ALINA GAVRILA, JEAN L. CHAN, NIKOS YIANNAKOURIS, MEROPI KONTOGIANNI, LISA C. MILLER, CHRISTINE ORLOVA, AND CHRISTOS S. MANTZOROS

0021-972X/03/$15.00/0 The Journal of Clinical Endocrinology & Metabolism 88(10):4823 4831 Printed in U.S.A. Copyright 2003 by The Endocrine Society doi: 10.1210/jc.2003-030214 Serum Adiponectin Levels Are Inversely Associated with Overall and Central Fat Distribution but Are Not Directly Regulated by Acute Fasting or Leptin Administration in Humans: Cross-Sectional and Interventional Studies ALINA GAVRILA, JEAN L. CHAN, NIKOS YIANNAKOURIS, MEROPI KONTOGIANNI, LISA C. MILLER, CHRISTINE ORLOVA, AND CHRISTOS S. MANTZOROS Division of Endocrinology and Metabolism (A.G., J.L.C., C.O., C.S.M.), Department of Internal Medicine, Beth Israel Deaconess Medical Center, Boston, Massachusetts 02215; Department of Home Economics and Ecology (N.Y., M.K.), Harokopio University, Athens, Greece 17671; and Department of Biostatistics (L.C.M.), Harvard School of Public Health, Boston, Massachusetts 02215 Adiponectin is an adipocyte-secreted protein that circulates in high concentrations in the serum and acts to increase insulin sensitivity. Previous studies have shown that serum adiponectin is inversely associated with fat mass and insulin resistance in humans and that acute fasting decreases adipose tissue adiponectin mrna expression in rodents. Whether acute energy deprivation, body fat distribution, or serum hormone levels are associated with circulating adiponectin in humans remains largely unknown. To identify predictors of serum adiponectin levels, we evaluated the association of adiponectin with several anthropometric, metabolic, and hormonal variables in a crosssectional study of 121 women without a known history of diabetes. We also performed interventional studies to assess whether fasting for 48 h and/or leptin administration regulates serum adiponectin in healthy men and women. Abbreviations: apo, Apolipoprotein; BIA, bioelectrical impedance analysis; BMI, body mass index; CV, coefficient of variation; DEXA, dual energy x-ray absorptiometry; HDL, high-density lipoprotein; HOMA- IR, homeostasis model assessment of insulin resistance; r-methuleptin, recombinant-methionyl human leptin; WC, waist circumference; WHR, waist-to-hip ratio. Our cross-sectional study shows that, in addition to overall obesity, central fat distribution is an independent negative predictor of serum adiponectin and suggests that adiponectin may represent a link between central obesity and insulin resistance. In addition, estradiol is negatively and independently associated with adiponectin, whereas there is no association between serum adiponectin and leptin, cortisol, or free testosterone levels. Our interventional studies demonstrate that neither fasting for 48 h, resulting in a low leptin state, nor leptin administration at physiological or pharmacological doses alters serum adiponectin levels. Further studies are needed to fully elucidate the physiology of adiponectin in humans and its role in the pathogenesis of insulin-resistant states. (J Clin Endocrinol Metab 88: 4823 4831, 2003) ADIPONECTIN (ADIPOQ, ACRP30, apm1) is a protein exclusively secreted by mature adipocytes that circulates in high concentrations in the blood (1). Accumulating evidence from animal and human studies shows that adiponectin increases insulin sensitivity (2 6), has antiinflammatory (7) and antiatherogenic effects (8, 9), and may improve the lipid profile (4, 5). Serum adiponectin levels are decreased in patients with obesity and type 2 diabetes mellitus (10 12) and are inversely associated with parameters of overall adiposity [e.g. body mass index (BMI), fat mass, and percentage of body fat] (2, 3, 10 15), as well as insulin resistance, independently of fat mass (13, 14, 16). Because central obesity and visceral fat are more closely associated with insulin resistance than sc or total fat (17, 18), it is important to evaluate whether adiponectin is associated primarily with central adiposity rather than overall adiposity and whether central fat affects insulin sensitivity by altering adiponectin levels. Although Cnop et al. (16) recently reported that intraabdominal fat is an independent negative predictor of serum adiponectin in Caucasians, a Japanese study has found an inverse correlation between adiponectin and waistto-hip ratio (WHR) in morbidly obese but not in overweight or moderately obese subjects (15). In addition, the entire spectrum of predictors of adiponectin levels remains to be fully elucidated. It has been previously proposed that a physiologically significant relationship may exist between leptin and adiponectin, because these two hormones have additive effects in normalizing insulin sensitivity in animals (4). Furthermore, studies in rodents support a possible role of leptin in regulating adiponectin, showing that fasting, a state that acutely decreases leptin expression and its serum concentration, also decreases adiponectin gene expression in adipose tissue, whereas refeeding normalizes the expression of both hormones (19). In contrast to adiponectin gene expression, however, neither changes of leptin levels induced by fasting or refeeding nor exogenously administered leptin has been shown to alter serum adiponectin levels in rodents (19). No interventional study has directly assessed the effect of altering leptin levels on serum adiponectin in humans, and two cross-sectional studies have reported discrepant results on the association between leptin and adiponectin (14, 16). 4823

4824 J Clin Endocrinol Metab, October 2003, 88(10):4823 4831 Gavrila et al. Predictors of Serum Adiponectin Furthermore, in vitro adipocyte exposure to glucocorticoids, which are known to regulate insulin sensitivity (20), inhibits adiponectin expression (21, 22), suggesting that glucocorticoids may influence insulin sensitivity through an effect on serum adiponectin levels. Thus, it would be of interest to study the potential association between adiponectin and steroid hormone levels in vivo in humans. Finally, although the gender dimorphism in adiponectin levels, with women having higher adiponectin levels than men independently of body fat mass or fat distribution (10, 13, 16, 23), has been attributed to differences in circulating estrogens or androgens, no previous study has jointly evaluated body fat mass, fat distribution, and sex steroids as predictors of serum adiponectin levels in humans; and studies that have focused on adiponectin levels in postmenopausal vs. premenopausal women have reported conflicting results (23, 24). We sought to assess the association of several anthropometric, metabolic, and hormonal variables with adiponectin levels in a cross-sectional study of 121 women without a known history of diabetes. In addition, to evaluate the