Plasma Adiponectin and Leptin Levels, Body Composition, and Glucose Utilization in Adult Women with Wide Ranges of Age and Obesity

Pathophysiology/Complications ORIGINAL ARTICLE

Plasma Adiponectin and Leptin Levels, Body Composition, and Glucose Utilization in Adult Women With Wide Ranges of Age and Obesity

1,2 3 ALICE S. RYAN, PHD RONALD L. GINGERICH, PHD he progressive impairment of carbo- 1,2 4 DORA M. BERMAN, PHD GRADY S. MENEILLY, MD hydrate intolerance with normal ag- 1,2 5 BARBARA J. NICKLAS, PHD JOSEPHINE M. EGAN, MD 3 6 ing is generally accompanied by ADHUR INHA PHD ARIUSH LAHI PHD T M S , D E , obesity and an increase in physical inac- tivity. These conditions promote hyperinsulinemia, insulin resistance, and the development of type 2 diabetes and car- OBJECTIVE — The purpose of this study was to determine the relationships between plasma diovascular disease. Obesity is a growing adiponectin and leptin levels, total and central obesity, and glucose utilization across the adult health concern because of these associ- age span. ated comorbidities. The adipose tissue secretes several RESEARCH DESIGN AND METHODS — We studied 148 women aged 18–81 years adipocytokines into the bloodstream with a BMI range of 17.2–44.3 kg/m2. Total percent body fat was determined by dual-energy X-ray absorptiometry and abdominal fat by computed tomography. Glucose tolerance in non– known to influence insulin resistance, in- type 2 diabetic volunteers was determined with an oral glucose tolerance test. Glucose utilization cluding leptin (1) and tumor necrosis fac- (M) was measured during the last 60 min of hyperinsulinemic-euglycemic clamps (240 pmol tor-␣ (2). The more recently identified mϪ2 minϪ1). Plasma adiponectin levels were measured by radioimmunoassay. The women adipocytokine, adiponectin, is a novel were separated into three age-groups: young, middle, and old (Ͻ40, 40–59, and Ն60 years, peptide expressed specifically and abun- respectively), as well as by glucose tolerance status. dantly in adipose tissue (3–6). Human adiponectin has 244 amino acids, and the RESULTS — Adiponectin concentrations did not differ by age-groups. There were significant molecular weight of the monomer is age effects for BMI, percent body fat, visceral fat, subcutaneous abdominal fat, VO2max, and M. Adiponectin levels were lower in the prediabetic women (n ϭ 18) than in the normal glucose- 26,413. However, it circulates in poly- tolerant women (n ϭ 108) and the women with type 2 diabetes (n ϭ 22) (both P Ͻ 0.05). meric form. Adiponectin appears to be Univariate correlations revealed significant negative relationships between plasma adiponectin linked to glucose homeostasis since levels and BMI, percent body fat, visceral fat, subcutaneous abdominal fat, fasting leptin, and plasma adiponectin levels are lower in di- fasting insulin and positive relationship with M (all P Ͻ 0.05). In a multiple stepwise regression abetic subjects (7,8) and are positively model to predict adiponectin, only M remained in the model at P Ͻ 0.001. Multivariate analyses correlated with glucose utilization (M) revealed a significant relation for M as a function of adiponectin, insulin, and VO2max. (9). Adiponectin levels are lower in obese (defined by BMI) than in nonobese sub- CONCLUSIONS — The data suggest that plasma adiponectin does not change with age but jects (6,10). Furthermore, adiponectin levels are negatively associated with percent body fat, visceral fat, subcutaneous abdominal fat, insulin, and leptin levels in women. Adiponectin is positively associated with M across the age levels increase after weight reduction span in women. (11). In addition, plasma adiponectin concentrations are negatively correlated Diabetes Care 26:2383–2388, 2003 with total body fat and waist-to-thigh ra- ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● tio (9). It is unknown whether adiponec- From the 1Department of Medicine, Division of Gerontology, University of Maryland School of Medicine, tin is related to direct measures of Baltimore, Maryland; the 2Baltimore Geriatric Research, Education, and Clinical Center (GRECC), VA abdominal obesity (e.g., visceral fat and Maryland Health Care System, Baltimore, Maryland; 3Linco Research, Saint Charles, Missouri; the 4Depart- subcutaneous abdominal fat), which is 5 ment of Medicine, University of British Columbia, Vancouver, Canada; the National Institute on Aging, known to be associated with insulin resis- Baltimore, Maryland; and the 6Department of Medicine, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts. tance (12). Moreover, to our knowledge, Address correspondence and reprint requests to Alice S. Ryan, PhD, Division of Gerontology, BT/18/GR, plasma adiponectin levels as a function of 10 N. Greene St., Baltimore Veterans’ Affairs Medical Center, Baltimore, MD 21201. E-mail: age have only been examined in a Japa- [email protected]. nese population (6). Although Arita et al. Received for publication 24 September 2002 and accepted in revised form 12 May 2003. M.S. is employed by Linco Research. R.L.G. is employed by, has been on an advisory panel, a standing (6) did not find an age effect, one might committee, or the board of directors of, and holds stock in Linco Research. expect that adiponectin would be lower in Abbreviations: FFM, fat-free mass; M, glucose utilization; NGT, normal glucose tolerance; SAT, subcu- older individuals because of an increase in taneous abdominal adipose tissue; VAT, visceral adipose tissue; VO2max, maximal oxygen uptake; WHR, obesity with aging. waist-to-hip ratio. A table elsewhere in this issue shows conventional and Syste`me International (SI) units and conversion We tested the hypothesis that adi- factors for many substances. ponectin levels would be negatively asso- © 2003 by the American Diabetes Association. ciated with visceral fat, positively

