Serum resistin correlates with central obesity but weakly with ... - Nature

International Journal of Obesity (2009) 33, 424–439 & 2009 Macmillan Publishers Limited All rights reserved 0307-0565/09 $32.00 www.nature.com/ijo

ORIGINAL ARTICLE Serum resistin correlates with central obesity but weakly with insulin resistance in Chinese children and adolescents M Li1,2,5, A Fisette3,5, X-Y Zhao4, J-Y Deng1,2, J Mi4 and K Cianflone3 1

Department of Endocrinology, Peking Union Medical College Hospital, Peking Union Medical College, Beijing, P.R. China; Department of Endocrinology, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences, Beijing, P.R. China; 3Centre de Recherche de l’Hoˆpital Laval, Universite´ Laval, Que´bec, Canada and 4Department of Epidemiology, Capital Institute of Pediatrics, Beijing, P.R. China 2

Objective: Resistin has been linked with obesity and hypothesized as a potential marker of insulin resistance in addition to being linked with acute inflammation. However, these links are still highly controversial in humans. Our goal was to examine resistin levels in relation to obesity, insulin resistance and inflammation markers in a large population of Asian children and adolescents. Methods: Children and adolescents (n ¼ 3472) aged 6–18 years, boys (n ¼ 1765) and girls (n ¼ 1707), were assessed for body size parameters, pubertal development, blood lipids, glucose, insulin, resistin, C-reactive protein (CRP), adiponectin and complement C3 (C3) levels. Results: Resistin increased with central obesity in both genders but not with simple adiposity in boys. Several markers associated with central obesity correlated in a gender-specific fashion with plasma resistin. Waist circumference, fat-mass percentage, waistto-height ratio and body mass index (BMI) positively correlated with resistin in both genders. Blood lipids such as triglycerides, nonesterified fatty acids (NEFA) and low-density lipoprotein cholesterol, diastolic and systolic blood pressure correlated positively with resistin in boys. NEFA, high-density lipoprotein cholesterol (negatively) and inflammation markers, such as CRP and C3, positively correlated with resistin in girls. There was no correlation between resistin and adiponectin, and no association of adiponectin with resistin quintiles in either boys or girls. In both boys and girls, resistin tended to decrease with age, with girls having higher levels than boys. Few indices of insulin resistance were linked with plasma resistin in either gender. Conclusion: In this population, plasma resistin levels are a weak biochemical marker of metabolic dysfunction defined by central obesity, adiposity and inflammation and does not predict insulin resistance. Only a small proportion of resistin variation can be explained by factors related to metabolic syndrome, suggesting that resistin is not strongly implicated in a concentrationdependent fashion in any of the examined pathologies. International Journal of Obesity (2009) 33, 424–439; doi:10.1038/ijo.2009.44; published online 17 March 2009 Keywords: resistin; childhood obesity; adolescent obesity; insulin resistance; metabolic syndrome

Introduction Resistin is a circulating 12.5 kDa protein of 114 amino acids that belongs to the resistin-like family.1,2 Considered an adipokine, resistin mRNA has been found in human adipose ˆ pital Laval, Correspondence: Dr K Cianflone, Centre de Recherche Ho Universite´ Laval, Y2186, 2725 Chemin Ste-Foy, Que´bec G1V 4G5, Canada. E-mail: katherine.cianflone@crhl.ulaval.ca or J Mi, Department of Epidemiology, Capital Institute of Pediatrics, Beijing, P.R. China. 5 These authors contributed equally to this work. Received 26 June 2008; revised 4 February 2009; accepted 8 February 2009; published online 17 March 2009

tissue and also in bone marrow, lung, nonfat cells of adipose tissue, placental tissue and pancreatic islets cells.3–7 Resistin is regulated by insulin, glucose, growth hormone and thiazolidinediones.8–11 A resistin receptor, tissue distribution of a receptor or the specific signaling pathway is unknown. The main function of resistin remains unclear. Resistin was initially hypothesized as a link between obesity and insulin resistance.1 In support of this theory, resistin is increased in mice with diet-induced obesity and in ob/ob mice.1 Studies also showed that mice injected with recombinant resistin or overexpressing resistin had impaired glucose tolerance and insulin action.1,12–14 In addition, resistin was found to be an in vitro antagonist of insulin on human preadipocytes.4

Resistin and central obesity in children M Li et al

425 Human hepatic cells overexpressing resistin had impaired glucose uptake and glycogen synthesis.15 By contrast, other groups have found the opposite or no correlation at all between resistin and insulin signaling.16–18 Resistin, also identified as ‘found in inflammatory zone 3’, has been linked with inflammation.19 It was positively correlated with pro-inflammatory factors in adults with pathophysiological conditions such as atherosclerosis, renal disease and inflammation of respiratory tracts.20–27 Proinflammatory molecules such as tumor necrosis factor-a, interleukin-6 and lipopolysaccharide regulate resistin gene expression in various cell models28 and reciprocal modulation has been hypothesized.29 Resistin is clearly involved in inflammation, but its specific function in that situation remains to be clarified. Many population studies suggest that resistin levels in adults are dependent on gender as women tend to have higher plasma resistin concentrations.30–32 This finding was not supported by another study.33 Resistin levels were also reported as not affected by age in adults.33 Most studies also found an increase in resistin in obese adults when compared to lean groups with a positive correlation between resistin and obesity indices.4,29,31,33–35 Because of its link with obesity and inflammation and its potential link with insulin resistance, resistin has been tagged as a potential metabolic syndrome (MetS) marker. Supporting this theory, adults with MetS tend to have higher resistin levels than their healthy counterparts.29 However, the correlation between insulin resistance and plasma resistin in adult humans remains controversial, and not in supported by some studies25,35–37 4,21,34,38,39 and this therefore weakens the relation others, between resistin and MetS. Numerous other adipokines have been shown to be associated with obesity and insulin resistance. In particular, adiponectin decreases with obesity and insulin resistance, and its function is associated with maintenance of insulin sensitivity.40 Thus, both resistin and adiponectin have been independently shown to be associated with insulin resistance, albeit in opposite directions. However, a clear functional link between these two hormones has yet to be extensively described, although suggested. Although some studies have found an inverse correlation between plasma levels of resistin and adiponectin,36,41–43 just as many others have found no statistical association24,44–47 or a positive correlation.48 All of these studies were conducted in adults. The relationships found were still significant after correction for insulin resistance factors. To our knowledge, no mechanism explaining a possible negative reciprocal cooperation or regulation between these adipokines has yet been proposed. A positive correlation mechanism, mediated through tumor necrosis factor-a secretion, has been suggested as plasma concentrations of both adipokines are regulated by this inflammatory cytokine.28,49 Most of the studies mentioned previously evaluated plasma resistin levels in a small numbers of subjects

(no350) and gender or obesity subgroups were relatively small for comprehensive statistical analysis. There is therefore little information about the relationship between resistin, inflammation, obesity and insulin resistance. The prevalence of obesity in children and adolescents and the associated risks of developing type 2 diabetes mellitus have been increasing dramatically for several decades.50 This relationship can be difficult to evaluate in children as acute insulin resistance consequent to obesity is usually only seen later in development. Resistin could potentially be a marker in obese children predicting the risks of transition to type 2 diabetes. As in adults, resistin levels in children tend to be higher in girls.32,51,52 A negative correlation of plasma resistin and age has also been found, suggesting a link with development.52 However, these relations with gender and age in children have not always been found.31,36,42 In contrast of their adult counterparts, obese children were reported to have similar resistin levels when compared to lean children.51,52 These findings come from few studies and, to our knowledge, all of them also involved limited subject numbers (fewer than 350 subjects), limiting the potential for subgroup analysis. We evaluated resistin levels and parameters linked to obesity, insulin resistance, inflammation and MetS in 3472 Chinese children and adolescents (1765 boys, 1707 girls), aged from 6 to 18 years. We also evaluated the possible correlation between resistin and adiponectin plasma concentrations in children. We hypothesized that resistin levels could be related to obesity levels, but also linked with early insulin resistance symptoms, inflammation and adiponectin plasma levels in Asian children and adolescents. In particular, the associations in very young children were compared to those in adolescents.

Methods Subjects Subjects were recruited from a cross-sectional populationbased survey: the Beijing Child and Adolescent Metabolic Syndrome (BCAMS) study.53 This study evaluated the presence of obesity and related metabolic abnormalities (hypertension, central obesity, type 2 diabetes, dyslipidemia) among a representative sample (n ¼ 19 593) of Beijing school children evaluated between April and October 2004. The cohort included boys and girls aged 6–18 years. Evaluation included physical examination, blood pressure, fasting finger capillary blood test for glucose, total cholesterol (TC) and triglycerides (TG). Physical examination included body mass index (BMI), waist circumference, fat-mass percentage (FMP) by bioimpedance analysis, systolic and diastolic blood pressure. Pubertal development was assessed by Tanner stage of breast development (girls) and testicular volume (boys).54 Within this large group of children and adolescents, 4500 were identified with risk factors defined as the presence of any one of the following: overweight based on BMI cutoffs, International Journal of Obesity


426 increased cholesterol (X5.2 mM), TG (X1.7 mM) or glucose (X5.6 mM). A parallel reference population of 1045 schoolaged children was also identified. Within these two groups, 2544 (BCAMS) and 981 (Reference) children/adolescents were recruited for blood samples. Within this total group, the cohort analyzed included 3472 children, boys (50.8%, n ¼ 1765) and girls (49.2%, n ¼ 1707), of normal weight, overweight and obese as defined by age- and gender-specific BMI cutoffs (recommended by the Working Group on Obesity in China55). Signed informed consent was obtained from all participants and/or their parents or guardians through all the study processes. The BCAMS study was approved by the ethics committee at Capital Institute of Pediatrics. The presence of pediatric MetS was defined by the presence of three or more of the following five components:56 (1) central obesity defined as X90th percentile for age and gender (established based on the BCAMS study); (2) elevated systolic and/or diastolic blood pressure X90th percentile for age, sex and height (according to the BCAMS study); (3) hypertriglyceridemia defined as TG X1.24 mM, equal to the 90th percentile of the reference population; (4) low serum HDL cholesterol (LHDL) defined as p1.03 mM, equal to 5th percentile of the reference population and (5) impaired fasting glucose (IFG) defined as X5.6 mM.