potential role of fasting and/or leptin in regulating serum adiponectin, we measured adiponectin levels at baseline and after 2 d of fasting, with and without leptin administration in physiological replacement doses, in healthy normalweight men. In a separate group of healthy men and women, we measured adiponectin levels in response to a pharmacological dose of leptin to evaluate whether leptin has a dose-related effect on adiponectin levels. Subjects and Methods Cross-sectional study One hundred thirty Greek Caucasian women without a known history of diabetes (age, 49.4 9.2 yr; BMI, 30.9 5.5 kg/m 2 ) were consecutively enrolled in this study, which was approved by the Ethics Committee at Harokopio University (Athens, Greece) and the Institutional Review Board at Beth Israel Deaconess Medical Center (Boston, MA). After giving written informed consent to participate, subjects provided a fasting blood sample and completed a self-administered questionnaire on demographic characteristics, general health, smoking status, and present medications, including hormone replacement and oral contraceptive treatment. Subjects were classified as premenopausal if they had regular menstrual periods, perimenopausal if they had irregular periods and/or elevated FSH levels, and postmenopausal if they reported no periods for at least 6 months. Weight to the nearest 0.5 kg, height to the nearest 0.5 cm, and waist and hip circumferences were measured. All subjects underwent bioelectrical impedance analysis (BIA) with a single-frequency bioimpedance analyzer (Model 101, RJL- Systems, Mt. Clemens, MI), and body fat mass (in kilograms and percentage of body weight) was calculated as previously described (25). To validate the BIA and anthropometric measurements, 60 subjects also underwent a dual energy x-ray absorptiometry (DEXA) scan performed with a Lunar DPX densitometer (software version 4.7e; Lunar Corp., Madison, WI) for assessment of total body and trunk fat mass (in kilograms and percentage of body weight) (26, 27). Estimates of fat mass and percentage of fat obtained in the 60 subjects who underwent all three assessments correlated strongly with each other (fat mass by BIA and DEXA: Pearson coefficient 0.91, P 0.001; percentage fat by BIA and DEXA: Pearson coefficient 0.88, P 0.001) and with BMI (fat mass or percentage of fat by BIA and BMI: Pearson coefficient 0.94, P 0.001; fat mass or percentage of fat by DEXA and BMI: Pearson coefficient 0.84, P 0.001). In addition, trunk fat, a measurement of central adiposity obtained by DEXA, correlated highly with waist circumference (WC) (Pearson coefficient 0.90, P 0.001) and waist-to-hip ratio (WHR) (Pearson coefficient 0.54, P 0.001). Thus, in our analysis, we used BMI and fat mass calculated by BIA as markers of overall adiposity, and WC and WHR as markers of central obesity. Analysis was restricted to 121 subjects for whom adiponectin data were available. Due to insufficient serum samples, six subjects did not have a complete hormonal and metabolic evaluation, and 10 subjects did not have either the fasting glucose or insulin levels needed to calculate the homeostasis model assessment of insulin resistance (HOMA-IR), but each statistical analysis included only subjects with complete data. Apolipoprotein (apo)a1 and apob levels were available for only a subset of 48 subjects. Interventional studies Two interventional studies investigating the effect of fasting and/or leptin administration on adiponectin levels were approved by the Beth Israel Deaconess Medical Center Institutional Review Board and were performed under an investigator-initiated new drug application. Clinical-quality recombinant-methionyl human leptin (r-methuleptin) was supplied by Amgen, Inc. (Thousand Oaks, CA). All subjects gave informed consent for participation. Fasting with or without administration of physiological doses of r-methuleptin. Eight normal-weight men (age, 23.4 1.5 yr; BMI, 23.2 1.4 kg/m 2 ) were screened for significant medical conditions, and none of them had abnormal glucose tolerance (random serum glucose, 77 12 mg/dl; range, 59 91 mg/dl). Subjects were admitted to the General Clinical Research Center on three separate occasions (separated by at least 7 wk to allow for recovery of hematocrit and normalization of their metabolic state) under three different conditions: baseline fed state, fasting with placebo administration, and fasting with r-methuleptin administration designed to restore the fasting-induced decline in leptin levels to physiological levels similar to those in the fed state. All eight subjects completed the fed and fasting/placebo admissions, and six subjects (age, 22.0 2.1 yr; BMI, 24.2 1.2 kg/m 2 ) completed the fasting/rmethuleptin admission. Subjects were admitted to the General Clinical Research Center the evening before study d 1 for each fed or fasting phase. Fasting blood samples for adiponectin and leptin measurements were obtained at 0800 h on d 1 and 3 of each admission. During the baseline fed state, subjects were placed on an isocaloric diet designed to maintain their admission body weight, with four standardized meals per day: 20% of calories from breakfast (0800 h), 35% from lunch (0100 h), 35% from dinner (1800 h), and 10% from a snack (2200 h). During the fasting studies, subjects received only caffeine-free and calorie-free liquids for 2 d, and NaCl (500 mg), KCl (40 meq), and a standard multivitamin with minerals daily. Starting at 0800 h on d 1 of the fasting/leptin admission, r-methuleptin was administered as a sc injection every 6 h for 2 d, at a dose of 0.04 mg/kg d on the first day and 0.1 mg/kg d on the second day. The dosing regimen used was based on previous detailed pharmacokinetic studies performed in the fasted state (Mantzoros, C. S., unpublished data). During the fasting/placebo admission, a buffer solution was administered sc every 6 h. Administration of a pharmacological dose of r-methuleptin in the fed state. Five normal-weight men (age, 22.2 2.0 yr; BMI, 22.0 1.0 kg/m 2 ), five obese men (age, 23.4 3.4 yr; BMI, 32.0 2.3 kg/m 2 ), and five normalweight women (age, 20.4 1.6 yr; BMI, 21.9 1.8 kg/m 2 ) were screened for significant medical problems. None of the subjects had abnormal glucose tolerance (lean men: random serum glucose, 73 19 mg/dl; range, 54 95 mg/dl; obese men: random serum glucose, 67 7 mg/dl; range, 57 74 mg/dl; lean women: random serum glucose, 76.8 7.1 mg/dl; range, 65 84 mg/dl). Subjects were admitted to the General Clinical Research Center during the evening and received one pharmacological dose of r-methuleptin (0.3 mg/kg) the following morning. Fasting blood samples were collected at 0800 h immediately before r-methuleptin administration, and at 6 and 12 h after the leptin dose. Subjects received an isocaloric diet during the study, as described above. Hormone measurements. Glucose and lipid profiles were measured using a photometric method with liquid reactants (Hitachi 917, Indianapolis, IN). Hormone concentrations were measured using commercially available RIA, as follows: adiponectin [Linco Research, St. Charles, MO; sensitivity, 2 g/ml; intraassay coefficients of variation (CV), 1.78 6.21%]; leptin (Linco Research; sensitivity, 0.5 ng/ml; intraassay CV, 8.3%); cortisol (DSL, Webster, TX; sensitivity, 0.5 g/dl; intraassay CV,