DIABETES CARE, VOLUME 26, NUMBER 8, AUGUST 2003 2383 Adiponectin, age, and insulin resistance associated with glucose utilization, and rowest point superior to the hip, divided volume. The blood samples were centri- decline with age. Additionally, we exam- by the circumference of the hip, and mea- fuged at 4°C, and a 1-ml aliquot of plasma ined the relationship between plasma adi- sured at its greatest gluteal protuberance was rapidly frozen (80°C) for subsequent ponectin and leptin levels. Thus, the to obtain waist-to-hip ratio (WHR) (15). hormone analysis. All determinations purpose of this study was to determine Fat mass, fat-free mass (FFM) as lean tis- were performed in duplicate. Plasma glu- the relationships among adiponectin lev- sue mass ϩ bone mineral content, was cose was measured with the glucose oxi- els, total and central obesity, and glucose determined by dual-energy X-ray absorp- dase method (Beckman Instruments, utilization across the adult age span. tiometry (DPX-L; Lunar Radiation, Madi- Fullerton, CA). Immunoreactive insulin son, WI) in 145 subjects. Computed and leptin were determined as previously RESEARCH DESIGN AND tomography scan of the abdomen (n ϭ described (13,20). Human adiponectin METHODS — One-hundred forty- 126) was performed using a PQ 6000 levels were determined by a newly devel- eight women (22 African Americans and Scanner (Marconi Medical Systems, oped specific human adiponectin radio- 126 Caucasians) with a mean age 50.3 Ϯ Cleveland, OH) to determine visceral ad- immunoassay (Linco Research, St. 1.25 years (range 18–81) and BMI ipose tissue (VAT) and subcutaneous ab- Charles, MO). The inter- and intra-assay 28.1 Ϯ 0.54 kg/m2 (range 17.2–44.3) dominal adipose tissue (SAT) areas as coefficient of variations were Ͻ10% at volunteered to participate in the study. previously described (16). ED20, ED50, and ED80 concentrations of Subjects were screened by medical his- the standard range (1–200 ng/ml). Since tory questionnaire, physical examination, Maximal oxygen uptake circulating levels are in ␮g/ml concentra- and fasting blood profile, and in 137 Maximal oxygen uptake (VO2max) was tions, samples were diluted 1/500 in assay women a graded exercise treadmill test measured using a continuous treadmill buffer before estimation. was performed in an attempt to exclude test protocol as previously described (17). those with cardiovascular disease. Validation for attainment of VO2max in- Statistical analyses Women were either sedentary (Ͻ20 min cluded meeting two of the following three The mean concentration of glucose and of aerobic exercise 2 times/week) or ath- criteria: 1) a plateau in oxygen uptake insulin was calculated for each sample letes (n ϭ 40; swimmers, runners, and with an increased workload as evidenced time point for the clamp. The trapezoidal triathletes), as previously described (13). by a difference in oxygen uptake of Ͻ2ml rule was used to calculate the integrated Ϫ Ϫ Unless the patient had known diabetes kg 1 min 1; 2) a respiratory exchange response over 30-min intervals for each and was being treated with oral hypogly- ratio Ͼ1.10; and 3) a maximal heart rate subject. The integrated response was di- cemic agents (n ϭ 9), subjects underwent within 10 bpm of the age-predicted max- vided by its time interval to compute a 75-g 2-h oral glucose tolerance test (14) imal value. mean