Fasting blood samples Venous blood samples were collected by direct venipuncture after an overnight (minimum 12 h) fast. The samples were centrifuged, aliquoted and immediately frozen for future analysis of lipids and hormones.

Analytical procedures Samples were analyzed for concentrations of plasma glucose, insulin, nonesterified fatty acids (NEFA), complement C3 (C3), resistin, C-reactive protein (CRP), and serum TG, TC, high-density lipoprotein cholesterol (HDL-Chol) and lowdensity lipoprotein cholesterol (LDL-Chol). Plasma glucose was determined by the glucose oxidase method. Serum TC and TG concentrations were determined using standard enzymatic methods. HDL-Chol and LDL-Chol were measured directly. The serum lipid levels and plasma glucose were assayed using the Hitachi 7060C Automatic Biochemistry Analysis System. Plasma NEFA was determined by colorimetric enzymatic assay (Wako Chemicals, Tokyo, Japan). Plasma C3 concentration was determined by turbidimetric assay using a polyclonal anti-human antibody specific against C3 (Linfei Co., P.R. China). For this last assay (C3), intra-assay coefficient of variation (CV) was o4% and interassay CV was o8%. Insulin was measured by monoclonal antibody-based sandwich enzyme-linked immunosorbent assays (ELISA), 57 which was developed in the Key Laboratory of Endocrinology, Peking Union Medical College Hospital. The assay had an inter-assay CV of o9.0% and had no crossInternational Journal of Obesity

reactivity to proinsulin (o0.05%). Adiponectin was measured by ELISA. All antibodies, reagents and adiponectin standard were purchased from Phoenix Pharmaceuticals Inc. (Belmont, CA, USA). Briefly, plates were coated with 10 mg ml1 monoclonal anti-human adiponectin antibody in phosphate-buffered saline (PBS), incubated overnight at 4 1C and then blocked for 1 h with 1% bovine serum albumin (BSA) in PBS. Standards (0, 1.56, 3.1, 6.25, 12.5, 25, 50 and 100 ng ml1) were prepared in 1% BSA–PBS. Plasma samples were diluted 1:500 in PBS. Twenty microliters of standards, diluted blood samples and quality control reagents were loaded. Biotin-labeled anti-adiponectin monoclonal antibodies (100 ml; diluted 1:1000) were added and incubated 1.5 h at 37 1C. The following steps were performed sequentially, with the plates washed between each step: (1) 100 ml of avidin-horseradish peroxidase solution (diluted 1:1000) was then added and incubated 30 min at 37 1C; (2) 100 ml of freshly mixed chromogenic substrate and development solution was added to each well, and plates were incubated 10 min at 37 1C and (3) 100 ml of 1 M H2SO4 was added to stop the reaction and OD was measured at 450 nm. The sensitivity of the assay was 0.1 ng ml1, the reaction was 100% specific to human adiponectin and intra-assay and inter-assay CVs were o5.4 and o8.5%, respectively. Serum resistin was measured using the ELISA kit by Phoenix Pharmaceuticals Inc. The intra-assay and inter-assay CVs of this assay were o10 and o5%, respectively.

Calculations and statistical analysis BMI was calculated as weight (kg) divided by height squared (m2). Insulin resistance index was calculated by homeostasis model assessment of insulin resistance (HOMA-IR) as (fasting insulin UI l1) (fasting glucose mmol l1)/22.5.58 Unless otherwise stated, all results are displayed as mean±standard error of the mean (s.e.m.). All data analyses were performed using SigmaStat version 3.5 and GraphPad Prism version 5.0. Statistical normality of the data distribution was verified using the Kolmogorov–Smirnov test. Nonparametric data not normally distributed were analyzed after being logarithmically transformed. Resistin values outside of mean±3s interval were excluded. Fifteen values were excluded from the male group out of 1780 (0.8%) whereas ten values were excluded from the female group out of 1715 (0.6%). Differences between two groups were evaluated with Mann–Whitney U-test for nonparametric data. Differences between three or more groups were evaluated using one-way analysis of variance (ANOVA) with Bonferroni post hoc comparison test or linear trend post hoc test for parametric data and using Kruskal–Wallis test with Dunn’s post hoc comparison test for nonparametric data. Differences between grouped data were evaluated with two-way ANOVA with Bonferroni post hoc comparison test. Correlations between nonparametric data were examined using Pearson’s correlation analysis whereas correlations involving one or two parametric data were examined using Spearman’s correlation


427 analysis. For contribution of multiple independent variables on one dependent variable, multiple regressions using forward stepwise regression analysis were used. A P-value of o0.05 was considered statistically significant for all analysis.

Results Baseline characteristics of the population A representative sample of 3472 children, boys and girls, ranging from 6 to 18 years was selected from a 2004 crosssectional population-based study of school children (n ¼ 19 593) that evaluated obesity and related metabolic abnormalities (hypertension, central obesity, IFG and dyslipidemia) in Beijing, P.R. China. Obesity was defined based on Centers for Disease Control (CDC) BMI standards for 6-yearold children,55 and Chinese BMI standards established for 7- to 18-year-old children and adolescents.59

Gender-related differences in resistin levels The clinical and laboratory data of all the 3472 subjects distributed according to their gender are shown in Table 1. Age, degree of pubertal development, height, weight, waist circumference, FMP, waist-to-height ratio, BMI, systolic and diastolic blood pressure, TG, HDL-Chol, fasting glucose and adiponectin significantly differed between boys and girls (plasma TG, Po0.05, other parameters, Po0.001). TC, LDLChol, fasting insulin, HOMA-IR and plasma NEFA were constant between genders. As seen in Figure 1, resistin distribution is skewed among boys and girls. As shown in Table 1, boys had slightly lower average resistin levels than girls (17.0±0.2 vs 17.8±0.2 ng ml1, P ¼ 0.004). Boys and

Table 1

girls were then separately divided based on Tanner stage for prepubertal (Tanner stage 1) and pubertal (Tanner stages 2–5) stages (Figures 1c and d). As seen in Figure 1, the genderrelated difference in plasma resistin levels was not quite significant when comparing prepubertal groups (average plasma resistin for boys and girls, 16.5±1.2 vs 19.0±0.6 ng ml1, P ¼ 0.055), but was significant within pubertal group (average plasma resistin for boys and girls, 17.1±0.3 vs 17.8±0.3 ng ml1, P ¼ 0.005).

Age-related differences in resistin levels In prepubertal children, overall, average resistin levels were higher (18.6±0.6 ng ml1) than in pubertal children (17.4±0.2 ng ml1). Resistin levels were negatively correlated with age (P ¼ 0.0003). This was also true when examining each gender separately (boys, P ¼ 0.003 and girls, P ¼ 0.021). Pubertal markers in boys and girls were also negatively correlated with plasma resistin levels, although to a lesser extent (testicular volume in boys, P ¼ 0.042, Tanner stage development in boys, P ¼ 0.096 and Tanner stage development in girls, P ¼ 0.046). Resistin linearly decreased with each Tanner stage when taking into account the whole cohort (P ¼ 0.0075). As showed in Figure 1, resistin also decreased linearly with each Tanner stage for girls (Figure 1f, P ¼ 0.015). This tendency was also observed in boys, but only after the first Tanner stage, at the onset of puberty (Figure 1e, P ¼ 0.023).

Relations between resistin and obesity The population was then divided into quintiles based on resistin levels. Values for several biochemical and clinical

Clinical data in boys and girls

Variables Age (years) Puberty (Tanner stage) BMI (kg/m2) Waist circumference (cm) Waist-to-height ratio Fat-mass percentage (%) Systolic blood pressure (mm Hg) Diastolic blood pressure (mm Hg) Triglycerides (mmol l1) Nonesterified fatty acids (mmol l1) Total cholesterol (mmol l1) HDL-cholesterol (mmol l1) LDL-cholesterol (mmol l1) Glucose (mmol l1) Insulin (mU l1) HOMA-IR Adiponectin (mg ml1) Resistin (ng ml1)

Boys (n ¼ 1765)

Girls (n ¼ 1707)

12.2±0.1 3.34±0.03 22.8±0.1 75.7±0.4 0.492±0.002 23.0±0.2 110.4±0.4 68.8±0.3 1.02±0.01 0.672±0.012 4.06±0.02 1.38±0.01 2.51±0.02 5.14±0.02 10.7±0.3 2.52±0.07 12.3±0.2 17.0±0.2

12.6±0.1 3.21±0.03 21.0±0.1 68.3±0.3 0.456±0.001 25.2±0.2 104.6±0.3 66.7±0.2 1.03±0.01 0.683±0.011 4.11±0.02 1.42±0.01 2.57±0.02 5.02±0.02 10.2±0.2 2.33±0.06 13.2±0.2 17.8±0.2

P-value

All (n ¼ 3472)

*** *** *** *** *** *** *** *** * NS NS *** NS *** NS NS *** **

12.4±0.1 3.28±0.02 21.9±0.1 72.1±0.2 0.475±0.001 24.1±0.2 107.7±0.2 67.8±0.2 1.03±0.01 0.677±0.008 4.09±0.01 1.40±0.01 2.54±0.01 5.09±0.01 10.5±0.2 2.44±0.04 12.8±0.1 17.4±0.2

Abbreviations: BMI, body mass index; HDL, high-density lipoprotein; HOMA-IR, homeostatic model assessment of insulin resistance; LDL, low-density lipoprotein; NS, not significant. All values are reported as mean±s.e.m. Significance was calculated by Mann–Whitney or Student’s t-test where differences between girls and boys groups are indicated as *Po0.05, **Po0.01 and ***Po0.001.

