HIV Articles  
Subcutaneous (lipoatrophy) Rather than Visceral Adipose Tissue Is Associated with Adiponectin Levels and Insulin Resistance in Young Men
  Journal of Clinical Endocrinology & Metabolism Oct 2009
"Focusing on insulin resistance, SAT rather than VAT and TFA appeared as an independent predictor of a higher HOMA-IR."
-from Jules and based on the previous article I sent that insulin resistance can impair exercise capacity, together these suggest insulin resistance can lead to frailty.

L. Frederiksen, T. L. Nielsen, K. Wraae, C. Hagen, J. Frystyk, A. Flyvbjerg, K. Brixen and M. Andersen
Department of Endocrinology (L.F., T.L.N., K.W., C.H., K.B., M.A.), Odense University Hospital, and Institute of Clinical Research, University of Southern Denmark, DK-5000 Odense C, Denmark; and The Medical Research Laboratories (J.F., A.F.), Clinical Institute and Medical Department M (Diabetes and Endocrinology), Aarhus University Hospital, DK-8000, Aarhus C, Denmark
Introduction: Studies on the association between adiponectin, body composition, and insulin resistance (IR) have been conflicting.
Aim: Our aim was to evaluate the impact of body composition on adiponectin and IR determined by homeostasis model assessment (HOMA) in a population-based study on relatively healthy young men, minimizing the possible effects of age, obesity, and severe comorbidity.
Design, Methods, and Subjects: A population-based, cross-sectional study of 783 men aged 20-29 yr, randomly drawn from the Danish Central Personal Registry. Adiponectin was assessed using an in-house assay, and IR was determined using HOMA. Central fat mass (CFM) and lower extremity fat mass (LEFM) was measured by dual-energy x-ray absorptiometry, and visceral adipose tissue (VAT), sc adipose tissue (SAT), and thigh fat area (TFA) were assessed by magnetic resonance imaging.
Results: Using multiple linear regression analysis, adiponectin correlated negatively with CFM (r = -0.27; P < 0.001) and SAT (r = -0.20; P < 0.001) and positively with LEFM (r = 0.19; P < 0.001) and TFA (r = 0.18; P < 0.001), whereas VAT did not associate significantly. In multiple linear regression analysis, HOMA-IR (dependent variable), correlated significantly with CFM (r = 0.27; P < 0.001) and SAT (r = 0.15; P < 0.001), whereas LEFM, VAT, or TFA did not correlate. Adiponectin was an independent predictor of HOMA-IR in both analyses (r = -0.14; P < 0.001).
Conclusion: SAT rather than VAT was inversely associated with adiponectin levels, and, interestingly, fat on the lower extremities was positively associated with adiponectin. Focusing on insulin resistance, SAT rather than VAT and TFA independently predicted a higher HOMA-IR. The observation that adiponectin was independently associated with lower HOMA-IR must be repeated in other populations.
Adiposity is a major component of the metabolic syndrome and a risk factor for the development of type 2 diabetes and cardiovascular disease (1, 2).
Adipose tissue is considered an endocrine organ with the secretion of bioactive molecules known as adipokines. Adipokines affect the metabolic function of other tissues and may thereby mediate the systemic effect of obesity on health (3).
The metabolic syndrome and obesity are accompanied by decreased circulating levels of the antiinflammatory adipokine adiponectin (4, 5). Several studies have demonstrated that serum adiponectin inversely associates with insulin resistance (IR) (6, 7, 8, 9) even independently of obesity (10). The possible increase in insulin sensitivity appears to be mediated by adiponectin receptors adipoR1 and adipoR2 in the skeletal muscle and liver, respectively (11). Serum adiponectin has been shown to correlate inversely with central fat (12), and these findings have been interpreted as negative consequences of visceral adiposity.
To study the association between fat and serum adiponectin, it is important to control for ethnicity, gender, and age differences in the amount of visceral and sc fat (13). However, previous studies have investigated older populations with a large age span (14, 15), included preferably overweight and obese subjects (16, 17), or reported combined analyses of men and women (12, 16, 18), although serum adiponectin levels are significantly lower in men compared with women (19).