concentrations. M for the last 60- with blood samples drawn at baseline and min interval of the clamp, calculated from at 120 min for measurement of plasma Hyperinsulinemic-euglycemic clamps the amount of glucose infused after cor- glucose levels. Volunteers were grouped All testing was performed in the morning rection for glucose-equivalent space (glu- by glucose tolerance status (normal glu- after a 12-h overnight fast. All subjects cose space correction), was used in the cose tolerance [NGT], prediabetic, and were weight stabilized (Ͻ1 kg) for at least analyses. Adiponectin and insulin con- diabetic) and by age (young Ͻ40 years, 2 weeks before metabolic testing. Periph- centrations were log transformed to middle 40–59 years, and old Ն60 years). eral tissue sensitivity to exogenous insulin achieve a more normal distribution. All None of the women with type 2 diabetes was measured in 140 women using the data were analyzed using SAS version 8.2 were receiving insulin or thiazolidine- hyperinsulinemic-euglycemic clamp (SAS, Cary, NC). The data were analyzed diones, 13 were newly diagnosed and not technique (18) with 10-min priming and for the total group, as well as by glucose on any medication, and 10 were on med- continuous infusion of insulin (240 pmol tolerance status (normal, prediabetic, and Ϫ Ϫ ication, including 5 on sulfonylureas and m 2 min 1, Humulin; Eli Lilly, India- diabetic) and age-groups (young Ͻ40 5 on biguanides. All subjects were non- napolis, IN) (1). Insulin concentrations years, middle 40–59 years, and old Ն60 smokers and had no evidence of cancer, have been previously reported for these years). Standard methods were used to liver, renal or hematological disease, or studies (13,19) and are ϳ480 pmol/l. In compute means, SE, and Pearson correla- other medical disorders with the excep- the diabetic volunteers, plasma glucose tion coefficients. The mean values for cat- tion of glucose intolerance. All methods was allowed to drop to 5.3 mmol/l and egorical variables were computed using and procedures for the study were ap- clamped at that level for the duration of ANOVA (proc generalized linear model proved by the Institutional Review Boards the study. The mean plasma glucose level using Bonferroni corrections). Multiple of the University of Maryland, the Univer- for all other volunteers averaged 5.2 regression models were used to examine sity of British Columbia, and the Johns mmol/l (13,19). The coefficient of vari- the effects of age, body fat, adiponectin, Hopkins Bayview Medical Center. Each ance did not exceed 5.0% in any of the leptin, fitness, basal glucose, and insulin participant provided written informed studies. levels on glucose utilization. All standard consent to participate in the study. tests were two tailed. Data are means Ϯ Analysis of blood samples SE, and P Ͻ 0.05 was regarded as statis- Body composition Blood samples were collected in heparin- tically significant. Height (cm) and weight (kg) were mea- ized syringes and placed in prechilled test sured to calculate BMI as weight (kg)/ tubes containing 1.5 mg EDTA/ml of RESULTS — The mean value and height (m2). In 126 women, waist blood and aprotonin (400 KIU/ml) in a range of the variables for the entire group circumference was measured at the nar- total volume that was 4% of the sample are presented in Table 1. As shown, there