International Journal of Obesity


428

Figure 1 Distribution, gender- and age-related differences in plasma resistin. Plasma resistin was measured in children and the distribution in boys (a) and in girls (b) is shown. Results are expressed as number of observations relative to resistin levels. Average resistin levels divided by gender in prepubertal children (c) and in pubertal children (d) are shown. Average resistin levels in each Tanner stage are shown for boys (e) and for girls (f). Results are expressed as mean±s.e.m. Significance was calculated by Kruskal–Wallis or Mann–Whitney tests where differences vs boys are expressed as **Po0.01 (c and d).

factors for each quintile can be found in Table 2 for the whole cohort, in Table 3 for boys and Table 4 for girls. A positive linear trend is observed between quintiles when comparing their average degree of obesity or BMI (for both, boys, Po0.0001 and girls, P ¼ 0.003). The highest resistin quintile in boys also showed a significantly increased degree of obesity (Po0.001), BMI (Po0.001), waist circumference (Po0.01), waist-to-height ratio (Po0.0001) and FMP (Po0.0001) compared to the first quintile. In girls, the International Journal of Obesity

fourth quintile showed, but to a lesser extent, an association with increased degree of obesity (Po0.05), BMI (Po0.05), waist circumference (Po0.05), waist-to-height ratio (Po0.01) and FMP (Po0.01) compared to the first quintile. In addition, subgroups based on the degree of obesity showed a significant increase in average plasma resistin levels for obese vs normal weight children (18.2±0.3 vs 16,9±0.3 ng ml1, Po0.0001). This increase was also true in boys, as seen in Figure 2 (18.0±0.4 vs 16.2±0.4 ng ml1,

Table 2

Average values for quintiles of resistin in the whole cohort

Continuous variables

Pearson vs resistin R (P)

1

Nonparametric variables Puberty (Tanner stage) Degree of obesity (% Normal/Owt/Ob) MetS incidence (%) Central obesity incidence (%) High blood pressure incidence (%) Impaired fasting glucose incidence (%) High triglycerides incidence (%) Low HDL-cholesterol incidence (%) No. of MetS components

0.061 0.068 0.053 0.091 0.091 0.028 0.050 0.063 0.102 0.047 0.039 0.061 0.0211 0.0128 0.005 0.0193

(0.0003) (o0.0001) (0.00171) (o0.0001) (o0.0001) (NS) (0.003) (0.0002) (o0.0001) (0.0054) (0.0212) (0.0003) (NS) (NS) (NS) (NS)

7.80±0.06 12.6±0.1 21.3±0.2 70.5±0.5 0.465±0.003 22.5±0.3 107±1 66.7±0.4 0.98±0.02 0.614±0.015 1.43±0.01 2.52±0.03 5.12±0.03 10.1±0.5 2.37±0.12 12.8±0.3 1.54±0.05

11.42±0.03 12.5±0.1 21.7±0.2 71.7±0.5 0.470±0.003* 23.6±0.3 107±1 67.5±0.4 1.00±0.02 0.679±0.023 1.41±0.01 2.50±0.03 5.10±0.02 10.6±0.4 2.46±0.11 12.9±0.3 1.57±0.04

14.84±0.04 19.80±0.07 33.25±0.38 12.4±0.1 12.4±0.1 12.1±0.1**,# 0.054 (0.001) 22.1±0.2* 22.3±0.2*** 22.3±0.2*** 0.077 (0.0001) 72.5±0.5 73.3±0.5** 72.7±0.5* 0.062 (0.003) ,## 0.476±0.003*** 0.480±0.003*** 0.484±0.003*** 0.097 (0.0001) 0.098 (0.0001) 24.4±0.3*** 24.9±0.3*** 25.0±0.3***,## 108±1 109±1 108±1 0.034 (0.0445) 68.4±0.4* 68.5±0.4* 68.2±0.4* 0.056 (0.001) ,# 1.04±0.02 1.06±0.02* 1.07±0.02*** 0.059 (0.0005) 0.703±0.019** 0.688±0.016* 0.702±0.016*** 0.082 (0.0011) 1.39±0.01 1.38±0.01 1.39±0.01 0.052 (0.002) 2.53±0.03 2.56±0.03 2.60±0.03# 0.04 (0.0184) 5.08±0.02 5.11±0.02 5.04±0.02#,& 0.03 (NS) 10.3±0.3 10.8±0.3 10.7±0.4 0.02 (NS) 2.41±0.10 2.49±0.08 2.46±0.09 0.01 (NS) 12.5±0.3 12.9±0.3 12.9±0.3 0.00003 (NS) 1.61±0.06 1.60±0.04 1.54±0.04 0.004 (NS)

Spearman vs resistin R (P) Resistin quintile 1 Resistin quintile 2 Resistin quintile 3 Resistin quintile 4 Resistin quintile 5 0.049 0.082 0.049 0.102 0.054 0.039 0.051 0.019 0.085

(0.0048) (o0.0001) (0.0041) (o0.0001) (0.0016) (0.0205) (0.0027) (NS) (o0.0001)

3.33±0.05 53/18/29 11 27 24 14 21 11 0.97±0.04

3.37±0.05 47/20/33 10 31 27 13 23 8 1.03±0.04

Linear trend R (P)

3.32±0.05 45/19/36 13 34 32* 11 24 9 1.11±0.04

3.25±0.05 44/18/38 13 39***,# 29 12 26 13 1.18±0.04**

#

3.15±0.05 42/18/40 15# 40***,## 30 10 28* 10 1.18±0.04**

Linear trend R (P) 0.050 0.083 0.046 0.099 0.045 0.038 0.052 0.015 0.073

(0.004) (0.0001) (0.0073) (0.0001) (0.0086) (0.026) (0.022) (NS) (0.0001)

ANOVA P-value

0.0104 0.0001 0.0015 o0.0001 o0.0001 NS 0.0034 0.0002 0.0007 NS 0.0474 0.0079 NS NS NS NS


Resistin (ng ml ) Age (years) BMI (kg/m2) Waist circumference (cm) Waist-to-height ratio Fat-mass percentage (%) Systolic blood pressure (mm Hg) Diastolic blood pressure (mm Hg) Triglycerides (mmol l1) Nonesterified fatty acids (mmol l1) HDL-cholesterol (mmol l1) LDL-cholesterol (mmol l1) Glucose (mmol l1) Insulin (mU l1) HOMA-IR Adiponectin (mg ml1) C3 (g l1)

Resistin quintile 1 Resistin quintile 2 Resistin quintile 3 Resistin quintile 4 Resistin quintile 5

ANOVA P-value 0.0263 NS NS o0.0001 0.0232 NS 0.0499 NS 0.0007

Abbreviations: ANOVA, analysis of variance; BMI, body mass index; HDL, high-density lipoprotein; HOMA-IR, homeostatic model assessment of insulin resistance; LDL, low-density lipoprotein; MetS, metabolic syndrome; NS, not significant; Owt, overweight; Ob, obese. All values are reported as mean±s.e.m. The cohort were separated based on resistin quintiles, where Q1 ¼ 3.20–9.92, Q2 ¼ 9.93– 12.93, Q3 ¼ 12.94–16.79, Q4 ¼ 16.80–23.53 and Q5 ¼ 23.54–77.46 ng ml1. Significance was calculated by Pearson’s regression, Kruskal–Wallis test with Dunn’s post hoc comparison or one-way ANOVA with linear trend post hoc test where differences vs quintile 1 are indicated as *Po0.05, **Po0.01 and ***Po0.001, differences vs quintile 2 are indicated as #Po0.05, ##Po0.01 and differences with quintile 4 are indicated as &Po0.05.

