A total of 113 obese children from our unit were enrolled in this study from January 1 2005 to November 31 2005. The criteria for obese diagnosis was defined that the body mass index (BMI) was over 95% percentile of the same gender and age. They included 35 girls and 78 boys aged from 6.0 to 16.0 years (mean 11.1 ± 2.3 years). In obese group, 55 cases were prepuberal (Tanner I) and 58 were puberal (51 of Tanner II and 7 of Tanner III) and the mean BMI was 27.71 ± 3.88 kg/m². Hyperlipemia in 57 cases (50.44%), nonalcoholic fatty liver disease (NAFLD) in 63 cases (55.75%), impaired glucose tolerance (IGT) in 11 cases (9.73%), hypertension in 31 cases (27.43%) were noted in obese group. The control group consisted of 37 nonobese children without other endocrine, metabolic or kidney disease, including 17 boys and 20 girls aged from 7.3 to 14.9 years (mean 10.3 ± 1.8 years). In controls, 25 cases were prepuberal (Tanner I) and 12 were puberal (11 of Tanner II and 1 of Tanner III) with a mean BMI of 15.52 ± 1.53 kg/m².

Blood samples were obtained for adiponectin, resistin, and other biochemical measurements in the morning after an overnight fasting. All sera were stored frozen at –80ºC until measurement. The sera were diluted by 100 or 5-fold for adiponectin or resistin measurement before assay. Serum levels of adiponectin and resistin were assayed by an enzyme linked immunosorbent assay (ELISA). Quantikine kits purchased from R&D Systems Inc (USA) were used to detect serum adiponectin and resistin according to the manufacturer’s protocol. Negative and positive controls were used. The intra-and interassay CVs of adiponectin were ranged from 2.5 to 4.7% and 5.8 to 6.9% respectively. The intra- and interassay CVs of resistin were ranged from 3.8 to 5.3% and 7.8 to 9.2% respectively. The assay sensitivities were ranged from 0.079-0.891µg/L (0.246 µg/L) for adiponectin and 0.010-0.055 µg/L (0.026 µg/L) for resistin respectively.

Samples for fasting glucose and insulin (FG, FI) after an overnight fasting were obtained. Plasma glucose levels were detected by glucose oxidase method (Beijing North Biotechnology Invest, China) with intra-assay and interassay CVs of 2.1% and 4.4%. Insulin serum levels were determined by radioimmunoassay (Beijing North Biotechnology Invest, China). The intra-assay and interassay CVs were 6.4% and 9.7% respectively. Insulin resistance and ß cell function index were calculated as follow: insulin resistance by homeostasis model (HOMA-IR) = FI × FG/22.5(10), whole body insulin resistance index (WBISI) = 10000/[FI (mLU/L) × FG (mg/dL) × mean insulin (mIU/L) × mean glucose (mg/dL)]1/2 [11]. Serum triglyceride (TG), alanine aminotransferase (ALT), and uric acid were measured in the clinical laboratories of our unit. Plasma apolipoprotein (Apo) A1, ApoB, high-density lipoproteins-cholesterol (HDL-C) and low-density lipoproteins-cholesterol (LDL-C) levels were detected as well.

Statistical analyses were conducted by using SPSS software (10.0). Pearson Chi-square was used to measure the enumeration data between subgroups. Quantitative data were presented as mean ± SD. Because the data for adiponectin, resistin, TG, ALT and insulin, HOMA-IR and WBISI were skewed, they were presented as median (mix-max). The statistical significance was estimated by unpaired Student’s t test or two independent nonparametric tests (Mann-Whitney U method). Bivariate correlation analysis for insulin resistance parameter (WBISI, HOMA-IR) and other factors (including adiponectin and resistin) were done using Spearman correlation. Stepwise multiple linear regression models were used to examine the determinant of WBISI after these skewed parameters were transformed logarith-mically. A 2-tailed P <0.05 was considered significant.

The characteristics of controls and obese children were shown in Table I. Compared with controls, higher HOMA-IR and lower WBISI were found in obese children (all P <0.001). Moreover, the adiponectin level in obese children was 3.63 µg/L, which was significantly lower than that of controls (5.79 µg/L, P <0.001). However, the serum resistin levels were not significantly different between obese children and controls (P = 0.948).

When analyzing the correlation between the insulin resistance parameter (WBISI, HOMA-IR) and other factors, significant correlations between insulin resistance parameter and age, stage of sexual development, BMI, TG, ApoB, LDL-C, HDL-C, ALT, uric acid, or adiponectin levels (all P <0.05) were found. However, no significant correlation was found between insulin resistance parameter and gender (treated female as 0, male as 1), ApoA1 or resistin (all P >0.05), as shown in Table II.

In this study, we noted lower adiponectin level in obese children, which is consistent with previous studies(7,8). Animal model and some human studies showed resistin may be associated with diet-induced obese(4,5). However, the resistin in obese children in this study was not different from that of controls. This is similar with previous reports and suggests adiponectin, not resistin plays role in obesity(6-7). It is still unclear that why adiponectin levels decreased with increased adipose tissue mass. However, it is clear that low levels of adiponectin decrease fatty acid oxidation and higher fatty acid is associated with insulin resistance(12-13). Recent data have demonstrated that adiponectin effects are mediated by the interaction with muscle and hepatic receptors through activation of AMP kinase, the cellular "fuel gauge", which in turn inhibits acetyl CoA carboxylase and increases fatty acid beta-oxidation(13-14).

Although many studies have been done about the insulin resistance in obese, the mechanism is not clear. As like some other studies, insulin resistance correlated with BMI, TG, ApoB, ALT and uric acid (9). We here seek to describe the relation between insulin resistance and the two serum adipocytokines (adiponectin and resistin) in obese children. We found that serum adiponectin levels were positively and significantly correlated with WBISI. This suggested that adiponectin, as an adipocytokine, might play a role in the pathophysiology of insulin resistance and obesity.

Recent studies found that circulating adiponectin is predominantly present as several characteristic multimers, including low molecular weight adiponectin (LMW form), middle molecular weight adiponectin (MMW form) and high molecular weight adiponectin (HMW form)(15-16). More-over, different forms of adiponectin might activate different signal transduction pathways and exert distinct functions on its target tissue. HMW complex is the most active form of adiponectin in depressing blood glucose levels(17-20). In the present study, although low correlation between WBISI and serum adiponectin was found, whether different ratio of the different forms which might exist have an influence on insulin resistance is still unknown, hence further study is required.

In summary, our data that adiponectin correlated with insulin resistance support that adiponectin may play a protective role in obese children. Since our samples are small and association studies are susceptible to various biases, confirmatory studies using larger population are needed to prove these results.

We thank all children and their parents for participating in this research project. We thank Drs. Dong Guanping, Wang Chunlin, and Zhu Jianfang for their exceptional patient care and organization. We also thank Chen Liqin for her assay expertise.

Contributors: All authors contributed to study design, conduct, data collection, analysis and drafting of the manuscript.

Funding: Zhejiang Science and Technology grants (2005C24001).

Competing Interests: None stated.