2384 DIABETES CARE, VOLUME 26, NUMBER 8, AUGUST 2003 Ryan and Associates

Table 1—Physical and metabolic characteristics of the women

Women (N ϭ 148) Young, Ͻ40 years Middle, 40–59 years Old, Ն60 years Normal subjects (n)295821 Prediabetic subjects (n) 0 12 6 Nondiabetic subjects (n)1912 Age (years) 50 Ϯ 1 (18–81) 25.7 Ϯ 1.3 (30) 51.0 Ϯ 0.6 (79)* 67.2 Ϯ 0.9 (39)†‡ Weight (kg) 75.4 Ϯ 1.5 (38.6–124.4) 58.3 Ϯ 1.8 (30) 80.2 Ϯ 2.0 (79)* 78.7 Ϯ 2.4 (39)‡ BMI (kg/m2) 28.1 Ϯ 0.5 (17.2–44.3) 21.4 Ϯ 0.6 (30) 29.6 Ϯ 0.7 (79)* 30.0 Ϯ 0.9 (39)‡ Percent body fat 37.2 Ϯ 1.1 (10.0–56.6) 22.8 Ϯ 1.6 (30) 40.9 Ϯ 1.3 (79)* 42.6 Ϯ 1.8 (30)‡ VAT (cm2) 108.6 Ϯ 5.9 (13.6–277.5) 34.8 Ϯ 4.8 (28) 122.0 Ϯ 6.8 (68)* 148.6 Ϯ 11.0 (29)‡ SAT (cm2) 335.6 Ϯ 16.6 (27.3–734.6) 125.8 Ϯ 15.7 (28) 394.3 Ϯ 20.5 (69)* 398.3 Ϯ 26.7 (29)‡ Ϫ1 Ϫ1 Ϯ Ϯ Ϯ Ϯ VO2max (ml kg min ) 30.3 0.8 (12.5–61.0) 48.3 2.1 (29) 26.7 1.3 (78)* 21.8 1.7 (29)†‡ Fasting plasma leptin (ng/ml) 20.0 Ϯ 1.3 (0.9–68.3) 7.1 Ϯ 1.3 (29) 22.4 Ϯ 1.7 (77)* 26.3 Ϯ 2.8 (29)‡ Fasting plasma glucose (mmol/l) 5.7 Ϯ 0.1 (4.4–12.5) 5.2 Ϯ 0.1 (29) 5.6 Ϯ 0.1 (79) 6.4 Ϯ 0.3 (39)†‡ Fasting plasma insulin (pmol/l) 61 Ϯ 3 (12–188) 41 Ϯ 2 (29) 63 Ϯ 4 (75)* 71 Ϯ 5 (39)‡ ␮ Ϫ1 Ϫ1 Ϯ Ϯ Ϯ Ϯ M ( mol kgFFM min ) 55.1 1.89 (2.55–110.2) 72.4 2.8 (29) 53.9 2.6 (74)* 44.0 3.2 (37)†‡ Adiponectin (␮g/ml) 5.75 Ϯ 0.55 (0.21–36.22) 3.76 Ϯ 0.5 (30) 6.88 Ϯ 0.89 (79) 5.00 Ϯ 0.79 (39) Data are means Ϯ SE (range) or means Ϯ SE (n). *P Ͻ 0.05, young vs. middle; †P Ͻ 0.05, old vs. middle; ‡P Ͻ 0.05, old vs. young. was a wide range of age, total body obe- There were 108 women with NGT, 18 women are shown in Table 3. Age per se sity, fat distribution (waist circumference women were prediabetic, and 22 had di- was not correlated with plasma adiponec- 60–130.7 cm and WHR 0.68–1.03), cen- abetes. Glucose tolerance, as expected, tin concentrations. However, BMI, per- tral fat, physical fitness, glucose utiliza- was associated with many of the variables. cent body fat, waist circumference, WHR, tion, and adiponectin concentrations. Differences in adiponectin levels were ob- VAT (Fig. 1A), SAT (Fig. 1B), and plasma Table 1 also shows the physical and served as a function of glucose tolerance leptin were negatively correlated with adi- metabolic characteristics, as well as adi- status. Adiponectin levels were lower in ponectin. In a univariate analyses, VO2max ponectin concentrations, in women the prediabetic women than those with was also significantly associated with adi- across the three age-groups. There were NGT but were not different in women ponectin levels. We then tested if the re- significant effects of age-groups on BMI, with NGT from women with diabetes. lationship between adiponectin and percent body fat, VAT, SAT, VO2max, fast- Comparisons of adiponectin (ANOVA) VO2max was the same across the age- ing leptin, glucose, insulin concentra- between NGT and prediabetic women groups. A positive relationship was found tions, and M. Nevertheless, plasma and between prediabetic and diabetic in the middle and old age-groups, while adiponectin concentrations did not differ women were significant at the 0.05 level. an inverse relationship was found in the between the age-groups (young versus Even if prediabetic and diabetic subjects young group, which was significantly dif- middle, P ϭ 0.45; young versus old, P ϭ are matched by BMI and age, the differ- ferent from the two older groups. The two 0.85; middle versus old, P ϭ 0.55). ences between groups remain significant older groups were not different from each The physical and metabolic charac- for adiponectin (P ϭ 0.03). The relation- other. Therefore, we could not adjust for teristics as well as adiponectin concentra- ships between age, body composition, fit- VO2max across the age span. Finally, tions in women grouped by glucose ness, and glucose metabolism variables plasma glucose concentrations were not tolerance status are presented in Table 2. and plasma adiponectin concentrations in associated with adiponectin levels, but in-