429


430

International Journal of Obesity Table 3 Average values for quintiles of resistin in boys Continuous variables


0.072 0.090 0.081 0.120 0.105 0.070 0.072 0.083 0.103 0.046 0.049 0.071 0.039 0.029 0.010 0.0002

(0.0027) (0.0002) (0.0007) (o0.0001) (o0.0001) (0.0034) (0.0024) (0.0005) (0.0026) (NS) (0.0408) (0.0028) (NS) (NS) (NS) (NS)

Resistin quintile 1 Resistin quintile 2 Resistin quintile 3 Resistin quintile 4 Resistin quintile 5 7.6±0.1 12.3±0.2 22.0±0.3 73.5±0.8 0.481±0.004 21.6±0.4 108±1 67.0±0.6 0.980±0.030 0.593±0.020 1.39±0.02 2.52±0.04 5.21±0.04 10.1±0.6 2.39±0.16 12.4±0.4 1.56±0.07

11.1±0.1 12.4±0.2 22.5±0.3 75.0±0.8 0.486±0.004 22.3±0.4 110±1 68.2±0.6 0.972±0.027 0.677±0.039 1.42±0.02 2.46±0.04 5.15±0.03 10.7±0.7 2.55±0.18 12.6±0.4 1.62±0.06

14.4±0.1 12.3±0.2 23.4±0.3 76.0±0.8 0.492±0.004 23.4±0.4 112±1** 69.8±0.6** 1.03±0.03 0.693±0.025* 1.40±0.02 2.54±0.04 5.10±0.03 10.7±0.6 2.55±0.17 12.0±0.4 1.52±0.06

19.4±0.1 12.3±0.2 23.5±0.3* 77.0±0.7* 0.495±0.004 23.5±0.4* 112±1** 69.9±0.5** 1.04±0.03 0.686±0.021* 1.35±0.02 2.50±0.04 5.19±0.02 11.1±0.5 2.60±0.11 12.2±0.4 1.58±0.05

32.7±0.6 11.7±0.2*,& 0.059 (0.0132) 24.2±0.3*** 0.102 (o0.0001) 77.3±0.8** 0.087 (0.0003) 0.509±0.004***,###,+ 0.127 (o0.0001) 0.117 (o0.0001) 24.2±0.5***,## 111±1 0.082 (0.0005) 69.5±0.5* 0.092 (0.0001) 1.10±0.03**,# 0.085 (0.0003) 0.690±0.021** 0.038 (NS) 1.37±0.02 0.041 (NS) # 2.60±0.04 0.045 (NS) 5.08±0.02&& 0.038 (NS) 11.2±0.5 0.037 (NS) 2.60±0.15 0.041 (NS) 12.4±0.4 0.000007 (NS) 1.50±0.05 0.010 (NS)

Spearman vs resistin R (P) Resistin quintile 1 Resistin quintile 2 Resistin quintile 3 Resistin quintile 4 Resistin quintile 5 0.041 0.109 0.077 0.130 0.080 0.036 0.071 0.009 0.107

(0.096) (o0.0001) (0.0012) (o0.0001) (o0.0001) (NS) (0.0030) (NS) (o0.0001)

3.35±0.07 45/18/37 14 29 25 17 22 14 1.08±0.06

3.46±0.07 38/21/31 11 33 27 15 21 10 1.07±0.06

3.35±0.07 38/18/44 14 36 36* 14 26 9 1.23±0.06

3.34±0.07 36/17/46 17 40* 34 18 26 14 1.32±0.06*,#

Linear trend R (P)

3.21±0.07 30/14/53***,# 20## 47***,##,+ 34 11 30 11 1.33±0.06*,#

Linear trend R (P) 0.032 0.214 0.106 0.141 0.087 0.033 0.077 0.011 0.143

(NS) (o0.0001) (o0.0001) (o0.0001) (0.0006) (NS) (0.0012) (NS) (o0.0001)

ANOVA P-value

0.024 0.001 0.008 o0.0001 o0.0001 0.002 0.0004 0.008 NS NS NS 0.0002 NS NS NS NS ANOVA P-value NS o0.0001 0.0005 o0.0001 0.0006 NS 0.021 NS o0.0001

Abbreviations: ANOVA, analysis of variance; BMI, body mass index; HDL, high-density lipoprotein; HOMA-IR, homeostatic model assessment of insulin resistance; LDL, low-density lipoprotein; MetS, metabolic syndrome; NS, not significant; Owt, overweight; Ob, obese. All values are reported as mean±s.e.m. Boys were separated based on resistin quintiles, where Q1 ¼ 3.39–9.61, Q2 ¼ 9.62–12.62, Q3 ¼ 12.63–16.41, Q4 ¼ 16.42–22.98 and Q5 ¼ 22.99–77.46 ng ml1. Significance was calculated by Pearson’s regression, Kruskal–Wallis test with Dunn’s post hoc comparison or one-way ANOVA with linear trend post hoc test, where differences vs quintile 1 are indicated as *Po0.05, **Po0.01, ***Po0.001; differences vs quintile 2 are indicated as #Po0.05, ##Po0.01; differences vs quintile 3 are indicated as +Po0.05 and differences with quintile 4 are indicated as &Po0.05, &&Po0.01.


Resistin (ng ml1) Age (years) BMI (kg/m2) Waist circumference (cm) Waist-to-height ratio Fat-mass percentage (%) Systolic blood pressure (mm Hg) Diastolic blood pressure (mm Hg) Triglycerides (mmol l1) Nonesterified fatty acids (mmol l1) HDL-cholesterol (mmol l1) LDL-cholesterol (mmol l1) Glucose (mmol l1) Insulin (mU l1) HOMA-IR Adiponectin (mg ml1) C3 (g l1)


Table 4 Average values for quintiles of resistin in girls Continuous variables


0.056 0.063 0.054 0.090 0.068 0.003 0.037 0.036 0.097 0.055 0.027 0.0412 0.002 0.003 0.013 0.042

(0.0206) (0.0095) (0.0257) (0.0002) (0.0055) (NS) (NS) (NS) (0.0091) (0.0237) (NS) (NS) (NS) (NS) (NS) (NS)

Resistin quintile 1 Resistin quintile 2 Resistin quintile 3 Resistin quintile 4 Resistin quintile 5 Linear trend R (P) 8.08±0.09 12.8±0.2 20.4±0.2 66.8±0.7 0.447±0.003 23.9±0.5 104±1 66.3±0.5 1.00±0.03 0.651±0.022 1.45±0.02 2.52±0.04 5.01±0.05 10.3±0.7 2.40±0.19 13.1±0.4 1.51±0.06

11.8±0.1 12.7±0.2 21.0±0.3 68.6±0.6 0.454±0.003 25.3±0.5 105±1 66.3±0.5 1.03±0.03 0.638±0.023 1.41±0.02 2.55±0.04 5.03±0.03 10.3±0.4 2.33±0.10 13.1±0.4 1.55±0.07

15.2±0.1 12.5±0.2 20.9±0.2 68.1±0.6 0.455±0.003 24.9±0.5 104±1 67.0±0.5 1.02±0.03 0.714±0.027 1.41±0.01 2.52±0.04 5.05±0.03 9.86±0.38 2.28±0.10 13.3±0.4 1.68±0.10

20.1±0.1 12.6±0.2 21.6±0.3* 69.7±0.7* 0.464±0.003** 26.4±0.5** 105±1 67.1±0.5 1.07±0.03 0.695±0.024 1.41±0.02 2.59±0.04 5.04±0.04 10.0±0.4 2.28±0.09 13.3±0.4 1.61±0.08

33.7±0.5 12.3±0.2 21.2±0.2 68.7±0.7 0.463±0.003** 25.6±0.5 105±1 67.2±0.5 1.05±0.03 0.714±0.024 1.40±0.02 2.60±0.04 4.99±0.03 10.5±0.5 2.39±0.12 13.3±0.4 1.58±0.07

0.050 0.073 0.062 0.095 0.082 0.015 0.041 0.055 0.093 0.054 0.043 0.030 0.016 0.012 0.00004 0.034

(0.038) (0.003) (0.011) (o0.0001) (0.001) (NS) (NS) (0.024) (0.013) (0.025) (NS) (NS) (NS) (NS) (NS) (NS)

Spearman vs resistin R (P) Resistin quintile 1 Resistin quintile 2 Resistin quintile 3 Resistin quintile 4 Resistin quintile 5 Linear trend R (P) 0.049 0.077 0.022 0.077 0.032 0.034 0.031 0.057 0.072

(0.0458) (0.0015) (NS) (0.0014) (NS) (NS) (NS) (0.0185) (0.0030)

3.28±0.08 62/17/21 9 26 24 11 20 7 0.88±0.06

3.30±0.08 52/19/29 9 30 26 8 26 8 0.98±0.06

3.25±0.08 55/20/25 9 28 26 9 20 7 0.90±0.06

320±0.08 51/20/29* 10 37* 26 7 25 12 1.08±0.06

3.08±0.08 52/19/29 11 35 28 8 26 10 1.07±0.06

0.048 0.072 0.018 0.074 0.029 0.032 0.034 0.054 0.062

(0.048) (0.003) (NS) (0.002) (NS) (NS) (NS) (0.025) (0.010)

ANOVA P-value

NS 0.0092 0.039 0.002 0.004 NS NS NS NS NS NS NS NS NS NS NS


Resistin (ng ml1) Age (years) BMI (kg/m2) Waist circumference (cm) Waist-to-height ratio Fat-mass percentage (%) Systolic blood pressure (mm Hg) Diastolic blood pressure (mm Hg) Triglycerides (mmol l1) Nonesterified fatty acids (mmol/L) HDL-cholesterol (mmol l1) LDL-cholesterol (mmol l1) Glucose (mmol l1) Insulin (mU l1) HOMA-IR Adiponectin (mg ml1) C3 (g l1)


ANOVA P-value NS 0.024 NS 0.010 NS NS NS NS 0.044

Abbreviations: ANOVA, analysis of variance; BMI, body mass index; HDL, high-density lipoprotein; HOMA-IR, homeostatic model assessment of insulin resistance; LDL, low-density lipoprotein; MetS, metabolic syndrome; NS, not significant; Owt, overweight; Ob, obese. All values are reported as mean±s.e.m. Girls were separated based on resistin quintiles, where Q1 ¼ 3.20–10.16, Q2 ¼ 10.17–13.30, Q3 ¼ 13.31–17.19, Q4 ¼ 17.20–24.07 and Q5 ¼ 24.08–77.07 ng ml1. Significance was calculated by Pearson’s regression, Kruskal–Wallis test with Dunn’s post hoc comparison or one-way ANOVA with linear trend post hoc test, where differences vs quintile 1 are indicated as *Po0.05, **Po0.01.