Gavrila et al. Predictors of Serum Adiponectin J Clin Endocrinol Metab, October 2003, 88(10):4823 4831 4825 5.3 8.4%), insulin (DSL; sensitivity, 1.3 IU/ml; intraassay CV, 8.3%); estradiol (DPC, Los Angeles, CA; sensitivity, 8 pg/ml; intraassay CV, 4.3 7%), and free testosterone (DPC; sensitivity, 0.15 pg/ml; intraassay CV, 8%). Insulin resistance was estimated using the HOMA with the following formula: HOMA-IR fasting insulin ( IU/ml) glucose (mmol/liter)/22.5 (28, 29). To minimize variability, hormone levels were measured in one assay for all subjects participating in the cross-sectional study and in one assay for each subject in the interventional studies. Data analysis. SPSS 8.0 software (SPSS Inc., Chicago, IL) was used for statistical analysis, and a P value 0.05 (two-tailed) was considered statistically significant for all analyses. Several variables were logarithmically transformed to obtain a normal distribution, as indicated in Table 2. In the cross-sectional study, we assessed the differences in body composition and metabolic and hormonal parameters according to menopausal status using one-way ANOVA tests with post hoc (least significant difference) analysis. We then calculated Pearson s correlation coefficients and performed bivariate and multivariate regression analyses to evaluate potential associations between serum adiponectin levels and anthropometric (BMI, fat mass, WC, and WHR), metabolic (fasting glucose, insulin, HOMA-IR index, and lipid profile) as well as hormonal (leptin, cortisol, estradiol, and testosterone) parameters, all expressed as continuous variables. We controlled our analyses for potential confounders, including age, menopausal status, fat mass, WHR, and estrogen levels, as indicated in Table 2. The sample size of 121 subjects provides more than 80% power to detect a correlation coefficient r of at least 0.22 at the conventional level of statistical significance 0.05. We used nonparametric tests for the interventional studies, because the number of subjects was relatively small, and the data were not normally distributed. We performed exact Wilcoxon signed rank tests to evaluate changes in adiponectin levels, leptin levels, and weight between d 1 and 3 of each fed or fasting admission in the first interventional study. We used Friedman tests to evaluate changes in adiponectin and leptin levels from baseline to 6 and 12 h after administration of the pharmacological dose of r-methuleptin in the second interventional study. Results Cross-sectional study Baseline characteristics of the 121 women are summarized in Table 1. Twelve subjects had a BMI lower than 25 (24.1 0.5) kg/m 2, 52 subjects had a BMI between 25 and 30 (27.6 1.5) kg/m 2, and 57 subjects had a BMI greater than 30 (35.4 4.8) kg/m 2. Mean serum adiponectin level was 10.4 7.1 g/ml, and its relationship with BMI, fat mass, WHR, serum leptin, serum insulin, and HOMA-IR index is shown in Fig. 1. Seventy-one percent of subjects were nonsmokers, 26% were current smokers, whereas 3% were past smokers. Although subjects did not have a known history of diabetes, we found four subjects with fasting glucose levels higher than 126 mg/dl, and data analysis with and without these four subjects resulted in similar results. Because five subjects were undergoing hormone replacement treatment and two subjects were undergoing oral contraceptive treatment at the time of the evaluation, we also analyzed the data with and without these subjects and then adjusted our analysis for a dummy variable (estrogen yes or no), and each time we obtained similar results. In addition, there was no significant difference in serum adiponectin levels between the women taking estrogens and those not on estrogen treatment (serum adiponectin, 11.83 4.40 vs. 10.39 7.28 g/ml; P 0.45). Thus, we report herein the entire group analysis, with and without adjustment for estrogen levels in several regression analysis models, as indicated in Table 2. We also report results obtained when adjusting our analysis for premenopausal vs. postmenopausal status; adjustment for premenopausal vs. perimenopausal and postmenopausal status gave similar results. Of note, there was no correlation between adiponectin levels and the day of menstrual cycle in the pre- and perimenopausal women, but the study was not adequately powered for this outcome. Postmenopausal women had significantly higher adiponectin, triglyceride, low-density lipoprotein, and total cholesterol levels, and lower estradiol levels compared with premenopausal women, whereas leptin, cortisol, free testosterone levels, and body composition parameters were not TABLE 1. Cross-sectional study: baseline characteristics (mean SD) and classification according to menopausal status All (n 121) Premenopausal (n 44) Perimenopausal (n 12) Postmenopausal (n 65) Overall P value Demographic and anthropometric parameters Age (yr) 49.4 9.2 41.2 7.6 c 46.4 2.5 55.5 5.8 b,d 0.0001 BMI (kg/m 2 ) 30.9 5.5 31.3 5.9 c 34.9 6.9 30.0 4.6 c 0.01 Fat mass (kg) 32.9 9.8 34.4 10.7 39.2 11.2 30.8 8.3 c 0.01 Fat mass (%) 41.3 4.2 41.3 4.4 43.7 4.0 40.9 3.9 c 0.09 Waist circumference (cm) 88.6 12.9 90.0 12.8 c 99.3 13.7 85.7 11.8 d 0.003 WHR 0.80 0.06 0.80 0.07 0.82 0.04 0.79 0.06 0.32 Hormonal and metabolic parameters Adiponectin ( g/ml) 10.4 7.1 8.8 6.0 6.8 4.0 12.2 7.9 a,c 0.001 Leptin (ng/ml) 23.4 13.7 23.6 14.1 25.8 10.2 22.7 14.1 0.49 Estradiol (pg/ml) 188.1 114.4 244.1 129.6 229.8 132.5 142.7 75.5 b,c 0.0001 Free testosterone (pg/ml) 1.4 0.9 1.5 0.8 1.2 0.6 1.4 1.0 0.63 Cortisol ( g/dl) 9.5 4.1 10.1 4.1 9.3 2.4 9.2 4.4 0.30 Fasting glucose (mg/dl) 99.8 16.3 98.7 12.9 105.7 18.1 99.4 18.0 0.41 Fasting insulin ( IU/ml) 7.9 4.1 8.6 4.4 8.5 4.3 7.3 3.9 0.16 Insulin resistance (HOMA) 2.0 1.3 2.1 1.3 2.3 1.3 1.9 1.2 0.29 Triglycerides (mg/dl) 102 56 85 36 c 124 64 110 64 a 0.02 HDL cholesterol (mg/dl) 55 12 53 10 47 12 57 13 c 0.02 LDL cholesterol (mg/dl) 141 37 124 32 144 32 153 37 a 0.0001 Total cholesterol (mg/dl) 216 42 196 36 217 40 231 42 b 0.0001 ApoA1 (mg/dl) 115 12 117 11 108 16 116 9 0.15 ApoB (mg/dl) 87 22 79 20 c 99 21 93 20 a 0.02 Overall P values derived by ANOVA. LDL, Low-density lipoprotein. a 0.05 P 0.001. b P 0.001 by post hoc t tests, compared to premenopausal group. c 0.05 P 0.001. d P 0.001 by post hoc t tests, compared to perimenopausal group.