Table 2—Physical and metabolic characteristics of women by glucose tolerance

Normal subjects Prediabetic subjects Diabetic subjects Age (years) 46.7 Ϯ 1.5 (108) 56.1 Ϯ 1.8 (18)* 62.1 Ϯ 2.8 (22)† Weight (kg) 71.9 Ϯ 1.7 (108) 88.8 Ϯ 2.7 (18)* 81.1 Ϯ 3.5 (22) BMI (kg/m2) 26.5 Ϯ 0.6 (108) 34.0 Ϯ 1.0 (18)* 30.7 Ϯ 1.1 (22)† Percent body fat 35.0 Ϯ 1.3 (108) 46.0 Ϯ 1.5 (17)* 41.5 Ϯ 2.0 (20) VAT (cm2) 95.2 Ϯ 6.3 (100) 172.4 Ϯ 12.2 (17)* 140.4 Ϯ 13.7 (8)† SAT (cm2) 301.7 Ϯ 18.8 (100) 457.9 Ϯ 26.6 (17)* 480.8 Ϯ 34.8 (9)† Fasting plasma leptin (ng/ml) 17.2 Ϯ 1.4 (108) 31.8 Ϯ 3.0 (17)* 29.0 Ϯ 4.6 (10)† Fasting plasma glucose (mmol/l) 5.3 Ϯ 0.04 (108) 5.5 Ϯ 0.1 (18) 8.0 Ϯ 0.4 (22)†‡ Fasting plasma insulin (pmol/l) 50 Ϯ 2 (107) 84 Ϯ 9 (18)* 96 Ϯ 8 (19)† ␮ Ϫ1 Ϫ1 Ϯ Ϯ Ϯ M ( mol kgFFM min ) 63.3 1.9 (103) 36.1 2.6 (17)* 29.4 3.2 (20)† Adiponectin (␮g/ml) 6.18 Ϯ 0.67 (108) 2.78 Ϯ 0.78 (18) 6.10 Ϯ 1.18 (22) Data are means Ϯ SE (n). *P Ͻ 0.05, prediabetic vs. normal subjects; †P Ͻ 0.05, diabetic vs. normal subjects; ‡P Ͻ 0.05, diabetic vs. prediabetic subjects.

DIABETES CARE, VOLUME 26, NUMBER 8, AUGUST 2003 2385 Adiponectin, age, and insulin resistance

Table 3—The relationships between plasma tolerance status was associated with adi- alysis revealed that only M is a signiﬁcant adiponectin concentrations and body compo- ponectin levels, with prediabetic volun- independent predictor of adiponectin. sition and glucose metabolism variables in teers having lower concentrations than Although we hypothesized that adi- women normal or diabetic volunteers. Visceral ponectin would decline with age because fat, subcutaneous abdominal fat, and fast- of the increase in obesity with age, our Pearson ing leptin were negatively associated with data failed to support this hypothesis. correlation adiponectin, and M was positively associ- Only one other study (6) has reported Variables coefﬁcients ated with adiponectin in women across that plasma adiponectin concentrations the age span. However, multivariate an- did not correlate with age when adjusted Age Ϫ0.02 Weight Ϫ0.30* BMI Ϫ0.32* Percent body fat Ϫ0.26* Waist circumference Ϫ0.28* WHR Ϫ0.30* VAT Ϫ0.28* SAT Ϫ0.27*

