study revealed no significant association between daily physical activity, as assessed by means of an interviewer-administered questionnaire (37,38), and serum leptin. Similarly, physiological studies on normal men have failed to show a change in leptin levels after exercise (53-56), although prolonged strenuous exercise, such as ultramarathon, may decrease leptin levels (57). In addition, leptin is not associated with total energy expenditure in healthy men (58). Both leptin (43) and steroid hormone secretion display a circadian variation (37,59). Previous studies have shown that a single blood sample drawn at a standardized and narrow time frame reliably reflects the mean hormonal concentrations for similar epidemiological studies (37,59). Additionally, circulating concentrations of the measured hormones vary considerably with age (59,60). The young age and narrow age range of the study subjects, the performance of appropriate statistical control, and the fact that timing of the blood draw was standardized limit errors from these sources (37). We have not measured thyroid hormones in this study because we have previously shown that shortterm hyperthyroxinemia that does not alter body weight has no effect on serum leptin concentrations (61). We presume some misclassification in the reporting of demographic asd lifestyle variables, the measurement of anthropometric indexes, and the laboratory determinations of hormones. Additionally, although leptin levels are not affected (4,62), circulating insulin levels may change in response to breakfast. As a result, it could be argued that measured insulin levels may not accurately reflect the true mean circulating insulin levels of study subjects and, thus, could have resulted in misclassification, which in turn, may have influenced the observed associations. The resulting errors in relation to anthropometric indexes and the laboratory determinations of insulin and leptin are clearly uncorrelated, however, because the laboratory tests were blindly performed and the personnel involved had no knowledge of either the identity of the subjects or their anthropometric, demographic, and lifestyle data. Such nondifferential misclassification should bias the regression coefficients toward the null value and should weaken the corresponding level of statistical significance (63). Furthermore, the extent of this misclassification is no less for insulin than for the other hormones. Thus, the above factors either have not affected the results of this study or, if anything, have weakened the associations studied (63). In summary, this study demonstrates that obesity, smoking, circulating insulin concentrations, and alcohol intake are all important independent predictors of circulating leptin levels in healthy young men. Both leptin (10) and smoking alter neuroendocrine function (64), cortisol (65), testosterone (66), and thyroid hormone levels (67,68). The precise mechanism by which smoking and alcohol intake act on the adipocyte to regulate leptin levels and the mechanism by which these effects on circulating leptin may affect human physiology will be the subject of future investigation. References 1. Zhang Y, Proenca R, Maffei M, Barone M, Leopold L, Friedman JM. Positional cloning of the mouse ob gene and its human homologue. Nature. 1994;372: Frederich RC, Hamann A, Anderson S, Lollman B, Lowell BB, Flier JS. Leptin reflects body lipid content in mice: evidence for diet-induced leptin resistance. Nature Med. 1995; I: Rink TJ. In search of a satiety factor. Nature. 1995;372: Caro JF, Sinha MK, Kolaczynski JW, Zhang PL, Considine RV. Leptin. The tale of an obesity gene. Diabetes. 1996; 45: Maffei M, Halaas J, Ravussin E, et al. Leptin levels in human and rodent: measurement of plasma leptin and ob RNA in obese and weight-reduced subjects. Nature Med. 1995;I: Halaas JL, Gajiwala KS, Maffei M, et al. Weight-reducing effects of the plasma protein encoded by the obese gene. Science. 1995;269: Pelleymounter MA, Cullen MJ, Baker MB, et al. Effects of the obese gene product on body weight regulation in ob/ob mice. Science. 1995;269: Campfield LA, Smith FJ, Guisez Y, Devos R, Burn P. Recombinant mouse ob protein: evidence for a peripheral signal linking adiposity and central neural networks. Science. 1995;269: Weigle DS, Bukowski TR, Foster DC, et al. Recombinant ob protein reduces feeding and body weight in the ob/ob mouse. J Clin Invest. 1995;96: Ahima R, Prabakaran D, Mantzoros C, et al. Role of leptin in the neuroendocrine response to fasting. Nature. 1996;382: I I. Barash la, Cheung CC, Weigle DS, et al. Leptin is a metabolic signal to the reproductive system. Endocrinology. 1996; 137: Chehab F, Lim M, Lu R. Correction of the sterility defect in homozygous obese female mice by treatment with the human recombinant leptin. Nat Genet. 1996;12: Perkins KA, Sexton JE, Epstein LH, et al. Acute thermogenic effects of nicotine combined with caffeine during physi- 184 OBESITY RESEARCH Vol. 6 NO.3 May 1998