4826 J Clin Endocrinol Metab, October 2003, 88(10):4823 4831 Gavrila et al. Predictors of Serum Adiponectin FIG. 1. Cross-sectional study. Relationship between serum adiponectin levels and BMI, fat mass, WHR, serum leptin, serum insulin, and HOMA-IR. significantly different using ANOVA with post hoc analysis (Table 1). The difference in serum adiponectin between premenopausal and postmenopausal women remained significant in multivariate linear regression analysis models adjusting for age and/or body composition but became nonsignificant after adjustment for estrogen levels. A posi-

Gavrila et al. Predictors of Serum Adiponectin J Clin Endocrinol Metab, October 2003, 88(10):4823 4831 4827 tive association of adiponectin with age [standardized coefficient (std ) 0.19; P 0.04] became borderline significant after adjusting for BMI (P 0.06), body fat (P 0.10), or WHR (P 0.05); moreover, there was no correlation between smoking and adiponectin. In addition, we found a strong negative correlation between adiponectin and estradiol levels by bivariate regression analysis, which remained significant after controlling for age, menopausal status, and body composition, but there was no relationship between adiponectin and leptin, cortisol, or free testosterone levels (Table 2). We found a strong negative association between serum adiponectin and parameters of central obesity (WHR and WC) independent of age, menopausal status, and estradiol levels (Table 2). Parameters of overall adiposity (BMI and fat mass) also correlated negatively with adiponectin in bivariate regression analyses, but these correlations appeared to be weaker than those between adiponectin and central obesity measurements. Although the association of adiponectin with BMI did not remain significant after controlling for age, menopausal status, and estradiol levels, the association with fat mass or estimates of central obesity remained significant after controlling for these potential confounders (Table 2). The negative associations between adiponectin and fasting glucose, insulin, or HOMA-IR index remained significant after adjusting for age, menopausal status, estrogen levels, and fat mass, but not WHR or WC, suggesting that adiponectin may be mediating, to a large extent, the effect of central obesity on insulin resistance and glycemia (Table 2). We also found a positive correlation of serum adiponectin with high-density lipoprotein (HDL) cholesterol and a negative correlation with triglyceride levels but no significant relationship with total or low-density lipoprotein cholesterol (Table 2). The association between adiponectin and triglycerides remained significant after adjusting for age, menopausal status, estrogen levels, but not for body composition. The strong association with HDL remained significant after controlling for all potential confounders, indicating that the association between HDL and adiponectin levels is independent of body fat distribution. The positive correlation of adiponectin with apoa1 in bivariate regression analysis (std 0.44; P 0.006) remained significant after adjustment for age, menopausal status, estrogen levels, and body composition (std 0.33; P 0.02), whereas the negative correlation between adiponectin and apobl tended to be significant in bivariate regression analysis (std 0.28; P 0.08), but was not significant after adjustment for the same potential confounders (std 0.13; P 0.45). Interventional studies Fasting with or without administration of physiological doses of r-methuleptin (Fig. 2). Weight did not change with isocaloric feeding (d 1, 73.0 8.1 kg vs. d 3, 73.1 8.4 kg; P 0.64) but decreased by approximately 2 kg during the fasting/placebo admission (d 1, 73.4 7.6 kg vs. d 3, 71.1 7.6 kg; P 0.01) and the fasting/r-methuleptin admission (d 1, 77.3 8.9 kg vs. d 3, 75.1 8.0 kg; P 0.06). Serum leptin levels did not change in the fed state (d 1, 1.6 1.6 ng/ml vs. d3,1.7 0.8 ng/ml; P 0.31) but decreased significantly to 33% of baseline after fasting for 2d(d1,2.1 1.8 ng/ml vs. d3,0.7 0.9 ng/ml; P 0.01). r-methuleptin administered in physiological doses during fasting increased serum leptin to levels higher than baseline but within the physiological range for lean men in the fed state (d 1, 2.3 1.9 ng/ml vs. d3,4.8 2.1 ng/ml; P 0.06). Serum adiponectin levels did not change during the fed (d 1, 11 5.1 g/ml vs. d 3, 12.9 6.4 g/ml; P 0.12) or fasting state (d 1, 11.1 7 g/ml vs. d 3, 10.1 6.2 g/ml; P 0.39), and r-methuleptin administration had no significant effect on adiponectin levels (d 1, 11.8 4 g/ml vs. d 3, 11.3 4.9 g/ml; P 0.56). Similar results were obtained when analysis was performed with data from all eight subjects or from the six subjects who TABLE 2. Cross-sectional study: bivariate and multivariate regression analyses of hormonal, anthropometric, and metabolic factors as predictors of serum adiponectin a levels 1 2 3 4 5 Leptin (ng/ml) a 0.14 0.14 0.10 0.07 0.02 Estradiol (pg/ml) a 0.36 d 0.31 c NA 0.28 b * 0.24 b ** Free testosterone (pg/ml) a 0.01 0.01 0.00 0.03 0.04 Cortisol ( g/dl) a 0.06 0.02 0.03 0.03 0.001 BMI (kg/m 2 ) 0.24 b 0.21 b 0.16 NA 0.02 Fat mass (kg) a 0.27 c 0.23 b 0.19 b NA 0.06 Waist circumference (cm) 0.32 c 0.28 c 0.25 c 0.42 b N/A WHR 0.33 d 0.33 d 0.31 c 0.28 c NA Fasting glucose (mg/dl) 0.24 b 0.24 b 0.23 b 0.20 b 0.14 Fasting insulin ( IU/ml) a 0.35 d 0.32 c 0.29 c 0.27 b 0.17 Insulin resistance (HOMA) a 0.37 d 0.34 d 0.31 c 0.31 c 0.20 Triglycerides (mg/dl) a 0.19 b 0.26 c 0.22 b 0.18 0.15 HDL cholesterol (mg/dl) 0.42 d 0.39 d 0.36 d 0.33 d 0.31 c LDL cholesterol (mg/dl) 0.08 0.04 0.02 0.02 0.01 Total cholesterol (mg/dl) 0.16 0.05 0.05 0.05 0.07 1, Bivariate standardized linear regression coefficient; 2, multivariate standardized linear regression coefficient adjusted for age and menopausal status; 3, adjusted for age, menopausal status, and estrogen levels; 4, adjusted for age, menopausal status, estrogen levels and fat mass; *, adjusted for age, menopausal status, and fat mass; 5, adjusted for age, menopausal status, estrogen levels and WHR; **, adjusted for age, menopausal status, and WHR. a Natural logarithmic transformation performed before analysis. b 0.05 P 0.01; c 0.01 P 0.001; d P 0.001.