VO2max 0.25* Fasting plasma leptin Ϫ0.28* Fasting plasma glucose 0.04 Fasting plasma insulin Ϫ0.24* M 0.28* *P Ͻ 0.005. sulin and M did correlate with plasma adiponectin (Fig. 1C). In a multiple stepwise regression model to predict adiponectin, we put in the variables age, M, percent body fat, VAT, SAT, insulin, glucose, leptin, and VO2max. Only M remained in the model at P Ͻ 0.001. To examine the contribution of the variables on M, a backward elimi- nation multiple regression analysis was performed with the following dependent variables: age, percent body fat, VAT, SAT, adiponectin, insulin, leptin, and VO2max. In the final model, percent fat, VAT, SAT, adiponectin, insulin, glucose, and VO2max all contributed significantly at P Ͻ 0.05. However, examination of the parameter coefficients clearly indicated that there was collinearity between percent fat, VAT, and SAT. Therefore, we ran three separate models when only one of the estimates of fatness (percent fat, VAT, or SAT) was entered into the model. VAT and percent fat did not contribute significantly to M in their final two correspond- ing models, whereas adiponectin, insulin, Figure 1—A: Relationsihp of VAT area with plasma adi- and VO2max all contributed significantly at P Ͻ 0.05. In the model with SAT, SAT ponectin concentrations in women (r ϭϪ0.28, P ϭ replaced VO2max so that adiponectin, insulin, and SAT contributed significantly 0.002). B: Relationship of SAT area with plasma adiponectin to M. concentrations in women (r ϭ Ϫ0.27, P ϭ 0.001). C: Rela- CONCLUSIONS — The results of tion of M with plasma adi- our study indicate that age does not influ- ponectin concentrations in ence plasma adiponectin levels. Glucose women (r ϭ 0.28, P Ͻ 0.001).

2386 DIABETES CARE, VOLUME 26, NUMBER 8, AUGUST 2003 Ryan and Associates for BMI. When we divided our women lization (9). We have previously reported Acknowledgments— This study was sup- into three age-groups, it appeared that the (20) the association between plasma lep- Ͻ ported by funds from National Institutes of youngest age-group ( 40 years) had the tin levels, hyperglycemia, hyperinsulin- Health Grants K01-AG00747 (to A.S.R.), R29- lowest adiponectin levels and women in emia, and body composition. In the A614066 (to B.J.N.), and K01-AG00685 (to the 40–59 age-group had the highest, but current study, plasma adiponectin levels D.M.B.); the Department of Veterans’ Affairs, the differences in adiponectin concentra- are negatively correlated with plasma lep- Geriatrics Research, Education, and Clinical tions across age were not statistically sig- tin levels, which is similar to that reported Center at Baltimore, Maryland; a grant from nificant. The younger women were in Japanese women (10). We also found the Canadian Diabetes Association; and the In- tramural Research Program of the National In- significantly leaner, had less abdominal an inverse association between plasma fat, were more insulin sensitive, and had a stitute on Aging. adiponectin concentration and fasting Our appreciation is extended to those higher VO . Thus, we expected to have 2max insulin levels and a direct association be- women and men who participated in this to correct for differences in adiposity to tween adiponectin and glucose utiliza- study. We are grateful to the nurses in the Ge- determine whether adiponectin declined tion. Because the women in our study riatrics Services at the Baltimore VA Medical with age, but this was not necessary. We included trained athletes across the age Center, Johns Hopkins Bayview Medical Cen- also cannot explain the differential effect span who were very insulin sensitive (13) ter, and the University of British Columbia for of age in the relation between VO and technical assistance. We also thank Adeola 2max and also included sedentary women with adiponectin among the age-groups, and Donsumu, Dana Jones, Agnes Kohler, Gail further studies are required. and without diabetes, the highest and Chin, Elizabeth Misiura, Elizabeth Bannon, lowest values of insulin-stimulated glu- and Chris Lockhart for laboratory assistance. Several studies have examined the re- ϳ lationship between plasma adiponectin cose disposal were 50-fold different. We thank Denis Muller, Metabolism Section, concentrations and adiposity by BMI Thus, across a wide range of insulin sen- National Institute on Aging, National Insti- sitivity, higher adiponectin levels