| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Departments of Nutritional Sciences and Medicine (A.J.G.H.), University of Toronto, Toronto, Ontario, Canada M5S 3E2; Division of Clinical Epidemiology (A.J.G.H., S.M.H.), University of Texas Health Science Center, San Antonio, Texas 78229-3900; Department of Biochemistry (D.B.) and Division of Public Health Sciences (L.E.W., A.B., C.L.), Wake Forest University School of Medicine, Winston-Salem, North Carolina 27157; State University of New York at Stony Brook (M.F.S.), Stony Brook, New York, New York 11794; Medical Genetics Institute (J.I.R., X.G., Y.-D.I.C.), Cedars-Sinai Medical Center, Los Angeles, California 90048; David Geffen School of Medicine at University of California Los Angeles (M.B.-A.), Los Angeles, California 90095; and Department of Preventive Medicine and Biometrics (J.M.N.), University of Colorado at Denver and Health Sciences Center, Denver, Colorado 80262
Address all correspondence and requests for reprints to: Steven Haffner, M.D., M.P.H., Division of Clinical Epidemiology, Department of Medicine, University of Texas Health Science Center, Mail Code 7873, 7703 Floyd Curl Drive, San Antonio, Texas 78229-3900. E-mail: haffner{at}uthscsa.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Objective: The aim of the study was to identify the correlates of adiponectin in 1636 nondiabetic Hispanics and African-Americans.
Design: A cross-sectional analysis of participants in the Insulin Resistance Atherosclerosis Family Study was conducted. SI was determined from frequently sampled iv glucose tolerance tests with minimal model analysis. Subcutaneous and visceral adipose tissues (SAT, VAT, respectively) were determined with computed tomography. Triglyceride, high-density lipoprotein, C-reactive protein, and adiponectin were measured in fasting samples. Generalized estimating equation (GEE) models were used to identify factors associated with adiponectin concentration.
Setting: A multicenter study using a family-based design was conducted.
Participants: A total of 1636 nondiabetic Hispanic and African-American subjects participated.
Main Outcome Measures: Circulating adiponectin concentration was measured.
Results: Age, female gender, high-density lipoprotein, SAT, and SI were positive independent correlates of adiponectin, whereas glucose, CRP, and VAT were negative independent correlates (all P < 0.05). Ethnicity was not an independent correlate of adiponectin in this model (P = 0.27); however, an ethnicity by VAT interaction term was retained, indicating a stronger negative association of VAT with adiponectin in African-Americans compared with Hispanics.
Conclusion: Directly measured SI, VAT, and SAT were independently correlated with adiponectin in Hispanic and African-American subjects. The inverse association of VAT with adiponectin was stronger in African-Americans compared with Hispanics, a finding that suggests possible ethnic differences in the effects of visceral obesity.
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Adiponectin circulates in high concentrations in the plasma. A number of factors have been studied for their role in determining circulating adiponectin concentrations, including genetic polymorphisms and chromosomal loci (16, 17), gender (18, 19), pharmacological agents (20, 21), sex hormones (22), insulin sensitivity (SI) (18, 23, 24), adiposity (1, 2, 3, 4, 5, 6, 7, 8, 18, 25, 26, 27), and ethnicity (11, 23, 28, 29, 30, 31, 32, 33, 34, 35). Although inverse associations with indirect measures of anthropometry and SI, including as body mass index (BMI), waist circumference, and insulin concentrations, have been consistently demonstrated (1, 2, 3, 4, 5, 6, 7, 8), fewer data on direct measures of adipose tissue depots and SI are available (18, 25, 26, 27). In addition, while a limited number of studies have reported lower adiponectin levels in African-Americans, Pima Indians, and South Asians compared with whites (11, 23, 28, 29, 30, 31, 32, 33, 34, 35), little information has been presented to explain the biological mechanisms underlying these ethnic differences. Furthermore, levels of adiponectin in Hispanics relative to other groups have received limited study (35). Understanding the determinants of nontraditional metabolic risk factors such as adiponectin in different ethnic groups is of value, given marked differences in the burden of chronic disease across these populations in Westernized countries (36, 37).