4828 J Clin Endocrinol Metab, October 2003, 88(10):4823 4831 Gavrila et al. Predictors of Serum Adiponectin significant increase in leptin levels after r-methuleptin administration (data not shown). FIG. 2. First interventional study. Serum leptin (A) and adiponectin (B) levels at baseline and after 2 d of isocaloric feeding, fasting with placebo administration, or fasting with administration of physiological doses of r-methuleptin. *, 0.05 P 0.01, compared with baseline value by Wilcoxon signed rank tests. completed the fasting/r-methuleptin admission, and thus results for all eight subjects are reported. Administration of pharmacological dose of r-methuleptin in the fed state (Fig. 3). After the pharmacological dose of r-methuleptin was administered, serum leptin levels increased significantly to 15 40 times baseline levels at 6 h and approximately 10 times baseline at 12 h in the lean men (baseline, 3 0.7 ng/ml; 6 h, 111.1 33.7 ng/ml; 12 h, 32.4 11.8 ng/ml; P 0.007) and the obese men (baseline, 16.24 5 ng/ml; 6 h, 242.9 38.8 ng/ml; 12 h, 114.9 22.1 ng/ml; P 0.007). Over the time frame in which leptin levels were markedly increased, serum adiponectin levels did not change significantly (lean men baseline, 15.4 8.2 g/ml; 6 h, 16.3 10 g/ml; 12 h, 17.6 10 g/ml; P 0.25; obese men baseline, 9.7 5.7 g/ml; 6 h, 9.8 7.1 g/ml; 12 h, 10 7 g/ml; P 0.82). Similar to findings in men, serum adiponectin levels did not change in lean women, despite a Discussion Our data show that, in addition to overall obesity, central fat distribution is an independent negative predictor of serum adiponectin and suggest that adiponectin may represent a link between central obesity and insulin resistance. Estradiol levels are also independently and negatively associated with adiponectin, and postmenopausal women have higher adiponectin levels compared with premenopausal women. In contrast, serum adiponectin is not associated with circulating leptin, cortisol, and/or free testosterone levels. Finally, neither fasting for 48 h, resulting in a low leptin state, nor leptin administration at physiological or pharmacological doses alter serum adiponectin levels in healthy men and women. Adiposity (BMI, fat mass, percentage of fat) has been previously reported to be inversely associated with serum adiponectin in Japanese people (10, 11, 13 15), Pima Indians (2, 3, 12), and Caucasians (12, 16). The etiology of the decreased adiponectin mrna expression in adipose tissue (30, 31) and the lower serum adiponectin levels reported in obese subjects (10 12) remains to be fully elucidated, but it may occur due to the increased expression of TNF-, which reduces adiponectin expression and secretion from adipocytes in vitro (32, 33). Although it has also been shown that glucocorticoids inhibit adiponectin expression in vitro (21, 22), serum cortisol was not a significant predictor of serum adiponectin levels in our study. It is also possible that local cortisol production by adipose tissue in amounts that do not increase serum cortisol levels (34) may still alter adiponectin expression in visceral fat (21, 22), but this requires further investigation. In addition to confirming the association between body fat mass and adiponectin, we found that central obesity is an independent negative predictor of serum adiponectin. The relationship of adiponectin with WHR and WC appears to be stronger than that with fat mass or BMI, indicating that central fat distribution is a better determinant of circulating adiponectin than total fat mass. Similar to our findings, Cnop et al. (16) have reported recently that intraabdominal fat (but not BMI) was significantly and independently associated with adiponectin, whereas a Japanese study (15) has reported an inverse association between adiponectin and WHR in morbidly obese patients but not in overweight and moderately obese patients. A strong relationship of adiponectin with abdominal visceral fat has been reported in HIVinfected patients who develop fat redistribution secondary to antiretroviral treatment (35). Previous studies have also shown that adiponectin secretion is differentially regulated in cultured human omental and sc adipocytes (36), and that in obese diabetic subjects there is a more pronounced decrease of adiponectin mrna expression in omental fat compared with sc fat (31). In contrast to adiponectin, leptin correlates positively and strongly with both total and sc fat mass but less strongly with visceral fat (37, 38). We did not find a significant association between serum leptin and adiponectin levels in our sample, whereas Matsubara et al. (14) reported that leptin is a neg-

Gavrila et al. Predictors of Serum Adiponectin J Clin Endocrinol Metab, October 2003, 88(10):4823 4831 4829 FIG. 3. Second interventional study. Serum leptin and adiponectin levels at baseline, 6 h, and 12 h after administration of a pharmacological dose of r-methuleptin (0.3 mg/kg) in lean men (A) and obese men (B). *, P 0.01, compared with baseline value by Friedman tests. ative predictor of adiponectin levels independently of body composition, and Cnop et al. (16) detected a significant association between leptin and adiponectin in men and women only when considered separately but not in the entire study cohort. These discrepant results may be explained by different fat distribution characteristics of the populations studied, despite their similar overall adiposity, because it is well known that leptin is preferentially produced by sc adipose tissue (39), whereas adiponectin is mainly produced by visceral adipose tissue (36). We subsequently explored whether the strong negative association between adiponectin and central fat distribution might account for the association between central obesity/ visceral fat and insulin resistance (18). We found that the negative associations between adiponectin and HOMA-IR index, fasting insulin or glucose levels remain significant after adjusting for fat mass but were not significant after adjusting for WHR or WC, suggesting that the relationship between central adiposity and insulin resistance is mediated, at least in part, by