are as- tutes of Health, for statistical assistance and (6,7,9) and percent body fat (9). We con- Andrew P. Goldberg, MD, for support. firm the established relation between sociated with greater insulin-stimulated plasma adiponectin levels and BMI and glucose disposal. In our multiple regres- percent body fat with a range of body fat sion analyses, adiponectin was an independent predictor of M. References (10–57%) larger than that previously re- 1. Girard J: Is leptin the link between obesity ported. To our knowledge, the relation- Adiponectin levels have a reported and insulin resistance? Diabetes Metab 23: ship between plasma adiponectin genetic heritability of 46% (21). More- 16–24, 1997 concentrations and direct measures of over, several quantitative trait loci have 2. Hotamisligil GS, Shargill NS, Spiegelman central obesity was not previously been identified that have significant evi- BM: Adipose expression of tumor necrosis known. As we hypothesized, we found dence of linkage for obesity-related phe- factor-␣: direct role in obesity-linked in- that women with lower amounts of VAT notypes with serum adiponectin levels sulin resistance. Science 259:87–91, 1993 and SAT had higher adiponectin concen- (21). Genetic polymorphisms in the adi- 3. Scherer PE, Williams S, Fogliano M, Bal- dini G, Lodish HF: A novel serum protein trations in a cohort of women with a large ponectin gene have been identified variability in abdominal fat. Plasma adi- similar to Clq, produced exclusively in (22,23) and shown to be associated with adipocytes. J Biol Chem 270:26746– ponectin levels were also negatively cor- obesity and insulin resistance (24). Yet, 26749, 1995 related with measures of abdominal girth the mechanism by which adiponectin in- 4. Maeda K, Okubo K, Shimomura I, Fu- (waist circumference) and fat distribution fluences insulin sensitivity in humans is nahashi T, Matsuzawa Y, Matsubara K: (WHR). Thus, adiponectin concentra- unclear. There are several disease states in cDNA cloning and expression of a novel tions are associated with total and central which a decrease in adiponectin is associ- adipose specific collagen-like factor, adiposity. ated with insulin resistance (e.g., lipodys- apM1 (adipose most abundant gene tran- Although plasma adiponectin con- script 1). Biochem Biophys Res Commun trophy) (8). Adiponectin treatment 221:286–289, 1996 centrations have been shown to be lower reverses this in animal models (25). It in diabetic than nondiabetic subjects (7– 5. Nakano Y, Tobe T, Miura NC, Mazda T, would be interesting to see if adiponectin 9), our results are inconsistent with these Tomita M: Isolation and characterization treatment reverses insulin resistance in of GBP28: a novel gelatin-binding protein reports. However, in our study, predia- humans. purified from human plasma. J Biochem betic women have adiponectin levels two- 120:803–812, 1996 fold lower than those in normal or We conclude from our data that plasma adiponectin levels do not decline 6. Arita Y, Kihara S, Ouchi N, Takahashi M, diabetic women. Differences in adiponec- Maeda K, Miyagawa J, Hotta K, Shimo- tin concentrations between individuals with age, are negatively associated with mura I, Nakamura T, Miyaoka K, with NGT and those patients with diabe- visceral and subcutaneous abdominal fat Kuriyama H, Nishida M, Yamashita S, tes disappear after adjustment for M and plasma leptin, and are positively as- Okubo K, Matsubara K, Muraguchi M, and/or fasting plasma insulin concentra- sociated with glucose utilization across Ohmoto Y, Funahashi T, Matsuzawa Y: tions, which suggested that plasma adi- the adult age span. Additional studies are Paradoxical decrease of an adipose-spe- ponectin concentrations are more closely needed to determine whether changes in cific protein, adiponectin, in obesity. Bio- adiponectin concentrations with treat- chem Biophys Res Commun 257:79–83, related to insulin sensitivity and fasting 1999 insulinemia than glycemia (9). Plasma ments or interventions are associated with 7. Hotta K, Funahashi T, Arita Y, Takahashi adiponectin levels are inversely correlated improvements in insulin sensitivity in hu- M, Matsuda M, Okamoto Y, Iwahashi H, with fasting glucose (7) and insulin (7,9) mans and the mechanisms whereby this Kuriyama H, Ouchi N, Maeda K, Nishida and positively correlated with glucose uti- occurs. M, Kihara S, Sakai N, Nakajima T, Hase-