The aim of this study was first to investigate whether serum adiponectin is associated with fat mass in the Odense Androgen Study population of relatively lean, healthy young men with a narrow age span.
A second aim was to investigate the association between serum adiponectin and fat distribution by using dual-energy x-ray absorptiometry (DXA) to measure central and lower extremity fat mass (CFM and LEFM) and magnetic resonance imaging (MRI) to assess the areas of sc adipose tissue (SAT), visceral adipose tissue (VAT), and thigh fat (TFA).
Third, we examined the relationship between homeostasis model assessment of insulin resistance (HOMA-IR) and body composition using both DXA and MRI to examine whether adiponectin was an independent predictor of HOMA-IR.
In this population-based, cross-sectional study, we found a significant and negative association between serum adiponectin and CFM in young men at the early stages of visceral fat accumulation (20, 23). SAT rather than VAT was inversely associated with serum adiponectin in these healthy and relatively lean young men.
In recent years, CFM has been extensively studied due to the strong association between abdominal obesity, metabolic syndrome, and cardiovascular disorders. The quantity of VAT and SAT, together constituting CFM, varies between individuals and it appears that the difference in biological effects of the various fat deposits are related to the anatomical sites (13).
The amount of adiponectin secreted from SAT and VAT varies. It has been shown in vitro that adiponectin gene expression is lower in VAT than in SAT, suggesting SAT to be more important for circulating adiponectin (24, 25). In vivo, however, the association between SAT, VAT, and adiponectin is not clear. In line with our findings, a recent study on 359 older, lean Asian males found only SAT and not VAT to negatively correlate with adiponectin (15). The association is present even in older subjects with more extensive VAT accumulation. Farvid et al. (14) found in mostly overweight and obese men that SAT rather than VAT was negatively associated with adiponectin.
Four studies have reported findings contradictory to ours. Two studies found the exact opposite results, that SAT correlated positively with serum adiponectin (16, 26), and two studies reported that VAT was negatively associated with serum adiponectin, whereas SAT had no significant impact (12, 18). These studies, however, presented combined analyses of male and female subjects disregarding the well known difference in serum adiponectin levels between men and women (the percentage of men varied from 37-58%). In contrast, Valsamakis et al. (27), who included only men, reported no significant relationship between adiponectin and VAT in 28 elderly, overweight, Caucasian men. SAT was not measured, and the lack of significance may be related to the limited size of the study.
There is a general consensus concerning the inverse relationship between IR and serum adiponectin (8, 9); our results agree even after adjustment for total body fat. The observation that SAT rather than VAT correlated significantly and negatively with adiponectin may be reflected in the association between SAT and IR. In multivariate analysis, CFM was positively associated with IR mirroring the negative relationship between fat mass and serum adiponectin. Incorporating MRI data in linear regression analysis, SAT but not VAT correlated positively with IR. SAT was corresponding to the association between SAT and adiponectin, an independent positive predictor of IR compared with VAT. In agreement with our results, two elegant studies have reported a positive association between SAT and IR using the hyperinsulinemic euglycemic glucose clamp method (28, 29). Abate et al. (28) included 39 males: 13 obese, 13 lightly overweight, and 13 lean middle-aged men thereby including men with greater amounts of SAT and VAT compared with our study. When focusing on SAT, children accumulate fat in the abdominal sc compartment and therefore have a smaller proportion of visceral fat compared with adults. In further support of our findings, Maffeis et al. (30) reported that SAT rather than VAT correlated inversely with insulin sensitivity in overweight prepubertal children.
We observed a positive association between fat on lower extremities, measured by both DXA and MRI, and serum adiponectin, suggesting a positive effect of LEFM. There was, however, no association between HOMA-IR and LEFM/TFA using multivariate analyses.