adiponectin. Several studies have reported that insulin resistance is negatively associated with adiponectin levels independently of BMI (13, 14), and a recent study has shown that this association is also independent of the amount of intraabdominal fat (16). Our findings are in accordance with previous data showing that the increase in adiponectin after weight reduction by bariatric surgery is associated with improvement of insulin sensitivity, as estimated by HOMA-IR (40), and improved -cell function (41). We confirm that adiponectin levels are related to triglyceride levels and are a strong positive and independent determinant of HDL cholesterol in Caucasian women, as previously reported in Japanese subjects (11, 13, 24). We also confirm the positive association of adiponectin with apoa1 and the negative association with apob reported previously in the Japanese population (24). Further studies are needed to fully explore these relationships and their implications,

4830 J Clin Endocrinol Metab, October 2003, 88(10):4823 4831 Gavrila et al. Predictors of Serum Adiponectin because the low adiponectin levels in subjects with cardiovascular disease (10 12) may be related to insulin resistance and/or dyslipidemia. Serum estradiol is a strong negative determinant of adiponectin in this study, and postmenopausal women have lower estradiol but higher adiponectin levels compared with premenopausal women. Using a multiple linear regression model, we show that the difference in serum adiponectin related to menopausal status could be explained by the different estradiol levels but not by differences in age or body composition between the pre- and postmenopausal women. Despite the negative association between estradiol and adiponectin, it has been reported that women have higher adiponectin levels compared with men (10, 13, 16, 23), suggesting that in addition to estrogens, other gender-dependent factors, such as body fat distribution or androgen and progesterone levels, may be of relevance. Although women have more sc fat and men have more visceral fat (42, 17), previous studies have found that the gender difference in adiponectin levels is independent of total body fat or fat redistribution (13, 16, 23). In addition, androgens have been shown to decrease adiponectin levels in rodents (23), but we did not find an association between adiponectin and free testosterone levels in women. Testosterone levels are lower and have a narrow range in women, however, and thus further studies in men, who have much higher testosterone levels, are warranted. Defining the relationship between adiponectin and body fat distribution and/or sex hormones is of clinical relevance because it may explain the observed gender differences in insulin sensitivity (43). We also report a weak positive correlation with age, but in contrast to Cnop et al. (16), the association between adiponectin and age is only of borderline significance after adjusting for body composition. Thus, it remains possible that the previously reported significant association with age is due to either body composition changes and/or selection bias (16). Further studies are needed to elucidate this point. Importantly, we did not find any evidence for regulation of adiponectin by acute fasting or leptin in our interventional studies, in contrast to leptin s recently demonstrated ability to regulate several neuroendocrine axes (44). More specifically, neither fasting-induced decrease in leptin levels nor r-methuleptin administration at physiological or pharmacological doses acutely alter serum adiponectin levels. Of note, previous studies in humans and animals have shown that prolonged caloric restriction, which leads to significant and prolonged weight loss and decreased leptin levels, induces adiponectin expression (11, 45 47). Studies in rodents have shown that fasting for 2 d diminishes both adiponectin and leptin gene expression, which results in decreased serum leptin levels but unchanged serum adiponectin levels (19). This has been explained by the fact that serum adiponectin concentrations are nearly 1000 times higher than those of leptin, and thus fasting-induced changes of adiponectin expression in adipose tissue may not be directly translated into appreciable alterations of serum adiponectin levels over a short period of time (19). Moreover, leptin administration to lipoatrophic mice, which have low leptin and adiponectin levels and are highly insulin resistant, increases their leptin levels and improves insulin resistance but does not increase their low adiponectin levels (48). A combination of physiological doses of adiponectin and leptin completely reverses the insulin resistance of lipoatrophic mice, whereas adiponectin or leptin alone are only partially effective in improving insulin resistance in rodents (4, 48, 49) and humans with congenital lipodystrophy (50). Because leptin does not appear to regulate adiponectin in humans, leptin and adiponectin may represent two different and independent pathways that control insulin sensitivity and need to be studied further. In summary, we provide evidence for a strong association of serum adiponectin with fat mass, central fat distribution, and estradiol levels, whereas serum adiponectin is not associated with cortisol, leptin, or testosterone levels in humans. Importantly, we found that neither short-term fasting nor leptin administration alters serum adiponectin levels. The cross-sectional nature of our first study limits determinations of temporality or causality; therefore, further interventional studies are necessary to evaluate the relationship of serum adiponectin with central obesity and insulin resistance. Elucidation of the full spectrum of determinants of circulating adiponectin in humans should be the focus of future studies because it may prove to have major physiological and pathophysiological importance. Acknowledgments We thank the Beth Israel Medical Center, General Clinical Research Center staff for their assistance with nursing support, nutrition support, and specimen processing. Received February 11, 2003. Accepted July 9, 