Objective: We investigated whether serum concentrations of adiponectin are determined by body fat distribution and compared the findings with leptin.
Research Methods and Procedures: Serum concentrations of adiponectin and leptin were measured by radioimmunoassay (n = 394) and analyzed for correlation with sex, age, and body fat distribution, i.e., waist-to-hip ratio, waist and hip circumference, and subcutaneous adipose tissue area of the lower leg as assessed by magnetic resonance imaging.
Results: After adjusting for sex and percentage of body fat, adiponectin was negatively (r = −0.17, p < 0.001) and leptin was positively (r = 0.22, p < 0.001) correlated with waist-to-hip ratio. Leptin, but not adiponectin, correlated with both waist (r = 0.49, p < 0.001) and hip circumference (r = 0.46, p < 0.001). Furthermore, leptin, but not adiponectin, correlated with the proportion of subcutaneous fat of the lower leg cross-sectional area (r = 0.37, p < 0.001).
Discussion: These data suggest that both adipocytokines are associated with central body fat distribution, and serum adiponectin concentrations are determined predominantly by the visceral fat compartment.
It is well established that obesity is the most important risk factor for the development of the metabolic syndrome, a cluster of diseases including insulin resistance, type 2 diabetes, dyslipidemia, hypertension, microalbuminuria, and atherosclerosis (1). In this context, factors released from adipose tissue are considered to be responsible for mediating insulin resistance. Moreover, it has been recognized that central obesity characterized by excess abdominal adipose tissue has a more severe impact on whole body insulin sensitivity than a peripherally pronounced distribution pattern with excess depots around hip and thigh (2). For many years, this has been explained by the higher lipolytic activity of intra-abdominal vs. subcutaneous fat (3, 4).
During the last decade, however, a more intense analysis of the secretory activity of adipose tissue revealed that not only free fatty acids, the product of lipolysis, but also a number of polypeptides are released from the adipocyte. Among these polypeptides, also known as adipocytokines, leptin and adiponectin are currently the most prominent ones. In humans, circulating leptin and adiponectin concentrations are strongly correlated with measures of obesity. Whereas serum leptin correlates positively (5), serum adiponectin correlates negatively with percentage of body fat (6). It was long believed that leptin impairs insulin signaling and confers insulin resistance. Leptin replacement in hypoleptinemic, lipoatrophic patients, however, improved glucose tolerance and decreased serum insulin concentrations (7, 8). This clearly suggests that leptin may possess insulin-sensitizing properties. Adiponectin, on the other hand, seems to have undisputed insulin-sensitizing properties in both humans (9) and animals (10, 11).
Conceivably, adipocytokines also play an important role in fat depot-related modulation of insulin sensitivity. It is unclear, however, whether circulating adipocytokine concentrations are determined by body fat distribution patterns in humans. We, therefore, quantified the levels of adiponectin (and leptin for direct comparison) in serum samples from 394 healthy, nondiabetic participants of the Tübingen Family Study for type 2 diabetes and analyzed them for correlation with sex, age, percentage body fat, waist-to-hip ratio (WHR)1, and waist and hip circumference. In a subgroup, the fat depot specificity of these hormones was further investigated by correlating their serum concentrations with a measure of the subcutaneous fat proportion of the lower leg using magnetic resonance imaging.
Research Methods and Procedures
In the Tübingen Family Study for type 2 diabetes, nondiabetic individuals with (and without) a family history of type 2 diabetes were recruited and metabolically characterized. For the present study, we included individuals based on availability of leptin and adiponectin measurements (Table 1). Subjects gave informed written consent before participating. The protocol was approved by the local ethical committee.
Table 1. Subject characteristics
Data are given as means ± SEM. p values were determined by Student's t test on log-transformed data.
Hormone concentrations were adjusted for percentage of body fat; p values were derived from multivariate linear regression analysis.
All determinations of blood parameters were performed after a 10-hour overnight fast. A 75-g oral glucose tolerance test (glucose and insulin determinations at 0, 30, 60, 90, and 120 minutes) was performed to exclude diabetes. Serum insulin was determined by microparticle enzyme immunoassay (Abbott Laboratories, Tokyo, Japan). Serum concentrations of leptin and adiponectin were determined by radioimmunoassay (Linco Research, St. Charles, MO).
Magnetic Resonance Imaging
Axial T1-weighted spin-echo images were recorded on a 1.5 T whole body imager (Magnetom Vision; Siemens, Erlangen, Germany). Volunteers were placed with the most extended part of their right lower leg in the center of the extremity coil. Determination of subcutaneous fat area was performed by manually drawing two regions of interest at the outer and inner boundaries of subcutaneous fat, as previously described (12), and calculating the percentage in relation to the complete cross-section of the lower leg.
For statistical analysis, all data were logarithmically transformed to approximate a normal distribution. The statistical software package JMP (SAS Institute, Cary, NC) was used to perform Student's t tests and to calculate multivariate linear models. To illustrate effects of a variable after adjusting for covariates, plots are shown as provided by the general linear model function of the JMP software package. A p value <0.05 was considered statistically significant.