431



432

Figure 2 Relationship between obesity and resistin. Average obesity index for each quintile of resistin in boys (a) and in girls (b) are shown (where obesity index 0 ¼ normal, 1 ¼ overweight and 2 ¼ obese) with same bin limits as in 2 and 3. Average plasma resistin levels in normal, overweight (Owt) and obese (Ob) boys (c) and girls (d) are shown. Results are expressed as mean±s.e.m. Significance was calculated by Kruskal–Wallis test, where differences vs first quintile are expressed as *Po0.05 and ***Po0.001 (a and b), and where differences vs normal weight children are expressed as *Po0.05 and ***Po0.001 (c and d).

Po0.0001). The same increase was observed in obese vs normal weight girls, but to a lesser extent (18.5±0.5 vs 17.4±0.3 ng ml1, P ¼ 0.007). Consistent with obesity associations, Pearson’s analysis showed that several obesity markers were positively correlated with resistin levels. In boys, waist-to-height ratio (Po0.0001), FMP (Po0.0001), BMI (P ¼ 0.0002) and waist circumference (Po0.0007) all correlated with plasma resistin levels. None of these associations was found when exclusively analyzing the male prepubertal group. In girls, waistto-height ratio (P=0.0002), FMP (P ¼ 0.006), BMI (P ¼ 0.010) and waist circumference (P ¼ 0.026) were positively correlated with resistin plasma levels. None of these associations was found when exclusively analyzing the female prepubertal group. Spearman’s analysis indicated that the degree of obesity also correlated positively with plasma resistin in both genders (boys, Po0.0001 and girls, P ¼ 0.002).

Relations between resistin and blood lipids and pressure Blood lipids such as plasma TG (P=0.0005), plasma NEFA (P ¼ 0.003) and LDL-Chol (P ¼ 0.041) positively correlated with resistin in boys (Table 3). The fifth quintile of resistin in boys showed increased TG (Po0.01) and NEFA (Po0.01) when compared to the first quintile. No correlation relative International Journal of Obesity

to blood lipids was found when only analyzing the prepubertal boys group. In girls, NEFA (P ¼ 0.009) positively correlated with resistin and HDL-Chol (P ¼ 0.024) negatively correlated (Table 4). There was no significant difference in girls resistin quintiles in relation to blood lipids. Only HDLChol negatively correlated with resistin in the female prepubertal group (P ¼ 0.012). In boys, blood pressure components such as diastolic blood pressure (P ¼ 0.002) and systolic blood pressure (P ¼ 0.003) were also positively and significantly correlated with plasma resistin. In girls, no link between resistin and blood pressure was observed.

Relations between resistin and glucose metabolism For boys, linear regression analysis showed that resistin levels surprisingly correlated negatively with glucose (P ¼ 0.003), but that insulin levels (P ¼ 0.106), HOMA-IR (P ¼ 0.229) and presence of IFG (P ¼ 0.136) did not correlate with blood resistin (Table 3). The same tendencies were observed if only examining the prepubertal boys. For girls, none of the aforementioned factors correlated with resistin (glucose, P ¼ 0.089, insulin levels, P ¼ 0.945, HOMA-IR, P ¼ 0.904 and presence of IFG, P ¼ 0.159; Table 4). However, in the prepubertal girls subgroup, insulin levels strongly and


433

Figure 3 Metabolic syndrome (MetS) and resistin association. Average plasma resistin levels in children and adolescents based on number of components of the MetS are shown for boys (a) and girls (b). Average plasma resistin in children and adolescents with impaired fasting glucose (IFG), high triglyceride (HTG), low HDLcholesterol (LHDL), high blood pressure (HBP), central obesity (C. Ob) compared to control group with zero MetS component is shown in boys (c) and girls (d). Results are expressed as mean±s.e.m. Significance was calculated by Kruskal–Wallis tests, where differences vs children with no MetS component are expressed as *Po0.05, **Po0.01 and ***Po0.001.

positively correlated with plasma resistin (Po0.0001). There was no correlation between resistin and adiponectin in either the whole cohort (Table 2) or in boys and girls separately (Tables 3 and 4).

Relations between resistin and MetS As shown previously, several MetS markers were correlated with resistin levels in a gender-specific way, including blood pressure, plasma TG, glucose, HDL-Chol and degree of obesity. We evaluated the average plasma resistin in groups based on the number of positive MetS components. When pooling both sexes, resistin levels were significantly increased in children with two and with three and more MetS components (for both, Po0.01). More gender-specific analyses are shown in Figures 3a and b, where boys with two MetS components (17.3±0.5 ng ml1, Po0.01) and with three and more MetS components (18.5±0.6 ng ml1, Po0.01) showed increased resistin levels compared to those without any MetS component (16.2±0.4 ng ml1). Girls with two MetS components also had significantly higher resistin levels than girls without any MetS component (18.8±0.6 and 17.4±0.4 ng ml1, respectively; Po0.05). Despite

a moderate increase, girls with three or more MetS components did not show significantly elevated plasma resistin (18.4±0.8 ng ml1). A positive linear trend was observed between resistin and the number of MetS components in both genders (boys, P ¼ 0.0014 and girls, P ¼ 0.048). We also evaluated the relation between plasma resistin and presence of specific MetS components. In boys, presence of central obesity (Po0.0001), high blood pressure (HBP) (Po0.0001), high triglyceride (HTG) (P ¼ 0.003), number of MetS components (Po0.0001) and presence of MetS (P ¼ 0.001) was correlated with resistin plasma levels (Table 3). When exclusively analyzing the male prepubertal group, none of these associations was found. In girls, presence of central obesity (P ¼ 0.001), presence of LHDL (P ¼ 0.019) and number of MetS components (Po0.003) correlated with plasma resistin levels (Table 4). In the prepubertal girls, only LHDL correlated with resistin (P ¼ 0.034). We evaluated the average plasma resistin in presence of each specific (but not exclusive) type of MetS component and analyzed the differences between each MetS component group and control group (number of MetS component ¼ 0). International Journal of Obesity


434 When examining the whole cohort, children with central obesity (18.5±0.3 ng ml1, Po0.001), HTG (18.0± 0.3 ng ml1, Po0.001) or HBP (17.9±0.3 ng ml1, Po0.001) had significantly higher plasma resistin than children without any MetS component (16.9±0.3 ng ml1). As seen in Figure 3, gender-specific subgroups showed increased average resistin levels in boys with central obesity (18.5± 0.4 ng ml1, Po0.001), HTG (18.0±0.5 ng ml1, Po0.01) and HBP (17.7±0.4 ng ml1, Po0.01) compared to the group without any MetS component (16.2±0.4 ng ml1). Groups with IFG and LHDL showed levels comparable to the control group. In girls, only those positive for central obesity had significantly elevated average resistin levels compared to the healthy control group (18.6±0.4 and 17.4±0.4 ng ml1, respectively; Po0.05). We subsequently divided the above MetS-specific component subgroups based on pubertal development (Tanner stages). The increases in plasma resistin in boys with central obesity, HTG or HBP were consistently observed in each stage of puberty (Po0.01 for all three parameters). Average resistin levels in boys with MetS components were also consistently higher than those without MetS components, throughout puberty. However, the increase in plasma resistin levels of boys with central obesity tended to be located primarily in the central Tanner stages (2–4) and therefore might be related to interactions between Tanner stage and MetS component (P ¼ 0.050). Prepubertal boys (Tanner stage 1) or boys that completed their puberty (Tanner stage 5) tended to have similar or slightly elevated average resistin levels when positive to the aforementioned MetS components. No such relation could be observed in girls, where average resistin levels were consistently higher in girls with central obesity independently of their pubertal stage (Po0.01) and with no interaction between Tanner stage and MetS component. To further distinguish the effects of central obesity on plasma resistin levels, we divided the population into three groups based on obesity index as described earlier. We then further divided these groups into two subgroups, negative or positive for central obesity (Figure 4). Children with central obesity had consistently increased resistin levels compared to those without central obesity, regardless of obesity index. Overweight or obese children with central obesity had significantly increased resistin when compared to lean children (Po0.01 and Po0.001, respectively), but overweight or obese children without central obesity had similar resistin levels. We then further divided our population based on gender. As shown in Figure 4a, boys with central obesity had a consistently elevated resistin compared to their counterparts without central obesity (Po0.01). Girls with central obesity also tended to higher average plasma resistin levels than those without central obesity, but the difference was not significant and only observed in lean or overweight girls (Figure 4b). Obese girls had elevated resistin levels compared to lean girls regardless of central obesity. International Journal of Obesity

We also evaluated the combined effects of puberty and central obesity by dividing the population according to Tanner stage, and then further dividing the group in two subgroups, negative or positive for central obesity. Resistin levels were significantly increased in the presence of central obesity (Po0.001), with significant increases at Tanner stages 2–4 (Po0.01 for all). When divided based on gender, as shown in Figures 4c and d, plasma resistin in boys with central obesity was also globally higher than their counterparts regardless of Tanner stage (Po0.01). Plasma resistin moderately increased with central obesity in Tanner stages 2 and 4 (Po0.05) and strongly increased in Tanner stage 3 (Po0.001). In girls, resistin was also globally increased in the presence of central obesity (Po0.01), but the specific differences between each pubertal subgroup were not as consistent as with the boys. The only significant increase between subgroups was seen in the fourth Tanner stage (Po0.05). We subsequently verified resistin contribution to HOMAIR after correction for waist circumference. In these circumstances, resistin did not significantly contribute to predicting HOMA-IR in boys. However, after correction, resistin still contributed to predicting HOMA-IR in girls, but in a linearly negative way (P ¼ 0.037). The contribution was extremely small (o1%).