2003. Address all correspondence and requests for reprints to: Christos S. Mantzoros, M.D., Division of Endocrinology and Metabolism, Beth Israel Deaconess Medical Center, 330 Brookline Avenue, Stoneman 820, Boston, Massachusetts 02215. E-mail: cmantzor@bidmc.harvard.edu. This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grant DK 58785, National Institutes of Health (NIH) Grant K30 HL04095, and in part by NIH Grant MO1-RR01032. References 1. Tsao TS, Lodish HF, Fruebis J 2002 ACRP30, a new hormone controlling fat and glucose metabolism. Eur J Pharmacol 440:213 221 2. Stefan N, Vozarova B, Funahashi T, Matsuzawa Y, Weyer C, Lindsay RS, Youngren JF, Havel PJ, Pratley RE, Bogardus C, Tataranni PA 2002 Plasma adiponectin concentration is associated with skeletal muscle insulin receptor tyrosine phosphorylation, and low plasma concentration precedes a decrease in whole-body insulin sensitivity in humans. Diabetes 51:1884 1888 3. Lindsay RS, Funahashi T, Hanson RL, Matsuzawa Y, Tanaka S, Tataranni PA, Knowler WC, Krakoff J 2002 Adiponectin and development of type 2 diabetes in the Pima Indian population. Lancet 360:57 58 4. Yamauchi T, Kamon J, Waki H, Terauchi Y, Kubota N, Hara K, Mori Y, Ide T, Murakami K, Tsuboyama-Kasaoka N, Ezaki O, Akanuma Y, Gavrilova O, Vinson C, Reitman ML, Kagechika H, Shudo K, Yoda M, Nakano Y, Tobe K, Nagai R, Kimura S, Tomita M, Froguel P, Kadowaki T 2001 The fat-derived hormone adiponectin reverses insulin resistance associated with both lipoatrophy and obesity. Nat Med 7:941 946 5. Kubota N, Terauchi Y, Yamauchi T, Kubota T, Moroi M, Matsui J, Eto K, Yamashita T, Kamon J, Satoh H, Yano W, Froguel P, Nagai R, Kimura S, Kadowaki T, Noda T 2002 Disruption of adiponectin causes insulin resistance and neointimal formation. J Biol Chem 277:25863 25866 6. Bluher M, Michael MD, Peroni OD, Ueki K, Carter N, Kahn BB, Kahn CR 2002 Adipose tissue selective insulin receptor knockout protects against obesity and obesity-related glucose intolerance. Dev Cell 3:25 38 7. Yokota T, Oritani K, Takahashi I, Ishikawa J, Matsuyama A, Ouchi N, Kihara S, Funahashi T, Tenner AJ, Tomiyama Y, Matsuzawa Y 2000 Adiponectin, a new member of the family of soluble defense collagens, negatively regulates the growth of myelomonocytic progenitors and the functions of macrophages. Blood 96:1723 1732

Gavrila et al. Predictors of Serum Adiponectin J Clin Endocrinol Metab, October 2003, 88(10):4823 4831 4831 8. Ouchi N, Kihara S, Arita Y, Maeda K, Kuriyama H, Okamoto Y, Hotta K, Nishida M, Takahashi M, Nakamura T, Yamashita S, Funahashi T, Matsuzawa Y 1999 Novel modulator for endothelial adhesion molecules: adipocytederived plasma protein adiponectin. Circulation 100:2473 2476 9. Ouchi N, Kihara S, Arita Y, Nishida M, Matsuyama A, Okamoto Y 2001 Adipocyte-derived plasma protein, adiponectin, suppresses lipid accumulation and class A scavenger receptor expression in human monocyte-derived macrophages. Circulation 103:1057 1063 10. Arita Y, Kihara S, Ouchi N, Takahashi M, Maeda K, Miyagawa J, Hotta K, Shimomura I, Nakamura T, Miyaoka K, Kuriyama H, Nishida M, Yamashita S, Okubo K, Matsubara K, Muraguchi M, Ohmoto Y, Funahashi T, Matsuzawa Y 1999 Paradoxical decrease of an adipose-specific protein, adiponectin, in obesity. Biochem Biophys Res Commun 257:79 83 11. Hotta K, Funahashi T, Arita Y, Takahashi M, Matsuda M, Okamoto Y, Iwahashi H, Kuriyama H, Ouchi N, Maeda K, Nishida M, Kihara S, Sakai N, Nakajima T, Hasegawa K, Muraguchi M, Ohmoto Y, Nakamura T, Yamashita S, Hanafusa T, Matsuzawa Y 2000 Plasma concentrations of a novel, adipose-specific protein, adiponectin, in type 2 diabetic patients. Arterioscler Thromb Vasc Biol 20:1595 1599 12. Weyer C, Funahashi T, Tanaka S, Hotta K, Matsuzawa Y, Pratley RE, Tataranni PA 2001 Hypoadiponectinemia in obesity and type 2 diabetes: close association with insulin resistance and hyperinsulinemia. J Clin Endocrinol Metab 86:1930 1935 13. Yamamoto Y, Hirose H, Saito I, Tomita M, Taniyama M, Matsubara K, Okazaki Y, Ishii T, Nishikai K, Saruta T 2002 Correlation of the adipocytederived protein adiponectin with insulin resistance index and serum highdensity lipoprotein-cholesterol, independent of body mass index, in the Japanese population. Clin Sci (Lond) 103:137 142 14. Matsubara M, Maruoka S, Katayose S 2002 Inverse relationship between plasma adiponectin and leptin concentrations in normal-weight and obese women. Eur J Endocrinol 147:173 180 15. Yang WS, Lee WJ, Funahashi T, Tanaka S, Matsuzawa Y, Chao CL, Chen CL, Tai TY, Chuang LM 2002 Plasma adiponectin levels in overweight and obese Asians. Obes Res 10:1104 1110 16. Cnop M, Havel PJ, Utzschneider KM, Carr DB, Sinha MK, Boyko EJ, Retzlaff BM, Knopp RH, Brunzell JD, Kahn SE 2003 Relationship of adiponectin to body fat distribution, insulin sensitivity and plasma lipoproteins: evidence for independent roles of age and sex. Diabetologia 46:459 469 17. Wajchenberg BL 2000 Subcutaneous and visceral adipose tissue: their relation to the metabolic syndrome. Endocr Rev 21:697 738 18. Ross R, Freeman J, Hudson R, Janssen I 2002 Abdominal obesity, muscle composition, and insulin resistance in premenopausal women. J Clin Endocrinol Metab 87:5044 5051 19. Zhang Y, Matheny M, Zolotukhin S, Tumer N, Scarpace PJ 2002 Regulation of adiponectin and leptin gene expression in white and brown adipose tissues: influence of 3-adrenergic agonists, retinoic acid, leptin and fasting. Biochim Biophys Acta 1584(2 3):115 122 20. Olefsky JM, Kruszynska YT 2001 Causes of insulin resistance. De Groot s textbook of endocrinology. 