The objective of the present study was to identify the correlates of adiponectin concentration in a large group of Hispanic and African-American participants in the Insulin Resistance Atherosclerosis (IRAS) Family Study. Subjects in this study have been extremely well characterized, including directly measured visceral and sc adipose tissue (VAT and SAT, respectively) and SI, in addition to a wide range of other traditional and nontraditional risk factors for diabetes and CVD. We hypothesize that previously reported lower concentrations of adiponectin among African-Americans may be explained by unique metabolic and anthropometric characteristics of this population.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
126 mg/dl), and for whom information was available on adiponectin, SI, SAT, VAT, and other variables of interest. The institutional review boards at the respective institutions approved the protocol, and informed consent was given by each subject. Fat mass in the abdominal region was measured by computed tomography (CT) at both the L2/L3 and L4/L5 vertebral regions. A standardized protocol was used at each of the three clinical centers. Scans were read centrally at the University of Colorado Health Sciences Center, Department of Radiology, for SAT and VAT, with bowel fat subtracted out from measure of VAT. The L4/L5 measure was used in the present analysis. However, 45 subjects had data for the L2/L3 region, but not the L4/L5 region. Because adipose tissue areas at the L2/L3 and L4/L5 regions were highly correlated (Spearman correlation: 0.95 for SAT, 0.90 for VAT), data for these latter individuals for the L4/L5 region were imputed using a simple linear model (38, 39). SI was determined using a frequently sampled iv glucose tolerance test (40, 41, 42). SI, expressed as the SI index, was calculated using minimal model analysis (43).
Adiponectin concentration was measured using a RIA (LINCO Research, Inc., St. Charles, MO), with interassay and intraassay coefficients of variation (CVs) of 9.3%, 6.9%, and 9.3%, and 3.6%, 6.2%, and 1.8%, respectively, at concentrations of 1.5, 3.0, and 7.5 µg/ml. C-reactive protein (CRP) was measured using an in-house ultrasensitive competitive immunoassay (antibodies and antigens from Calbiochem, La Jolla, CA), with an interassay CV of 8.9% (44). Plasma glucose was measured using the glucose oxidase technique on an autoanalyzer. Impaired fasting glucose was defined as fasting glucose 100 and less than 126 mg/dl. Plasma insulin was measured using the dextran-charcoal RIA (45), which has a 19% interassay CV. Lipids were determined using standard laboratory procedures. Height and weight were measured to the nearest 0.5 cm and 0.1 kg, respectively. Waist circumference was measured at the natural indentation or a level midway between the iliac crest and lower edge of the rib cage if no natural indentation was present. Duplicate measures were made following a standardized protocol and averages used in the analysis. Resting blood pressure (systolic and fifth-phase diastolic) was recorded with a standard mercury sphygmomanometer after a 5-min rest. The average of the second and third measurements was used in the analysis. Ethnicity, smoking, and menopausal status were assessed by self-report.
Statistical analysis
SAS version 8.02 (SAS Institute, Inc., Cary, NC) was used for all statistical analyses. Initial univariate analyses were conducted to assess the similarity of associations of metabolic and anthropometric variables with adiponectin among Hispanic participants from the San Antonio and San Luis Valley centers. In addition, subgroup analyses, restricted to these two Hispanic groups, were conducted to determine whether there were any statistical interactions of clinical center (San Antonio vs. San Luis Valley) by VAT, SAT, or SI on adiponectin concentration.
GEEs were used to test for associations between adiponectin and the various independent variables while accounting for the familial correlations. In this application, the identity link was used. The familial correlation was modeled assuming exchangeable correlation for the working correlation matrix, and the variance was estimated via the sandwich estimator. GEEs are a standard approach to the analysis of correlated data such as family data and can be intuitively thought of as similar to linear regression models except that they account for the correlation among pedigrees. The natural logarithm of adiponectin was the response or dependent variable for all models, in which the log transformation was computed to best approximate the modeling assumptions (i.e. conditional normality, homogeneity of variance). The performance of each model was examined using standard regression diagnostics (i.e. colinearity, influence, homogeneity of variance, residuals). A multistage approach to model building was used to investigate the joint relationships of adiposity measures (VAT and SAT), SI, and ethnicity on the natural log of adiponectin. To minimize the influence of observations in the tail of the skewed distributions, VAT, SAT, and SI + 1, were also modeled on the natural log scale. The first model examined whether there were meaningful departures from linearity for each continuous variable adjusting for gender and ethnicity. Specifically, in addition to the linear effects, the corresponding quadratic effects of age, (log) VAT, (log) SAT, and (log) (SI + 1), were modeled as the squared deviation from the respective means. The nonlinear quadratic terms were statistically significant for age, (log) SAT, and (log) (SI +1), and were retained in subsequent modeling. The second model tested for the main effects of (log) VAT, (log) SAT, (log) (SI +1), and ethnicity on adiponectin. Covariates in this model