DIABETES CARE, VOLUME 26, NUMBER 8, AUGUST 2003 2387 Adiponectin, age, and insulin resistance

gawa K, Muraguchi M, Ohmoto Y, Naka- 881, 2001 D, Takahashi M, Kihara S, Tanaka S, Mat- mura T, Yamashita S, Hanafusa T, 14. Expert Committee on the Diagnosis and suzawa Y, Blangero J, Cohen D, Kissebah Matsuzawa Y: Plasma concentrations of a Classification of Diabetes Mellitus: Report A: The genetic basis of plasma variation in novel, adipose-specific protein, adi- of the Expert Committee on the Diagnosis adiponectin, a global endophenotype for ponectin, in type 2 diabetic patients. Ar- and Classification of Diabetes Mellitus. obesity and the metabolic syndrome. terioscler Thromb Vasc Biol 20:1595– Diabetes Care 20:1183–1197, 1997 J Clin Endocrinol Metab 86:4321–4325, 1599, 2000 15. Lohman TG, Roche AF, Martell R (Eds.): 2001 8. Haque W, Shimomura I, Matsuzawa Y, The Anthropometric Standardization Refer- 22. Takahashi M, Arita Y, Yamagata K, Mat- Garg A: Serum adiponectin and leptin ence Manual. Champaign, IL, Human Ki- sukawa Y, Okutomi K, Horie M, Shimo- levels in patients with lipodystrophies. netics Publishing, 1988 mura I, Hotta K, Kuriyama H, Kihara S, J Clin Endocrinol Metab 87:2395–2398, 16. Ryan AS, Nicklas BJ: Age-related changes Nakamura T, Yamashita S, Funahashi T, 2002 in fat deposition in mid-thigh muscle in Matsuzawa Y: Genomic structure and mu- 9. Weyer C, Funahashi T, Tanaka S, Hotta K, women: relationships with metabolic car- tations in adipose-specific gene, adi- Matsuzawa Y, Pratley RE, Tataranni PA: diovascular disease risk factors. Int J Obes ponectin. Int J of Obes 24:861–868, 2000 Hypoadiponectimia in obesity and type 2 Relat Metab Disord 23:126–132, 1999 23. Schaffler A, Langmann T, Palitzsch K-D, diabetes: close association with insulin re- 17. Ryan AS, Pratley RE, Elahi D, Goldberg Scholmerich J, Schmitz G: Identification sistance and hyperinsulinemia. J Clin En- AP: Resistive training increases fat-free and characterization of the human adipo- docrinol Metab 86:1930–1935, 2001 mass and maintains RMR despite weight cyte apM-1 promoter. Biochi Biophys Acta 10. Matsubara M, Maruoka S, Katayose S: In- loss in postmenopausal women. J Appl 1399:187–197, 1998 verse relationship between plasma adi- Physiol 79:818–823, 1995 24. Stumvoll M, Tschritter O, Fritsche A, ponectin and leptin concentrations in normal-weight and obese women. Eur J 18. DeFronzo RA, Tobin JD, Andres R: Glu- Staiger H, Renn W, Weisser M, Machicao Endocrinol 147:173–180, 2002 cose clamp technique: a method for quan- F, Ha¨ring H: Association of the T-G poly- 11. Yang W, Lee W, Funahashi T, Tanaka S, tifying insulin secretion and resistance. morphism in adiponectin (exon 2) with Matsuzawa Y, Chao C, Chen C, Tai T, Am J Physiol 237:E214–E223, 1979 obesity and insulin sensitivity: interaction Chuang L: Weight reduction increases 19. Ryan AS, Nicklas BJ, Berman DM, Ferrell with family history of type 2 diabetes. Di- plasma levels of an adipose-derived anti- RE: The insertion/deletion polymorphism abetes 51:37–41, 2002 inflammatory protein, adiponectin. J Clin of the ACE gene is related to insulin sen- 25. Yamauchi T, Kamon J, Waki H, Terauchi Endocrinol Metab 86:3815–3819, 2001 sitivity in overweight women. Diabetes Y, Kubota N, Hara K, Mori Y, Ide T, Mu- 12. Abate N, Garg A, Peshock RM, Stray-Gun- Care 24:1646–1652, 2001 rakami K, Tsuboyama-Kasaoka N, Exaki dersen J, Grundy SM: Relationships of 20. Ryan AS, Elahi D: The effects of acute hy- O, Akanuma Y, Gavrilova O, Vinson C, generalized and regional adiposity to in- perglycemia and hyperinsulinemia on Reitman ML, Kagechika H, Shudo K, sulin sensitivity in men. J Clin Invest 96: plasma leptin levels: its relationships with Yoda M, Nakano Y, Tobe K, Nagai R, 88–98, 1995 body fat, visceral adiposity, and age in Kimura S, Tomita M, Froguel P, Kad- 13. Ryan AS, Muller DC, Elahi D: Sequential women. J Clin Endocrinol Metab 81:4433– owaki T: The fat-derived hormone adi- hyperglycemic-euglycemic clamp to as- 4438, 1996 ponectin reverses insulin resistance sess ␤-cell and peripheral tissue: studies 21. Comuzzie AG, Funahashi T, Sonnenberg associated with both lipoatrophy and in female athletes. J Appl Physiol 91:872– G, Martin LJ, Jacob HJ, Black AEK, Maas obesity. Nat Med 7:941–946, 2001

2388 DIABETES CARE, VOLUME 26, NUMBER 8, AUGUST 2003