7 cal activity in male and female smokers. Am J Clin Nutr. 1994;60: Neese RA, Benowitz NL, Hoh R, Faix D, LaBua A, Hellerstein MK. Metabolic interactions between surplus dietary energy intake and cigarette smoking or its cessation. Am J Physiol. 1994;267:EI Grassi G, Seravalle G, Calhoun DA, et ai. Mechanisms responsible for sympathetic activation by cigarette smoking in humans. Circulation. 1994;90: Perkins KA, Epstein LH, Marks BL, Stiller RL, Jakob RG. The effect of nicotine on energy expenditure during light physical activity. N Engl ] Med. 1989;320: Audrain JE, Klesges RC, DePue K, Klesges LM. The individual and combined effects of cigarette smoking and food on resting energy expenditure. lnt J Obes. 1991;15: Perkins KA. Metabolic effects of tobacco smoking. J Appl Physiol. 1992;72: Flegal KM, Troiano RP, Kuczmarski RJ, Campbell SM. The influence of smoking cessation on the prevalence of overweight in the United States. N Engl J Med. 1995;333: Williamson DF, Madans J, Anda RF, Kleinman JC, Giovino GA, Byers T. Smoking cessation and severity of weight gain in a national cohort. N Engl J Med. 1991;324: Cusin I, Sainsbury A, Doyle P, Rohner-Jeanrenaud F, Jeanrenaud B. The ob gene and insulin. Diabetes. 1995;44: Saladin R, De Vos P, Guerre-Millo M, et ai. Transient increase in obese gene expression after food intake or insulin administration. Nature. 1995;377: Kolaczynski JW, Nyce MR, Considine RV, et ai. Acute and chronic effect of insulin on leptin production in humans: studies in vivo and in vitro. Diabetes. 1996;45: De Vos P, Saladin R, Auwerx J, Staels B. Introduction of ob gene expression by corticosteroids is accompanied by body weight loss and reduced food intake. J Bioi Chern. 1995;270: Murakami T, Iida M, Shima K. Dexamethasone regulates obese expression in isolated rat adipocytes. Biochem Biophys R~s Commun. 1995;214: Slieker LJ, Sloop KW, Surface PL, et ai. Regulation of expression of ob mrna and protein by glucocorticoids and c AMP. ] Bioi Chern. 1996;271: Collins S, Kuhn CM, Petro AE, Swick AG, Chrunyk BA, Surwit RS. Role of leptin and fat regulation. Nature. 1996; 380: Trayhurn P, Duncan JS, Rayner DV. Acute cold-induced suppression of ob (obese) gene expression in white adipose tissue of mice: mediation by the sympathetic system. Biochem t. 1996;311: Mantzoros CS, Frederich RC, Qu D, et al, Activation of [33 adrenergic receptors suppresses leptin expression and mediates a leptin-independent inhibition of food intake in mice. Diabetes. 1996;45: Haffner SM, Stern MP, Miettinen H, Wei M, Gingerich RL. Leptin concentrations in diabetic and nondiabetic Mexican-Americans. Diabetes. 1996;45: Dagogo-Jack S, Fanelli C, Paramore D, Brothers J, Landt. Plasma leptin and insulin relationships in obese and nonobese humans. Diabetes. 1996;45: Haffner SM, Miettinen H, Karhapaa P, Mykkanen L, Laasko M. Leptin concentrations, sex hormones and cortisol in nondiabetic men. J Clin Endocrinol Metab. 1997;82: Laughlin GA, Yen SS. Hypoleptinemia in women athletes: absence of a diurnal rhythm with amenorrhea. J Clin Endocrinol Metab. 1997;82: Tuominen GA, Ebeling P, Laquier FW, Heiman ML, Stephens T, Koivisto VA. Serum leptin concentrations and fuel homeostasis in healthy men. Eur J Clin Invest. 1997;27: Mantzoros C, Moschos S, Avramopoulos I, et ai. Leptin concentrations in relation to body mass index and the tumor necrosis factor-a system in humans. J Clin Endocrinol Metab. 1997;82: World Health Organization. Alcohol drinking. IARC Monogr. 1988;44: Mantzoros CS, Georgiadis EI. Body mass and physical activity are important predictors of serum androgen concentrations in young healthy men. Epidemiology. 1995;6: Mantzoros C, Georgiadis E, Trichopoulos D. Contribution of dihydrotestosterone to male sexual activity. BM]. 