included fasting glucose, gender, current cigarette smoking, systolic and diastolic blood pressure, and the natural logs of triglyceride (Tg), high-density lipoprotein (HDL), and CRP. The third set of models included tests for the following interactions: (log) VAT by (log) SAT, (log) VAT or (log) SAT by ethnicity, (log) VAT or (log) SAT by gender, and (log) (SI + 1) by ethnicity. A backward elimination procedure was used, with interactions maintained with a P value < 0.01 and main effect variables retained if any corresponding interaction was significant, or the main effect in the absence of a corresponding interaction was significant at the P value < 0.05. Results for the interaction variables indicated that only the VAT by ethnicity interaction was statistically significant under these criteria, and, therefore, this variable was retained in the final model. Separate models were also generated for the two ethnic groups using main effect variables identified during backward elimination for the full cohort. Finally, the aforementioned model was rerun using the VAT to SAT ratio in place of separate variables for VAT and SAT.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Anthropometric and metabolic characteristics are shown in Table 1
, stratified by gender and ethnicity. Significant gender differences included greater VAT and waist circumference in males, and greater SAT in females (all P < 0.0001). Males had higher systolic and diastolic blood pressure, and higher concentrations of glucose and Tg, while females had higher HDL, CRP, and adiponectin (all P < 0.0001). A number of ethnic differences were also noted, including greater VAT and SI, and higher concentrations of Tg and adiponectin among Hispanics, and higher HDL among African-Americans (all P < 0.0001).
|
). Inverse associations with other metabolic variables, including Tg, CRP, and blood pressure, although significant, were more modest (r = –0.11 to –0.17). The magnitude and direction of these associations were consistent in separate analyses by ethnic group, the only notable differences being stronger associations of SI and VAT with adiponectin among African-Americans compared with Hispanics (Table 2).
|
. Smoking status and Tg concentration were removed during the backward elimination procedure, and, thus, the final model included age, female gender, HDL, SAT, and SI as positive independent correlates, and glucose, CRP, and VAT as negative independent correlates. Ethnicity was not an independent correlate of adiponectin (P = 0.27). Quadratic terms for age, SAT, and SI were also retained in the model, indicating nonlinear associations of these variables with adiponectin concentration. The functional forms of these relationships for SAT and SI demonstrated positive associations of SI and SAT with adiponectin across the majority of the ranges of these variables, except at the highest levels of SI and the lowest levels of SAT (data not shown). In addition, the ethnicity by VAT interaction term was retained in the multivariable GEE model, indicating that the negative association of VAT with adiponectin differed between Hispanics and African-Americans.
|
, indicate a much stronger inverse association of VAT with adiponectin in African-Americans compared with Hispanics. Age, female gender, HDL, and SI were significant positive correlates of adiponectin in both groups. Fasting glucose was a weak inverse determinant in both Hispanics (P = 0.04) and African-Americans (P = 0.12), while CRP was inversely related in Hispanics (P = 0.0003), but only marginally so in African-Americans (P = 0.045).
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
A number of previous studies have reported lower concentrations of adiponectin in African-American subjects compared with Caucasians and other ethnic groups (11, 28, 30, 31, 32, 34, 35, 46). Various potential mechanisms have been proposed to explain these lower concentrations, including ethnic variation in adipose tissue mass or distribution (31), or differences in other metabolic syndrome components known to be related to adiponectin concentrations (32, 34). It is also likely that genetic factors are playing a role in determining adiponectin levels in African-Americans. For example, it has been previously demonstrated that the heritability of adiponectin levels is high (h2 = 0.82) among African-American subjects in the IRAS Family Study cohort (35, 46). Furthermore, it was recently reported that variants of the adiponectin gene are associated with adiponectin concentrations in white and African-American adolescents in the Princeton School District Study (17). In the present study, univariate analysis confirmed that adiponectin concentrations were significantly lower in male and female African-American participants compared with their Hispanic counterparts. Furthermore, the univariate correlates of adiponectin in African-Americans included HDL and SI (positive), as well as BMI, waist circumference, fasting glucose and insulin, VAT, Tg, and CRP (inverse), with the direction and magnitude of these correlations being very similar between the two ethnic groups. However, African-American subjects did not have significantly lower adiponectin concentrations compared with Hispanics when an ethnicity by VAT interaction term was included in the model. This observation indicates that the inverse association of VAT with adiponectin was modified by ethnicity and highlights a substantially stronger adiponectin lowering effect of VAT in African-American subjects compared with Hispanics. Therefore, the extensive risk factor characterization in the IRAS Family Study and the large sample size has allowed a determination of the possible factors that explain ethnic differences in adiponectin concentrations.