Large hip circumference has been found to associate with lower morbidity and mortality with regard to diabetes and cardiovascular disease in women (31). The possible protective effect of a large hip circumference (LEFM/TFA) with regard to IR; may be due to a high lipoprotein lipase activity and low fatty acid turnover of gluteofemoral adipose tissue (32). It has been suggested that peripheral sc fat mass can reduce IR through the decrease of visceral fat (13). In line with these findings, the relative lack of sc fat on the extremities seen in e.g. lipodystrophy has been associated with increased IR (13). Patients with the Prader-Willi syndrome have high fat percentages and relatively high levels of serum adiponectin and reduced IR. Prader-Willi syndrome patients are characterized by low amounts of central fat and quite large amounts of peripheral fat compared with BMI-matched controls (33, 34).
The associations between the variables were significant, although relatively modest, explaining 1-10% of the variation in adiponectin and HOMA-IR.
Our study has some limitations. The study design was cross-sectional and thus does not allow us to conclude cause and effect. Second, our participants were a Caucasian cohort, and the results may not be projected to other populations. Similar findings, however, have been reported in an Asian population (15).
Our study, however, also had several strengths. First, we included a population-based sample of young men matching the background population with regard to demography, smoking, social status, and health as previously described (20). Second, the young and relatively lean population with a narrow age span minimized the effect of aging, obesity, and the possible influences of comorbidity. Third, body composition was evaluated using two different image techniques to give additional information on the distribution of body fat in participants with similar waist circumference.
We conclude that SAT rather than VAT is inversely associated with serum adiponectin levels. Moreover, LEFM was positively associated with serum adiponectin. Focusing on insulin resistance, SAT rather than VAT and TFA appeared as an independent predictor of a higher HOMA-IR.

Serum adiponectin and body composition
In the univariate analyses, serum adiponectin correlated negatively and significantly with waist circumference, waist to hip ratio, BMI, total fat mass, and LEFM (Table 2Go).
Using multiple linear regression analysis with serum adiponectin as the dependent variable, CFM was significantly negatively correlated with adiponectin. By contrast, LEFM correlated significantly and positively with serum adiponectin (Table 3Go).
Serum adiponectin correlated significantly and negatively with visceral fat and sc fat in the univariate analyses (Table 2Go). TFA did not correlate significantly with adiponectin in the univariate analysis.
In the multiple linear regression analysis with serum adiponectin as the dependent variable, SAT correlated negatively and significantly with adiponectin, whereas VAT did not correlate significantly with adiponectin. TFA correlated positively and significantly with serum adiponectin (Table 3Go).
HOMA-IR, adiponectin, and body composition (DXA and MRI scans)
DXA scans - In univariate analysis, HOMA-IR was significantly positively associated with total fat, CFM, and LEFM (Table 4Go).
Using multiple linear regression analysis (Table 5Go) with HOMA-IR as the dependent variable, CFM was significantly positively correlated with HOMA-IR, whereas the association with LEFM was not significant. After introducing adiponectin to the multivariate linear regression analysis as independent variable, CFM remained a positive correlate of HOMA-IR, and adiponectin was a negative independent predictor of HOMA-IR (Table 6Go).
MRI scans - Univariate analysis showed significant positive associations between HOMA-IR and VAT, SAT, and TFA (Table 4Go). In multiple linear regression analysis (Table 5Go) incorporating VAT, SAT, and TFA as independents, SAT correlated significantly and positively with HOMA-IR, whereas VAT and TFA did not significantly predict HOMA-IR.
Introducing serum adiponectin into the multivariate linear regression analysis as an independent variable, SAT remained positively correlated with HOMA-IR, whereas VAT and TFA was not. Finally, serum adiponectin was an independent negative predictor of HOMA-IR (Table 6Go).
The analyses above have been repeated excluding the subjects on drugs known to influence adiponectin and glucose metabolism. The results were unchanged (data not shown).
Subjects and Methods

The Odense Androgen Study is a population-based, prospective, observational study on the interrelationship between endocrine status, body composition, muscle function, and bone metabolism in young men. A detailed description of the design and inclusion of participants has been published elsewhere (20).
In brief, at the time of this study, 28,310 men at the age of 20-29 yr were living in Fnen County, Denmark. Of 28,310 men, 3000 were randomly selected from the central personal registration database and invited by mail to answer a questionnaire. A total of 2042 men returned the questionnaire and were invited by letter; 859 men provided their telephone number, of which 827 were invited for an interview (19 declined, 10 could not be included, and 43 did not show up), leaving 783 men, who gave written informed consent to participate in the study.