Relations between resistin and inflammation In boys, there was no correlation between plasma resistin and pro-inflammatory molecules such as C3 or CRP. Previously, a C3 cutoff point of 1.8 g l1 has been used to differentiate varying levels of coronary artery disease and vascular events.60 Similarly, in boys with high C3 levels (X1.8 g l1), no difference was found in average resistin levels and no group-specific correlation could be made between resistin and obesity, insulin resistance or MetS markers. However, significant positive correlations were found in girls between resistin and C3 (P ¼ 0.009) or CRP (P ¼ 0.004) after puberty onset (Tanner stages 2–5). In girls with pathological C3 levels, resistin was positively correlated with IFG (P ¼ 0.003). In the same group, Pearson’s analysis showed a link between resistin and fasting plasma glucose (P ¼ 0.023), but not with any other factor including obesity markers. They also had higher average resistin levels compared to girls with nonpathological C3 levels (18.9±0.8 vs 17.3±0.5 ng ml1), but the difference was not significant (P ¼ 0.316).

Predicting plasma resistin values Forward stepwise regression showed that resistin levels for boys can be predicted (R ¼ 0.161) using a linear combination of age (Po0.001), weight (Po0.001), fasting glucose (P ¼ 0.033), degree of obesity (P ¼ 0.006) and testicular volume (P ¼ 0.020). In prepubertal boys, no factor significantly predicted resistin levels. Resistin levels for girls can be


435

Figure 4 Central obesity (C. Ob) and resistin association. Average plasma resistin levels in normal weight, overweight (Owt) and obese (Ob) children with or without C. Ob are shown in boys (a) and in girls (b). Average plasma resistin levels in normal and positive for C. Ob children separated by Tanner stage are shown in boys (c) and in girls (d). Results are expressed as mean. Significance was calculated by two-way analysis of variance (ANOVA) test, where differences are expressed as *Po0.05, **Po0.01 and ***Po0.001.

predicted (R ¼ 0.204) with a linear combination of plasma NEFA (P ¼ 0.047), FMP (Po0.001), breast development based on Tanner stage (P ¼ 0.017) and HOMA-IR (P ¼ 0.034). In prepubertal girls, FMP (P ¼ 0.027) and IFG (P ¼ 0.028) significantly predicted resistin levels (R ¼ 0.306). Overall, obesity, insulin resistance and pubertal markers are the best predictors for resistin levels variation in both boys and girls, although the contribution is low.

Discussion Gender-specific differences As suggested previously in the literature, girls in our study also had higher average plasma resistin than boys. Interestingly, the difference seems more important in prepubertal children, despite the nonsignificant analysis. The difference between average resistin levels in girls and boys decreased through pubertal stages but might remain in adulthood, as previously proposed. Evidence, such as the significant expression of resistin mRNA in human maternal placenta,6 resistin regulation by gonadal hormones61 and genderrelated increase in resistin mRNA found in mice1 or rat adipose tissue61 suggests that the molecular pathway by which resistin levels are controlled could differ between males and females. This information could potentially explain the disparity between genders in resistin associations and stepwise regression analysis results. The very low amount of resistin variation explained with our analysis

also suggests that the main resistin determinants are still unknown and that the parameters measured are indirectly related. Realizing a stepwise regression with these factors, as many of them are known to be modulated differently by subtle and intrinsic gender-specific changes in lipid and carbohydrate metabolism, would obviously result again in important disparity between genders.

Age relation Many hormones such as ghrelin, leptin and adiponectin have been related to development,62 and resistin is now also a likely candidate. A strong link between resistin and development has been found in rats, where resistin mRNA was found to reach a peak at puberty and then decrease in the adipose tissue.61 As mentioned previously, increased maternal resistin levels through placental production also show evidence of a relationship between resistin and development. In the present study, our large sample size allowed us to find an inverse correlation that most other studies did not find, suggesting that human resistin levels decrease with age and puberty in both genders. In adults, resistin levels may no longer correlate with age, as reported in other studies. As in mice or rats, the mechanism by which this is achieved remains unidentified in humans.

Obesity correlation Plasma resistin correlated very strongly with many obesity markers in the present study, independently of gender and International Journal of Obesity


436 age, with waist-to-height ratio and FMP being the strongest factors. We have been able to distinguish the effects of simple elevated adiposity vs central obesity in the present study. There seems to be no increase in plasma resistin strictly related to adiposity alone in boys. However, the presence of central obesity in boys strongly associated with high resistin levels. In girls, adiposity and central obesity both contributed to elevated plasma resistin, providing another interesting gender-specific difference. The association of central obesity with resistin is supported in both genders by the strong linear correlation between plasma resistin and waist-to-height ratio, a good pediatric physical indicator of central obesity. Also supporting our findings, the amount of resistin mRNA expressed by visceral and subcutaneous abdominal fat was found vastly superior to that found in thigh or breast fat.63

Insulin resistance correlation Predicting the risk of developing insulin resistance and type 2 diabetes is increasingly important in every country, especially in the developing world. In China, diabetes among children is increasing in a dramatic way. In 2030, it is projected that diabetes prevalence will double in the Chinese adult population.50 In addition, Asian children have been found to be more prone to show impaired fasting glucose and high fasting insulin than Caucasian children.64 With resistin labeled as a potential insulin resistance marker, our study examined these relationships. To date, few studies have shown a positive correlation between insulin resistance markers and resistin in adults, and none has showed it in children. Moreover, very few children-specific studies have identified factors that could be related to the development of insulin resistance. Some studies have implicated fat distribution as a factor, as many children with central obesity show insulin resistance.65 However, it has been suggested that, in children, overall adiposity is a better insulin resistance marker.66,67 In the present study, there was no further increase of resistin when IFG was present in addition to central obesity, and independent resistin contribution to insulin resistance seemed inexistent or very small. Our large number of subjects allowed us to look extensively at many parameters to identify any significant differences. However, very little association was found between resistin and insulin resistance. Resistin is therefore, independently of gender or age, not an independent predictor of an obese and insulinresistant phenotype in children. However, we cannot rule out that resistin could still have a function in the development of a dysfunctional insulin signaling pathway. The lack of association between resistin and adiponectin is an interesting one considering the proposed, although not definitive, functions of these adipokines in protecting from (adiponectin) or promoting (resistin) insulin resistance. However, considering that insulin resistance parameters and plasma resistin concentrations in our study were independent, the absence of correlation with adiponectin International Journal of Obesity

is not that surprising. In fact, based on literature searches, the lack of correlation between these two parameters is quite common,24,44-47 leading us to hypothesize that although both adipokines may have specific functions in metabolism generally related to obesity and insulin homeostasis, their functions may be distinctly different and often unrelated.

MetS correlation Although some studies found links between MetS and resistin in adults, none to our knowledge has found a correlation in children. Interestingly, our sample size allowed us to find specific correlations among our groups. In boys, plasma resistin is strongly correlated with three MetS components: central obesity, HBP and HTG. Therefore, although resistin is globally a good predictor of MetS for boys, it is closely related only to some factors implicated in the diagnosis of MetS. For girls, central obesity was the only determining factor related to the MetS that was positively correlated with high resistin levels. Plasma resistin is therefore a weak MetS predictor, particularly in girls. Inflammation factors The presence of inflammatory cytokines tends to increase with age throughout adult life.68 However, resistin levels follow the opposite path in children and remain stable in adulthood, which indicates that resistin levels are not solely regulated by pro-inflammatory molecules as has been previously suggested. As many studies have shown in an adult population, we also found that resistin is increased in children with acute inflammation, but in a gender-dependent fashion. Interestingly, pubertal girls with acute inflammation showed no correlation between resistin levels and obesity markers, suggesting that the effects of pro-inflammatory molecules on resistin levels were not dependent on obesity. On the other hand, the strong correlation of resistin with central obesity, which is associated with inflammation,69 suggests that the increased plasma resistin observed in children with central obesity is linked to that proinflammatory state, which may emerge at a later time point. Supporting this idea, resistin is expressed at the highest levels in macrophages, cells of the immune system found in higher amount in adipose tissue in central obesity.70,71 Macrophages could therefore be another source contributing to the increased plasma resistin levels found in subjects with central obesity. It is relevant to note that an increase in inflammatory markers is not correlated with insulin resistance in children 66 and that resistin, therefore, cannot be linked with insulin resistance through inflammation.