4th ed. Philadelphia: WB Saunders Co.; 780 782 21. Halleux CM, Takahashi M, Delporte ML, Detry R, Funahashi T, Matsuzawa Y, Brichard SM 2001 Secretion of adiponectin and regulation of apm1 gene expression in human visceral adipose tissue. Biochem Biophys Res Commun 288:1102 1107 22. Fasshauer M, Klein J, Neumann S, Eszlinger M, Paschke R 2001 Hormonal regulation of adiponectin gene expression in 3T3-L1 adipocytes. Biochem Biophys Res Commun 290:1084 1089 23. Nishizawa H, Shimomura I, Kishida K, Maeda N, Kuriyama H, Nagaretani H, Matsuda M, Kondo H, Furuyama N, Kihara S, Nakamura T, Tochino Y, Funahashi T, Matsuzawa Y 2002 Androgens decrease plasma adiponectin, an insulin-sensitizing adipocyte-derived protein. Diabetes 51:2734 2741 24. Matsubara M, Maruoka S, Katayose S 2002 Decreased plasma adiponectin concentrations in women with dyslipidemia. J Clin Endocrinol Metab 87:2764 2769 25. Segal KR, Van Loan M, Fitzgerald PI, Hodgdon JA, Van Itallie TB 1988 Lean body mass estimation by bioelectrical impedance analysis: a four-site crossvalidation study. Am J Clin Nutr 47:7 14 26. Clasey JL, Hatman ML, Kanaley J, Wideman L, Teates CD, Bouchard C, Weltman A 1997 Body composition by DEXA in older adults: accuracy and influence of scan mode. Med Sci Sports Exerc 29:560 567 27. Economos CD, Nelson ME, Fiatarone MA, Dallal GE, Heymsfield SB, Wang J, Yasumara S, Ma R, Vaswani AN, Russell-Aulet M, Pierson RN 1997 A multi-center comparison of dual energy x-ray absorptiometers: in vivo and in vitro soft tissue measurement. Eur J Clin Nutr 51:312 317 28. Howard G, Bergman R, Wagenknecht LE, Haffner SM, Savage PJ, Saad MF, Laws A, D Agostino Jr RB 1998 Ability of alternative indices of insulin sensitivity to predict cardiovascular risk: comparison with the Minimal Model. Ann Epidemiol 8:358:369 29. Radziuk J 2000 Insulin sensitivity and its measurement: structural commonalities among the methods. J Clin Endocrinol Metab 85:4426 4433 30. Hu E, Liang P, Spiegelman BM 1996 AdipoQ is a novel adipose-specific gene dysregulated in obesity. J Biol Chem 271:10697 10703 31. Statnick MA, Beavers LS, Conner LJ, Corominola H, Johnson D, Hammond CD, Rafaeloff-Phail R, Seng T, Suter TM, Sluka JP, Ravussin E, Gadski RA, Caro JF 2000 Decreased expression of apm1 in omental and subcutaneous adipose tissue of humans with type 2 diabetes. Int J Exp Diabetes Res 1:81 88 32. Maeda N, Takahashi M, Funahashi T, Kihara S, Nishizawa H, Kishida K, Nagaretani H, Matsuda M, Komuro R, Ouchi N, Kuriyama H, Hotta K, Nakamura T, Shimomura I, Matsuzawa Y 2001 PPAR ligands increase expression and plasma concentrations of adiponectin, an adipose-derived protein. Diabetes 50:2094 2099 33. Kappes A, Loffler G 2000 Influences of ionomycin, dibutyryl-cycloamp and tumour necrosis factor- on intracellular amount and secretion of apm1 in differentiating primary human preadipocytes. Horm Metab Res 32:548 554 34. Masuzaki H, Paterson J, Shinyama H, Morton NM, Mullins JJ, Seckl JR, Flier JS 2001 A transgenic model of visceral obesity and the metabolic syndrome. Science 294:2166 2170 35. Addy CL, Gavrila A, Tsiodras S, Brodovicz K, Karchmer AW, Mantzoros CS 2003 Hypo-adiponectinemia is associated with insulin resistance, hypertriglyceridemia, and fat redistribution in human immunodeficiency virusinfected patients treated with highly active antiretroviral therapy. J Clin Endocrinol Metab 88:627 636 36. Motoshima H, Wu X, Sinha MK, Hardy VE, Rosato EL, Barbot DJ, Rosato FE, Goldstein BJ 2002 Differential regulation of adiponectin secretion from cultured human omental and subcutaneous adipocytes: effects of insulin and rosiglitazone. J Clin Endocrinol Metab 87:5662 5667 37. Considine RV, Sinha MK, Heiman ML, Kriauciunas A, Stephens TW, Nyce MR, Ohannesian JP, Marco CC, McKee LJ, Bauer TL 1996 Serum immunoreactive-leptin concentrations in normal-weight and obese humans. N Engl J Med 334:292 295 38. Cnop M, Landchild MJ, Vidal J, Havel PJ, Knowles NG, Carr DR, Wang F, Hull RL, Boyko EJ, Retzlaff BM, Walden CE, Knopp RH, Kahn SE 2002 The concurrent accumulation of intra-abdominal and subcutaneous fat explains the association between insulin resistance and plasma leptin concentrations: distinct metabolic effects of two fat compartments. Diabetes 51:1005 1015 39. Gottschling-Zeller H, Birgel M, Scriba D, Blum WF, Hauner H 1999 Depotspecific release of leptin from subcutaneous and omental adipocytes in suspension culture: effect of tumor necrosis factor- and transforming growth factor- 1. Eur J Endocrinol 141:436 442 40. Faraj M, Havel PJ, Phelis S, Blank D, Sniderman AD, Cianflone K 2003 Plasma acylation-stimulating protein, adiponectin, leptin, and ghrelin before and after weight loss induced by gastric bypass surgery in morbidly obese subjects. J Clin Endocrinol Metab 88:1594 1602 41. Guldstrand M, Ahren B, Adamson U 2003 Improved -cell function after standardized weight reduction in severely obese subjects. Am J Physiol Endocrinol Metab 284:E557 E565 42. Blaak E 2001 Gender differences in fat metabolism. Curr Opin Clin Nutr Metab Care 4:499 502 43. Robertson MD, Livesey G, Mathers JC 2002 Quantitative kinetics of glucose appearance and disposal following a 13C-labelled starch-rich meal: comparison of male and female subjects. Br J Nutr 87:569 577 44. Chan JL, Heist K, DePaoli AM, Veldhuis JD, Mantzoros CS 2003 The role of falling leptin levels in the neuroendocrine and metabolic adaptation to shortterm starvation in healthy men. J Clin Invest 111:1409 1421 45. Yang WS, Lee WJ, Funahashi T, Tanaka S, Matsuzawa Y, Chao CL, Chen CL, Tai TY, Chuang LM 2001 Weight reduction increases plasma levels of an adipose-derived anti-inflammatory protein, adiponectin. J Clin Endocrinol Metab 86:3815 3819 46. Hulver MW, Zheng D, Tanner CJ, Houmard JA, Kraus WE, Slentz CA, Sinha MK, Pories WJ, MacDonald KG, Dohm GL 2002 Adiponectin is not altered with exercise training despite enhanced insulin action. Am J Physiol Endocrinol Metab 283:E861 E865 47. Berg AH, Combs TP, Du X, Brownlee M, Scherer PE 2001 The adipocytesecreted protein Acrp30 enhances hepatic insulin action. Nat Med 7:947 953 48. Colombo C, Cutson JJ, Yamauchi T, Vinson C, Kadowaki T, Gavrilova O, Reitman ML 2002 Transplantation of adipose tissue lacking leptin is unable to reverse the metabolic abnormalities associated with lipoatrophy. Diabetes 51:2727 2733 49. Schwartz MW, Baskin DG, Bukowski TR, Kuijper JL, Foster D, Lasser G 1996 Specificity of leptin action on elevated blood glucose levels and hypothalamic neuropeptide Y gene expression in ob/ob mice. Diabetes 45:531 535 50. Oral EA, Simha V, Ruiz E, Andewelt A, Premkumar A, Snell P 2002 Leptin replacement therapy for lipodystrophy. N Engl J Med 346:570 578