The cohort reflects the background population of 20- to 29-yr-old men from Fnen, Denmark, as previously reported (20).
Information on chronic illness and medication was available from all subjects. A total of 690 participants received no drugs (90.9%), 45 participants (5.9%) used one drug, and 24 participants (3.2%) used two drugs or more (20). Ten subjects were on inhalation steroids, one subject was on a lipid-lowering drug, 11 subjects were on antiepileptics/antidepressants, seven subjects were on antacids, and five subjects were on antihypertensive drugs, two of them on angiotensin-converting enzyme inhibitors. Two subjects had type 1 diabetes and were treated with insulin. The subjects with type 1 diabetes were excluded before analysis of the associations with insulin sensitivity.
Three subjects were not analyzed due to abuse of anabolic steroids.
The clinical and biochemical characteristics of the study subjects are shown in Table 1Go. Fifty-five (7%) of the study subjects were obese [body mass index (BMI) >30 kg/m2].
The study was approved by the local ethics committee and is listed in (NCT00150163), and all participants provided written and verbal informed consent.
Anthropometric measurements
Body weight was measured using standardized equipment (Seca, Denmark) to the nearest 100 g, and height was measured by a stadiometer (Harpenden Holtain, Crymych, UK). BMI was calculated from the formula (body weight in kilograms)/(height in meters)2. Waist circumference was measured to the nearest centimeter between the iliac crest and the lower rib. Hip circumference was measured to the nearest centimeter, and the waist to hip ratio was calculated.
Measurement of adiponectin and insulin and calculation of HOMA-IR
Blood samples were collected after an overnight fast, and plasma glucose was determined with a routine clinical chemistry laboratory analyzer. Serum levels of insulin were analyzed by time-resolved immunofluorometric assay. The intraassay coefficient of variation (CV) was 2.1-3.7%, and the interassay CV was 3.4-4.0%. IR was estimated using HOMA-IR (21) and calculated from fasting plasma insulin and fasting plasma glucose concentrations and calculated as follows: HOMA-IR fasting insulin (mIU/liter) x fasting glucose (mmol/liter)/22.5. Adiponectin was determined by an in-house time-resolved immunofluorometric assay as described elsewhere (22). Intra- and interassay CV averaged less than 5 and 10%, respectively.
Total fat mass and lean body mass was measured by DXA using a Hologic 4500 device (Waltham, MA). CFM and LEFM was computed from the DXA images. The CV was 8% for CFM and 9% for LEFM, respectively.
MRI was performed with an open, low-field (0.2 Tesla) MR unit (Magnetom Open Viva; Siemens AG, Berlin, Germany). Three abdominal slices (10 mm thick, 20 mm apart, lower slice at the dorsal, intervertebral space of L4/L5) were recorded using an axial, T1-weighted gradient-echo sequence (repetition time 450 msec, echo time 15 msec, acquisition matrix 512 x 288, field of view 400 mm). Computer software was used to trace the different compartments of fat on the abdomen and for assessment of the areas of SAT and VAT. The TFA was determined in one femoral slice (equidistant from the major trochanter and patella) using a T1-weighted gradient-echo sequence (repetition time 370 msec, echo time 15 msec, acquisition matrix 512 x 512, field of view 230 mm).
Statistical analysis
The statistical analysis was performed by SPSS version 17. Statistics were performed using continuous natural logarithm transformed data to approximate normal distribution. P values <0.05 were considered statistically significant. Variables were expressed as median, minimum, and maximum.
In univariate analysis, we used the Pearson correlation, and results are presented as correlation coefficients (r). DXA data (CFM and LEFM) and MRI data (VAT, SAT, and TFA) were analyzed separately with regard to adiponectin and HOMA-IR, using multiple linear regression analysis to avoid the possible influence of covariance. Results are presented as partial correlation coefficients (r).
  iconpaperstack view older Articles   Back to Top