1995; 310: Houseknecht KL, Mantzoros CS, Kuliawat R, Hadro E, Flier JS, Kahn BB. Evidence for leptin binding to proteins in serum of rodents and humans: Modulation with obesity. Diabetes. 1996;45: Mantzoros C, Flier JS, Lesem MD, Brewrton TD, Jimerson DC. Cerebrospinal fluid leptin in anorexia nervosa: correlation with nutritional status and potential role in resistance to weight gain. J Clin Endocrinol Metab. 1997;82: Pagano M, Gauvreau K. Principles of Biostatistics. Wadsworth Inc.; 1993; Larsson H, Ahren B. Short-term dexamethasone treatment increases plasma leptin independently of changes in insulin sensitivity in healthy women.v Clin Endocrinol Metab. 1996; 81: Sinha MK, Ohanessian JP, Heinman ML, et ai. Nocturnal rise of leptin in lean, obese and non-insulin dependent diabetes mellitus subjects. J Clin Invest. 1996;97: Nolan JJ, Olefsky JM, Nyce MR, Considine RV, Caro JF. Effect of troglitazone on leptin production. Studies in vitro and in human subjects. Diabetes. 1996;45: Larsson H, Elmstahl S, Ahren B. Plasma leptin levels correlate to islet function independently of body fat in postmenopausal women. Diabetes. 1996;45: Rosenbaum M, Nicolson M, Hirsch J, et ai. Effects of gender, body composition and menopause on plasma concentrations of leptin.v Clin Endocrinol Metab. 1996;81: Tremblay A, Wonters E, Wenker M, St-Pierre S, Bouchard C, Despres JP. Alcohol and a high-fat diet: a combination favoring overfeeding. Am]Clin Nutr. 1995;62: Suter PM, Schutz Y, Jequier E. The effect of ethanol on fat storage in healthy subjects. N Engl ] Med. 1992;326: Avogaro A, Beltramello F, Gundi L, et ai. Alcohol intake OBESITY RESEARCH Vol. 6 NO.3 May

8 impairs glucose counterregulation during acute insulininduced hypoglycemia in 100M patients. Diabetes. 1993;42: K1esges RC, Measler CZ, Klesges LM. Effects of alcohol intake on resting energy expenditure in young women social drinkers. Am J Clin Nutr. 1994;59: Gettys TW, Harkness PJ, Watson PM. The ~3- adrenergic receptor inhibits insulin-stimulated leptin secretion from isolated rat adipocytes. Endocrinology. 1996;137: Wei M, Stern MP, Haffner SM. Serum leptin levels in Mexican Americans and non-hispanic whites: association with body mass index and cigarette smoking. Ann Epidemiol. 1997; 7: Hickey MS, Houmard JA, Considine RV, et al. Genderdependent effects of exercise training on serum leptin levels in humans. Am J Physiol. 1997;272:E Perusse L, Collier G, Gagnon J, et al, Acute and chronic effects of exercise on leptin levels in humans. J Appl Physiol. 1997;83: Racette SB, Coppack SW, Landt M, Klein S. Leptin production during moderate-intensity aerobic exercise. J Clin Endocrinol Metab. 1997;82: Kohrt WM, Landt M, Birge SR. Serum leptin levels are reduced in response to exercise training, but not hormone replacement therapy, in older men. J Clin Endocrinol Metab. 1996;81: Landt M, Lawson GM, Helgeson JM, et al. Prolonged exercise decreases serum leptin concentrations. Metabolism. 1997;46: Nicklas BJ, Toth MJ, Poehlman ET. Daily energy expenditure is related to plasma leptin concentrations in older Af- rican-american women but not men. Diabetes. 1997;46: Parker LN. Adrenal Androgens in Clinical Medicine. San Diego: Academic Press; 1989; Orentreich N, Brind JL, Rizer RL, Vogelman VH. Age changes and sex differences in serum dehydroepiandrosterone sulfate concentrations throughout adulthood. J Clin Endocrinol Metab. 1984;59: Mantzoros CS, Rosen HN, Greenspan SL, Flier JS, Moses AC. Short-term hyperthyroidism has no effect on 1eptinlevels in man. J Clin Endocrinol Metab. 1997;82: Considine RV, Sinha MK, Heiman ML, et al. Serum immunoreactive-ieptin concentrations in normal-weight and obese humans. N Engl J Med. 1996;334: Rothman K. Modern Epidemiology. Boston: Little Brown and Co.; Pomerleau OF, Fertig JB, Seyler LE, Jaffe J. Neuroendocrine reactivity to nicotine in smokers. Psychopharmacology. 1983;81: Kirschbaum C, Wust S, Strasburger CJ. Normal cigarette smoking increases free cortisol in habitual smokers. Life Sci. 1992;50: Sofikitis N, Miyagawa I, Dimitriadis D, Zavos P, Sikka S, Hellstrom W. Effects of smoking on testicular function, semen quality and sperm fertilizing capacity. J Urol. 1995;154: Bertelsen JB, Hegedus L. Cigarette smoking and the thyroid. Thyroid. 1994;4: Winsa B, Mandahl A, Karlsson FA: Graves' disease, endocrine opthalmopathy and smoking. Acta Endocrinol. 1993; 128: OBESITY RESEARCH Vol. 6 NO.3 May 1998