Individuals with visceral obesity, documented either indirectly using measures such as waist circumference or directly using imaging technologies such as CT, are at increased risk for type 2 diabetes and CVD (47). Furthermore, these individuals carry a heavier burden of metabolic risk factors, including insulin resistance, dyslipidemia, elevated free fatty acids, and subclinical inflammation, relative to those with lower body obesity (47, 48). More recently, hypoadiponectinemia has been added to the list of metabolic disorders associated with increased visceral adiposity. Significant inverse associations of adiponectin with surrogate measures of VAT, including waist circumference and waist-to-hip ratio, have been consistently reported in studies of human subjects, with associations between waist or waist-to-hip ratio and adiponectin often being of greater magnitude than associations with measures of overall body mass (1, 2, 3, 4, 5, 6, 7, 8). These observations have been extended by the results of a growing body of studies in which adipose tissue depots have been directly assessed using technologies such as CT (18, 25, 26, 27). In these studies, inverse correlations with adiponectin have been reported for both SAT and VAT in univariate analysis, with the magnitude of the associations often stronger for VAT. These findings are consistent with those from the current study, which involved a large sample of nondiabetic Hispanic and African-American adults. Studies of cultured visceral and sc adipocytes from humans and animal models have indicated that while both adiponectin gene expression and secretion are higher in visceral adipocytes (49, 50), adiponectin secretion from adipocytes in this depot decreases with increasing BMI (48, 51). This observation suggests a possible mechanism to explain the hypoadiponectinemia that characterizes total, and especially abdominal, obesity.
The majority of previous studies using direct assessments of adipose tissue have further indicated that the inverse associations of VAT most often remain statistically significant in multivariable analysis, whereas the magnitude and significance of SAT were often greatly attenuated. In contrast, in the present study, the association of SAT with adiponectin was significantly positive after adjustment for VAT and other variables. This observation is possibly a reflection of the adiponectin secretory capacity of adipose tissue after accounting for the down-regulating effects of visceral obesity.
SI has consistently been shown to be positively associated with adiponectin levels independently of other factors, most notably obesity measures. However, the majority of previous studies have documented this independent association using surrogate measures of SI, such as fasting insulin or indices derived from oral glucose tolerance tests. A limited number of studies have reported positive associations between adiponectin and direct measures of SI using glucose or insulin clamps, or frequently sampled iv glucose tolerance tests (23, 24) adjusted for simple measures of obesity, including BMI and waist circumference. Very limited data are available that confirm this relationship after considering direct measures of adipose tissue distribution (18). The results of the present study, in which a strong association between adiponectin and directly measured SI was documented in both Hispanics and African-Americans after adjustment for a broad range of potential confounding factors, most notably directly measured VAT and SAT, represent an important extension of this literature.
In addition to significant associations with VAT and SAT, multivariable analysis in the present study indicated that female gender, HDL (positive), and CRP (inverse) were independently associated with adiponectin concentrations. Previous studies have been remarkably consistent in documenting lower adiponectin levels in males (18, 19), an effect that may be related to the adiponectin-suppressing effect of androgens (19, 52). A strong, positive association of adiponectin with HDL has also been frequently reported in the literature (53, 54, 55, 56). The current findings and those from previous studies have shown that this association is maintained after adjustment for SI, which suggests that adiponectin may have an independent effect on hepatic lipoprotein metabolism (18, 24). A recent paper reported that hypoadiponectinemia was independently associated with increased post-heparin plasma hepatic lipase activity, which in turn could result in reductions in HDL cholesterol (57). Alternatively, this association might be explained by increases in the ligand activity of peroxisome proliferator-activated receptor-
by adiponectin (50). Activation of peroxisome proliferator-activated receptor- is known to influence the expression of genes encoding for proteins involved in HDL metabolism (58). Finally, a growing body of literature has reported inverse associations of adiponectin with CRP and other markers of subclinical inflammation (59), observations that highlight the antiinflammatory properties of adiponectin (60).
Two issues should be kept in mind when reviewing the results of this study. First, adiponectin was measured using a commercially available RIA that determines total circulating adiponectin concentrations. Adiponectin isoform distribution (i.e. the amounts or proportions of the high and low-molecular weight forms of the protein) cannot be determined using this assay. Recent evidence suggests that the adiponectin isoform distribution, particularly the proportion of the high-molecular weight form relative to the total (an index referred to as SA), may have important physiological implications with regard to changes in SI (1) and possibly ethnic variation in adiponectin levels (61). Second, the cross-sectional design of this study does not allow for conclusions regarding the temporality of these associations. However, evidence from previous studies of animal models and humans has suggested that a decline in adiponectin concentrations may be an early event in the pathogenesis of diabetes (62), and that higher baseline adiponectin concentrations are associated with improvements over time in SI and HDL (54, 63, 64), but not with changes in body mass (54, 65). Additional human studies using longitudinal designs are needed to further clarify these relationships.