In the 394 nondiabetic individuals examined, we found significant sex-dependent differences for percentage of body fat, WHR, fasting serum adiponectin, and leptin concentrations (Table 1). After adjustment for percentage of body fat, serum insulin was significantly higher in men than in women. On the other hand, the adjusted serum adiponectin and leptin concentrations were significantly lower in men than in women (Table 1). The female subgroup was, on average, 4 years older than the male subgroup. Plasma adiponectin concentrations were inversely correlated with measures of obesity and serum insulin in both women and men (simple linear correlations in Table 2).
Table 2. Simple linear correlations with plasma adiponectin concentration (gender segregated)
Body mass index
Percentage body fat
Fasting serum insulin
2-hour insulin concentration
Serum insulin area under the curve
In multivariate linear regression analyses, adiponectin and leptin were independently predicted by sex, percentage of body fat, and WHR, but not by age. As illustrated in Figure 1A, adiponectin concentrations correlated negatively with WHR after adjustment for sex and percentage of body fat. The likewise adjusted leptin concentrations were positively correlated with WHR (r = 0.22, p < 0.001). In a general linear model with fasting insulin concentrations and the insulin area under the curve as dependent variables, adiponectin remained significant after adjusting for sex, percentage of body fat, and WHR (p = 0.16 and p = 0.36, respectively).
To further explore depot specificity of these hormones, we analyzed their concentrations (again adjusted for sex and percentage of body fat) for correlation with waist circumference, as a measure for subcutaneous plus visceral fat, and hip circumference, as a measure exclusively for subcutaneous fat. As depicted in Figure 1 B and C, adiponectin revealed no significant correlation with either waist or hip circumferences. In contrast, leptin correlated significantly with both hip (r = 0.49, p < 0.001) and waist circumference (r = 0.46, p < 0.001). To further explore these differences, we quantified the subcutaneous fat area of the lower leg in transverse magnetic resonance images in a representative subgroup (99 individuals). In these subjects, the correlations of adiponectin and leptin with WHR were also present (data not shown). As illustrated in Figure 2A, serum adiponectin concentrations (adjusted for sex and percentage of body fat) clearly did not correlate with the subcutaneous fat area. The likewise adjusted leptin concentrations, however, were significantly correlated with subcutaneous fat (Figure 2B).
We addressed the question of whether serum concentrations of adiponectin are related to central body fat. For comparison, we also included leptin in the study. We used WHR as a parameter for central body fat. Adiponectin concentrations were negatively correlated with WHR, independent of gender and overall adiposity. This was demonstrated earlier in Pima Indians and a small group of whites, but no adjustments for gender and percentage of body fat were made (6). Consequently, one might speculate that serum concentrations of adiponectin positively correlate with peripheral fat, e.g., the fat depot around the hip, resulting in this negative correlation with WHR. However, the lack of a correlation between adiponectin and hip circumference (Figure 1C) clearly argues against this possibility. In agreement with the hypothesis that adiponectin is down-regulated in hypertrophic adipose tissue by a negative feedback mechanism (13, 14), we conclude that serum adiponectin concentrations are an inverse function of central body fat mass.
Central body fat consists of abdominal subcutaneous and intra-abdominal visceral adipose tissue. To determine predominant sites of release, we analyzed correlations with waist circumference, representing both subcutaneous and visceral adipose tissue, and hip circumference as a measure of subcutaneous fat alone. It is well established that leptin mRNA levels and secretion rates are higher in subcutaneous than in visceral adipose tissue (15, 16, 17, 18, 19). The positive correlation between leptin and waist circumference may, therefore, reflect the nonnegligible contribution of subcutaneous abdominal fat mass. However, lacking a direct measure of visceral fat, we cannot exclude the possibility that serum leptin concentrations are determined by visceral fat mass as well. In striking contrast, serum adiponectin levels showed no correlation with hip circumference. This strongly suggests that secretion of adiponectin into the bloodstream is not regulated by subcutaneous, but rather by visceral, adipose tissue. This is in agreement with in vitro findings showing lower adiponectin mRNA levels in omental vs. subcutaneous adipose tissue from type 2 diabetics (20). Moreover, the proportion of subcutaneous fat of the lower leg was not a significant determinant of serum adiponectin concentrations, again in contrast to leptin. This provides further evidence, albeit indirect, that adiponectin secretion in humans is a function of visceral fat mass.
In conclusion, our data suggest that serum concentrations of both adipocytokines are associated with central body fat distribution, that serum adiponectin concentrations are determined predominantly by the visceral fat compartment, and that serum leptin concentrations are determined mainly by subcutaneous adipose tissue, although a role of visceral fat cannot be excluded.
Dr. Stumvoll is supported by a Heisenberg grant from the Deutsche Forschungsgemeinschaft.