Conclusion Our study was able to demonstrate that the typical profile of an Asian male child with high resistin levels (fifth quintile) is more likely to display increased waist circumference, FMP,


437 diastolic blood pressure, plasma TG and plasma NEFA. Very high levels of resistin are a global indicator of metabolic dysfunctions associated with visceral obesity in boys. On the other hand, the typical profile of an Asian girl with high resistin levels is less alarming. Resistin levels above the median are a weak predictor of adiposity, translated by mildly increased waist circumference, BMI and FMP. Children of either gender with high plasma resistin are not more likely to show insulin resistance. Classical markers such as plasma TG, HDL-Chol, LDL-Chol, fasting glucose or fasting insulin are far better markers than resistin for most parameters related to obesity, MetS and insulin resistance in the present cohort. Moreover, resistin seems to be strongly correlated with variables that are easily measurable by physical examination, such as obesity, FMP and blood pressure. Although resistin is not closely associated with global metabolic dysfunctions, it is a marker of fat distribution because it is specifically associated with abdominal fat depots. This correlation with central obesity could also be a link between the function of resistin in obesity and inflammation, mediated by macrophages. The elevated plasma resistin levels observed in children with central obesity, a known determinant of insulin resistance development in adults,72 could also be the indirect link between resistin and insulin resistance as the children mature to adults. The nature of the direct link between insulin resistance metabolism and resistin in humans, if there is any, remains unclear. Despite our large population, our study was not able to find any indices, supporting the existence of such a direct link in Asian children. With the very small amount of variation in resistin levels that can be explained with the parameters mentioned throughout our study, it is likely that resistin is implicated in a related pathway but that its main function remains unknown. The lack of information about resistin receptor prevents us from drawing stronger conclusions, as alterations to resistin sensitivity through receptor regulation could have an important function.

Acknowledgements This work was supported by grants from CIHR (KC: No. MOP-64446), Beijing Municipal Science and Technology Commission (JM: No. H030930030031), National Natural Science Foundation of China (JM: Nos. 30671804 and 30470645); Hubei Provincial Science Foundation (No. 2003CA022) and the FRSQ-NSFC Quebec-China exchange program (KC).

References 1 Steppan CM, Bailey ST, Bhat S, Brown EJ, Banerjee RR, Wright CM et al. The hormone resistin links obesity to diabetes. Nature 2001; 6818: 307–312. 2 Steppan CM, Lazar MA. Resistin and obesity-associated insulin resistance. Trends Endocrinol Metab 2002; 1: 18–23.

3 Patel L, Buckels AC, Kinghorn IJ, Murdock PR, Holbrook JD, Plumpton C et al. Resistin is expressed in human macrophages and directly regulated by PPAR gamma activators. Biochem Biophys Res Commun 2003; 2: 472–476. 4 Kim KH, Lee K, Moon YS, Sul HS. A cysteine-rich adipose tissuespecific secretory factor inhibits adipocyte differentiation. J Biol Chem 2001; 14: 11252–11256. 5 Fain JN, Cheema PS, Bahouth SW, Hiler ML. Resistin release by human adipose tissue explants in primary culture. Biochem Biophys Res Commun 2003; 3: 674–678. 6 Yura S, Sagawa N, Itoh H, Kakui K, Nuamah MA, Korita D et al. Resistin is expressed in the human placenta. J Clin Endocrinol Metab 2003; 3: 1394–1397. 7 Minn AH, Patterson NB, Pack S, Hoffmann SC, Gavrilova O, Vinson C et al. Resistin is expressed in pancreatic islets. Biochem Biophys Res Commun 2003; 2: 641–645. 8 Kawashima J, Tsuruzoe K, Motoshima H, Shirakami A, Sakai K, Hirashima Y et al. Insulin down-regulates resistin mRNA through the synthesis of protein(s) that could accelerate the degradation of resistin mRNA in 3T3-L1 adipocytes. Diabetologia 2003; 2: 231–240. 9 Shojima N, Sakoda H, Ogihara T, Fujishiro M, Katagiri H, Anai M et al. Humoral regulation of resistin expression in 3T3-L1 and mouse adipose cells. Diabetes 2002; 6: 1737–1744. 10 Delhanty PJD, Mesotten D, McDougall F, Baxter RC. Growth hormone rapidly induces resistin gene expression in white adipose tissue of spontaneous dwarf (SDR) rats. Endocrinology 2002; 6: 2445–2448. 11 Wolfing B, Neumeier M, Buechler C, Aslanidis C, Scholmerich J, Schaffler A. Interfering effects of insulin, growth hormone and glucose on adipokine secretion. Exp Clin Endocrinol Diabetes 2008; 1: 47–52. 12 Moon B, Kwan JJM, Duddy N, Sweeney G, Begum N. Resistin inhibits glucose uptake in L6 cells independently of changes in insulin signaling and GLUT4 translocation. Am J Physiol Endocrinol Metab 2003; 1: E106–E115. 13 Palanivel R, Maida A, Liu Y, Sweeney G. Regulation of insulin signalling, glucose uptake and metabolism in rat skeletal muscle cells upon prolonged exposure to resistin. Diabetologia 2006; 1: 183–190. 14 Palanivel R, Sweeney G. Regulation of fatty acid uptake and metabolism in L6 skeletal muscle cells by resistin. FEBS Lett 2005; 22: 5049–5054. 15 Sheng CH, Di J, Jin Y, Zhang YC, Wu M, Sun Y et al. Resistin is expressed in human hepatocytes and induces insulin resistance. Endocrine 2008; 33: 135–143. 16 Rajala MW, Lin Y, Ranalletta M, Yang XM, Qian H, Gingerich R et al. Cell type-specific expression, coregulation of murine resistin, resistin-like molecule-alpha in adipose tissue. Mol Endocrinol 2002; 8: 1920–1930. 17 Makimura H, Mizuno TM, Bergen H, Mobbs CV. Adiponectin is stimulated by adrenalectomy in ob/ob mice, is highly correlated with resistin mRNA. Am J Physiol Endocrinol Metab 2002; 6: E1266–E1271. 18 Qiao XZ, Wang XF, Xu ZR, Yang YM. Resistin does not down-regulate the transcription of insulin receptor promoter. J Zhejiang Univ Sci 2008; 4: 313–318. 19 Holcomb IN, Kabakoff RC, Chan B, Baker TW, Gurney A, Henzel W et al. FIZZ1, a novel cysteine-rich secreted protein associated with pulmonary inflammation, defines a new gene family. EMBO J 2000; 15: 4046–4055. 20 Harsch IA, Koebnick C, Wallaschofski H, Schahin SP, Hahn EG, Ficker JH et al. Resistin levels in patients with obstructive sleep apnoea syndromeFthe link to subclinical inflammation? Med Sci Monit 2004; 9: CR510–CR515. 21 Reilly MP, Lehrke M, Wolfe ML, Rohatgi A, Lazar MA, Rader DJ. Resistin is an inflammatory marker of atherosclerosis in humans. Circulation 2005; 7: 932–939.



438 22 Axelsson J, Bergsten A, Qureshi AR, Heimburger O, Barany P, Lonnqvist FL et al. Elevated resistin levels in chronic kidney disease are associated with decreased glomerular filtration rate, inflammation, but not with insulin resistance. Kidney Int 2006; 3: 596–604. 23 Bo S, Gambino R, Pagani A, Guidi S, Gentile L, Cassader M et al. Relationships between human serum resistin, inflammatory markers, insulin resistance. Int J Obes 2005; 11: 1315–1320. 24 Shetty GK, Economides PA, Horton ES, Mantzoros CS, Veves A. Circulating adiponectin, resistin levels in relation to metabolic factors, inflammatory markers, vascular reactivity in diabetic patients, subjects at risk for diabetes. Diabetes Care 2004; 10: 2450–2457. 25 Al-Daghri N, Chetty R, McTernan PG, Al-Rubean K, Al-Attas O, Jones AF et al. Serum resistin is associated with C-reactive protein & LDL cholesterol in type 2 diabetes, coronary artery disease in a Saudi population. Cardiovasc Diabetol 2005; 5: 4–10. 26 Stejskal D, Adamovska S, Bartek J, Jurakova R, Proskova J. ResistinFconcentrations in persons with diabetes mellitus of type 2 and in individuals with acute inflammatory disease. Biomed Pap 2003; 1: 63–69. 27 Filippidis G, Liakopoulos V, Mertens PR, Kiropoulos T, Stakias N, Verikouki C et al. Resistin serum levels are increased but not correlated with insulin resistance in chronic hemodialysis patients. Blood Purif 2005; 6: 421–428. 28 Kaser S, Kaser A, Sandhofer A, Ebenbichler CF, Tilg H, Patsch JR. Resistin messenger-RNA expression is increased by proinflammatory cytokines in vitro. Biochem Biophys Res Commun 2003; 2: 286–290. 29 Aquilante CL, Kosmiski LA, Knutsen SD, Zineh I. Relationship between plasma resistin concentrations, inflammatory chemokines, components of the metabolic syndrome in adults. Metab Clin Exp 2008; 4: 494–501. 30 Kunnari A, Ukkola O, Paivansalo M, Kesaniemi YA. High plasma resistin level is associated with enhanced highly sensitive Creactive protein, leukocytes. J Clin Endocrinol Metab 2006; 7: 2755–2760. 31 Norata GD, Ongari M, Garlaschelli K, Raselli S, Grigore L, Catapano AL. Plasma resistin levels correlate with determinants of the metabolic syndrome. Eur J Endocrinol 2007; 2: 279–284. 32 Martos-Moreno GA, Barrios V, Argente J. Normative data for adiponectin, resistin, interleukin 6, leptin/receptor ratio in a healthy Spanish pediatric population: relationship with sex steroids. Eur J Endocrinol 2006; 3: 429–434. 33 Vilarrasa N, Vendrell J, Maravall J, Broch M, Estepa A, Megia A et al. Distribution, determinants of adiponectin, resistin, ghrelin in a randomly selected healthy population. Clin Endocrinol (Oxf) 2005; 3: 329–335. 34 Degawa-Yamauchi M, Bovenkerk JE, Juliar BE, Watson W, Kerr K, Jones R et al. Serum resistin (FIZZ3) protein is increased in obese humans. J Clin Endocrinol Metab 2003; 11: 5452–5455. 35 Menzaghi C, Coco A, Salvemini L, Thompson R, De Cosmo S, Doria A et al. Heritability of serum resistin and its genetic correlation with insulin resistance-related features in nondiabetic Caucasians. J Clin Endocrinol Metab 2006; 7: 2792–2795. 36 Wasim H, Al-Daghri NM, Chetty R, McTernan PG, Barnett AH, Kumar S. Relationship of serum adiponectin and resistin to glucose intolerance and fat topography in South-Asians. Cardiovasc Diabetol 2006; 2: 5–10. 37 Burnett MS, Devaney JM, Adenika RJ, Lindsay R, Howard BV. Cross-sectional associations of resistin, coronary heart disease, and insulin resistance. J Clin Endocrinol Metab 2006; 1: 64–68. 38 Kunnari A, Ukkola O, Kesaniemi YA. Resistin polymorphisms are associated with cerebrovascular disease in Finnish type 2 diabetic patients. Diabet Med 2005; 5: 583–589. 39 Heilbronn LK, Rood J, Janderova L, Albu JB, Kelley DE, Ravussin E et al. Relationship between serum resistin concentrations and insulin resistance in nonobese, obese, and obese diabetic subjects. J Clin Endocrinol Metab 2004; 4: 1844–1848.