In conclusion, we found that age, female gender, HDL concentration, SAT, and SI were significant, independent positive correlates of adiponectin, while VAT, CRP, and glucose concentrations were significant negative correlates in a large, well-characterized sample of nondiabetic Hispanic and African-American adults. Understanding the factors related to variation in adiponectin concentrations is of value in light of evidence from prospective studies indicating that hypoadiponectinemia is a significant predictor of type 2 diabetes and possibly CVD, independent of other risk factors (9, 10, 11, 12, 13, 14, 15). A number of strategies have been successful in increasing adiponectin levels, including weight loss and treatment with insulin sensitizing agents, such as thiazolidinediones (20, 21, 66). Our observation that increased VAT had a stronger adiponectin lowering effect in African-American subjects compared with Hispanics is of particular interest and highlights the importance of preventing weight gain, especially in the abdominal regions, in this population.
Furthermore, this paper provides evidence needed to motivate further studies to understand how and why ethnicity alters this relationship and ultimately influences diabetes risk.
|
Footnotes |
|---|
Disclosure Statement: The authors have nothing to disclose.
First Published Online April 10, 2007
Abbreviations: BMI, Body mass index; CRP, C-reactive protein; CT, computed tomography; CV, coefficient of variation; CVD, cardiovascular disease; GEE, generalized estimating equation; HDL, high-density lipoprotein; IRAS, Insulin Resistance Atherosclerosis; SAT, sc adipose tissue; SI, insulin sensitivity; Tg, triglyceride; VAT, visceral adipose tissue.
Received November 28, 2006.
Accepted April 4, 2007.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
agonist, rosiglitazone, increases plasma levels of adiponectin in type 2 diabetic patients. Diabetes Care 25:376–380 activation and high-density lipoprotein metabolism. Am J Cardiol 88:24N–29NThis article has been cited by other articles:
 |
|
 |
|
  L. Frederiksen, T. L. Nielsen, K. Wraae, C. Hagen, J. Frystyk, A. Flyvbjerg, K. Brixen, and M. Andersen Subcutaneous Rather than Visceral Adipose Tissue Is Associated with Adiponectin Levels and Insulin Resistance in Young Men J. Clin. Endocrinol. Metab., October 1, 2009; 94(10): 4010 - 4015. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. H. Jain, J. M. Massaro, U. Hoffmann, G. A. Rosito, R. S. Vasan, A. Raji, C. J. O'Donnell, J. B. Meigs, and C. S. Fox Cross-Sectional Associations Bet ween Abdominal and Thoracic Adipose Tissue Compartments and Adiponectin and Resistin in the Framingham Heart Study Diabetes Care, May 1, 2009; 32(5): 903 - 908. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. A. Aguilar-Salinas, E. G. Garcia, L. Robles, D. Riano, D. G. Ruiz-Gomez, A. C. Garcia-Ulloa, M. A. Melgarejo, M. Zamora, L. E. Guillen-Pineda, R. Mehta, et al. High Adiponectin Concentrations Are Associated with the Metabolically Healthy Obese Phenotype J. Clin. Endocrinol. Metab., October 1, 2008; 93(10): 4075 - 4079. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. R. Kizer, J. I. Barzilay, L. H. Kuller, and J. S. Gottdiener Adiponectin and Risk of Coronary Heart Disease in Older Men and Women J. Clin. Endocrinol. Metab., September 1, 2008; 93(9): 3357 - 3364. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M.-F. Hivert, L. M. Sullivan, C. S. Fox, D. M. Nathan, R. B. D'Agostino Sr., P. W. F. Wilson, and J. B. Meigs Associations of Adiponectin, Resistin, and Tumor Necrosis Factor-{alpha} with Insulin Resistance J. Clin. Endocrinol. Metab., August 1, 2008; 93(8): 3165 - 3172. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. A. Kosmiski, P. Bacchetti, D. P. Kotler, S. B. Heymsfield, C. E. Lewis, M. G. Shlipak, R. Scherzer, and C. Grunfeld Relationship of Fat Distribution with Adipokines in Human Immunodeficiency Virus Infection J. Clin. Endocrinol. Metab., January 1, 2008; 93(1): 216 - 224. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