40 Kadowaki T, Yamauchi T. Adiponectin and adiponectin receptors. Endocr Rev 2005; 3: 439–451. 41 Lewandowski KC, Szosland K, O’Callaghan C, Tan BK, Randeva HS, Lewinski A. Adiponectin and resistin serum levels in women with polycystic ovary syndrome during oral glucose tolerance test: a significant reciprocal correlation between adiponectin and resistin independent of insulin resistance indices. Mol Gen Metab 2005; 1: 61–69. 42 Frankel DS, Vasan RS, D’Agostino RB, Benjamin EJ, Levy D, Wang TJ et al. Resistin, adiponectin, and risk of heart failure. J Am Coll Cardiol 2009; 53: 754–762. 43 Fargnoli JL, Fung TT, Olenczuk DM, Chamberland JP, Hu FB, Mantzoros CS. Adherence to healthy eating patterns is associated with higher circulating total and high-molecular-weight adiponectin and lower resistin concentrations in women from the Nurses’ Health Study. Am J Clin Nutr 2008; 5: 1213–1224. 44 Silha JV, Krsek M, Skrha JV, Sucharda P, Nyomba BL, Murphy LJ. Plasma resistin, adiponectin and leptin levels in lean and obese subjects: correlations with insulin resistance. Eur J Endo 2003; 4: 331–335. 45 Pagano C, Marin O, Calcagno A, Schiappelli P, Pilon C, Milan G et al. Increased serum resistin in adults with Prader-Willi syndrome is related to obesity and not to insulin resistance. J Clin Endocrinol Metab 2005; 7: 4335–4340. 46 Hasegawa G, Ohta M, Ichida Y, Obayashi H, Shigeta M, Yamasaki M et al. Increased serum resistin levels in patients with type 2 diabetes are not linked with markers of insulin resistance and adiposity. Acta Diabetol 2005; 42: 104–109. 47 Yaturu S, Daberry RP, Rains J, Jain S. Resistin and adiponectin levels in subjects with coronary artery disease and type 2 diabetes. Cytokine 2006; 34: 219–223. 48 Makimura H, Mizuno TM, Bergen H, Mobbs CV. Adiponectin is stimulated by adrenalectomy in ob/ob mice and is highly correlated with resistin mRNA. Am J Physiol Endocrinol Metab 2002; 6: E1266–E1271. ¨ ffler G. Influences of ionomycin, dibutyryl49 Kappes A, Lo cycloAMP and tumour necrosis factor-alpha on intracellular amount and secretion of apM1 in differentiating primary human preadipocytes. Horm Metab Res 2000; 11–12: 548–554. 50 Wild S, Roglic G, Green A, Sicree R, King H. Global prevalence of diabetesFestimates for the year 2000 and projections for 2030. Diabetes Care 2004; 5: 1047–1053. 51 Reinehr T, Roth CL, Menke T, Andler W. Resistin concentrations before and after weight loss in obese children. Int J Obes 2006; 2: 297–301. 52 Gerber M, Boettner A, Seidel B, Lammert A, Ba¨r J, Schuster E et al. Serum resistin levels of obese and lean children and adolescents: biochemical analysis and clinical relevance. J Clin Endocrinol Metab 2005; 90: 4503–4509. 53 Mi J, Cheng H, Hou DQ, Duan JL, Teng TT, Wang YF. Prevalence of overweight and obesity among children and adolescents in Beijing in 2004. Chin J Epidemiol 2006; 27: 474–479. 54 Marshall WA, Tanner JM. Puberty. In: Falkner F, Tanner JM (eds). Human Growth Vol 2 Plenum: New York, 1986. pp 171–210. 55 Group of China Obesity Task Force. Body mass index reference norm for screening overweight and obesity in Chinese children and adolescents. China J Epidemiol 2004; 25: 97–102. 56 Wan NJ, Mi J, Wang TY, Duan JL, Li M, Gong CX et al. Metabolic syndrome in overweight and obese schoolchildren in Beijing. Zhonghua Er Ke Za Zhi 2007; 45: 417–421. 57 Li M, Wu C, Song A, Zhang K. Development and preliminary application of enzyme-linked immunosorbent assay for human net insulin in serum. Chin J Endocrinol Metab 1997; 13: 214–217. 58 Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC. Homeostasis model assessmentFinsulin resistance and beta-cell function from fasting plasma-glucose and insulin concentrations in man. Diabetologia 1985; 7: 412–419.


439 59 Kuczmarski RJ, Ogden CL, Grummer-Strawn LM, Flegal KM, Guo SS, Wei R et al. CDC growth charts: United States. Adv Data 2000; 314: 1–27. 60 Szeplaki G, Prohaszka Z, Duba J, Rugonfalvi-Kiss S, Karadi I, Kokai M et al. Association of high serum concentration of the third component of complement (C3) with pre-existing severe coronary artery disease and new vascular events in women. Atherosclerosis 2004; 2: 383–389. 61 Nogueiras R, Gualillo O, Caminos E, Casanueva FF, Die´guez C. Regulation of resistin by gonadal, thyroid hormone, and nutritional status. Obes Res 2003; 11: 408–414. 62 Budak E, Fernandez Sanchez M, Bellver J, Cervero A, Simon C, Pellicer A. Interactions of the hormones leptin, ghrelin, adiponectin, resistin, and PYY3-36 with the reproductive system. Fertil Steril 2006; 6: 1563–1581. 63 McTernan PG, McTernan CL, Chetty R, Jenner K, Fisher FM, Lauer M et al. Increased resistin gene and protein expression in human abdominal adipose tissue. J Clin Endocrinol Metab 2002; 5: 2407–2410. 64 Whincup PH, Gilg JA, Papacosta O, Seymour C, Miller GJ, Alberti KGMM et al. Early evidence of ethnic differences in cardiovascular risk: cross sectional comparison of British South Asian and white children. Br Med J 2002; 7338: 635–638. 65 Freedman DS, Khan LK, Dietz WH, Srinivasan SR, Berenson GS. Relationship of childhood obesity to coronary heart disease risk factors in adulthood: the Bogalusa Heart Study. Pediatrics 2001; 3: 712–718.

66 Goran MI, Bergman RN, Gower BA. Influence of total vs. visceral fat on insulin action and secretion in African American and white children. Obes Res 2001; 8: 423–431. 67 Maffeis C, Manfredi R, Trombetta M, Sordelli S, Storti M, Benuzzi T et al. Insulin sensitivity is correlated with subcutaneous but not visceral body fat in overweight and obese prepubertal children. J Clin Endocrinol Metab 2008; 93: 2122–2128. 68 Miles EA, Rees D, Banerjee T, Cazzola R, Lewis S, Wood R et al. Age-related increases in circulating inflammatory markers in men are independent of BMI, blood pressure and blood lipid concentrations. Atherosclerosis 2008; 1: 298–305. 69 Fontana L, Eagon JC, Trujillo ME, Scherer PE, Klein S. Visceral fat adipokine secretion is associated with systemic inflammation in obese humans. Diabetes 2007; 4: 1010–1013. 70 Weisberg S, Mc Cann D, Desai M, Murphy E, Rosenbaum M, Leibel R et al. Obesity is associated with macrophage accumulation in adipose tissue. J Clin Invest 2003; 12: 1796–1808. 71 Xu HY, Barnes GT, Yang Q, Tan Q, Yang DS, Chou CJ et al. Chronic inflammation in fat plays a crucial role in the development of obesity-related insulin resistance. J Clin Invest 2003; 12: 1821–1830. 72 Lapidus L, Bengtsson C, Larsson B, Pennert K, Rybo E, Sjostrom L. Distribution of adipose-tissue and risk of cardiovascular-disease and deathFa 12 year follow up of participants in the population study of women in Gothenburg, Sweden. Br Med J 1984; 6454: 1257–1261.
