SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Methods and Procedures
  4. Results
  5. Discussion
  6. SUPPLEMENTARY MATERIAL
  7. Disclosure
  8. REFERENCES
  9. Supporting Information

Minor allele A of single-nucleotide polymorphism (SNP) 11391 G/A of ADIPOQ gene (rs17300539) has been consistently associated with higher adiponectin levels in adults and children. The aim of this study was to investigate the metabolic role of this variant in a large cohort of children of European origin. A total of 1,852 children from two general populations in Verona and in Fleurbaix–Laventie and from the Lille childhood obesity cohort, were genotyped and pooled together after checking for the absence of genetic heterogeneity for rs17300539 between Italian and French children. The genotype of rs17300539 was studied in relation to circulating adiponectin levels, BMI, fasting plasma glucose, fasting serum insulin (FSI), insulin resistance index (homeostasis model assessment of insulin resistance (HOMAIR)), high-density lipoprotein cholesterol, and triglycerides. After adjustment for known confounders, rs17300539 GA+AA carriers had 1.6 µg/ml higher adiponectin levels (P = 6 × 10−8) than GG carriers. They also showed higher BMI (B = 0.97, P = 0.015) and higher prevalence of obesity (OR = 1.35 (1.06–1.85), P = 0.015) than GG carriers. Before adjusting for obesity status, GA+AA carriers had higher FSI (B = 1.10, P = 0.040) and higher HOMAIR (B = 0.31, P = 0.020) than GG carriers. After adjustment for obesity status, they did not differ from GG carriers for any metabolic parameter, either among obese or nonobese children. The rs17300539-A variant, though consistently associated with higher adiponectin levels, does not exert any appreciable protective metabolic effect in children, either in the presence or absence of obesity. In contrast, this SNP may increase the risk for childhood obesity and related insulin resistance.

Adiponectin is an adipocyte-secreted hormone with well-known pleiotropic metabolic effects (1). Even during childhood and adolescence, adiponectin levels are, independent of the adiposity status, associated with metabolic syndrome components (2), and with risk markers for atherogenesis (3).

Genetic factors account for a large amount of adiponectin variance (4) and variation at the ADIPOQ locus, the adiponectin-encoding gene, has been shown to modulate adiponectin concentration in both adults (5) and children (6,7). One of the single-nucleotide polymorphisms (SNPs) located in the promoter region of this gene, ADIPOQ −11391 G/A (rs17300539) has been consistently found associated with plasma adiponectin concentration in adults and children (5,6,7,8). This SNP also confers enhanced transcriptional activity to ADIPOQ promoter in “in vitro” assays (6). The metabolic impact of rs17300539 is still elusive. The minor A allele has been associated with lower risk of type 2 diabetes and insulin resistance phenotypes in obese adults (9,10,11) and with lower insulin levels and homeostasis model of insulin resistance (HOMAIR) in obese children (7). On the other hand, given adiponectin effect on insulin sensitivity and adipocyte proliferation and differentiation (12,13), rs17300539-A variant was previously hypothesized to play a detrimental role in the risk for early onset obesity, although this was not supported probably because of a lack of statistical power (6).

This study aimed (i) to analyze further the role of rs17300539-A variant in adiponectin modulation in a large sample-set of European children; (ii) to search for a possible role of this variant in the risk for childhood obesity; (iii) to assess whether and how the possible association between rs17300539 and obesity modulates metabolic effects on fasting plasma glucose, fasting serum insulin (FSI), insulin resistance index HOMAIR and lipid profile in childhood; (iv) to search for possible protective independent associations of rs17300539-A variant with these parameters in obese and nonobese children separately.

Methods and Procedures

  1. Top of page
  2. Abstract
  3. Methods and Procedures
  4. Results
  5. Discussion
  6. SUPPLEMENTARY MATERIAL
  7. Disclosure
  8. REFERENCES
  9. Supporting Information

Subjects

Participants were children and adolescents coming from three independent recruitments (Table 1): (i) 644 Italian children coming from the general population of Verona, whose families were randomly extracted from the registry office database of the town, and contacted by post; (ii) 446 French children and adolescents coming from 294 nuclear families recruited in the North Region of France, as participants to the Fleurbaix and Laventie Ville Santé Study II, a longitudinal epidemiological study on the determinants of weight gain (14); (iii) 762 French children and adolescents, coming from 449 nuclear families with at least one obese offspring, recruited in the Paediatric Endocrine Unit of Jeanne de Flandres Hospital, Lille or through national media campaign (Lille cohort).

Table 1.  Characteristics of French, Italian, and total pooled populations
inline image

Exclusion criteria were non-Caucasian ethnicity, a history of diabetes or any other chronic disease, dietary regimen at present or followed in the past.

Height and weight were measured during clinical examination. BMI was calculated as weight (kilograms) divided by height (meters) squared. Obesity was assessed according to the International Obesity Task Force recommendations, by using the Cole's cutoffs for BMI (15).

Pubertal stage was recorded according to the Tanner classification (16).

In each country, a local Committee approved the study, which was sponsored by local Universities: Lille 2 University (France) and University of Verona (Italy). All children's families provided their written informed consent and the study was performed according to the Helsinki Declaration agreements.

Biochemical measurements

Biochemical measurements were performed in two local laboratories, one for each country. In both countries, fasting plasma glucose was dosed using the glucose oxidase method, and triglycerides and high-density lipoprotein-cholesterol were dosed using enzymes. FSI and serum adiponectin were dosed by the same kits in the two laboratories (Bi-Insulin IRMA kit; Sanofi Pasteur, Marnes la Coquette, France and Adiponectin RIA kit; Linco Research, St Charles, MO, respectively).

Male gender was significantly associated with lower adiponectin levels (B = −1.6 µg/ml, P = 0.001) in Fleurbaix–Laventie cohort only, probably because of the higher mean age of this cohort in respect to the other two ones, implying a more important pubertal influence of gender on adiponectin levels.

HOMAIR was calculated as plasma glucose (mmol/l) × serum insulin (mU/ml)/22.5 (17).

None of the dosed metabolites had significantly different means across the two laboratories, after adjustment for age, gender, and BMI (all P > 0.05).

Genotyping

DNA was extracted from peripheral leukocytes. Primers were synthesized by MWG Biotech. Rs17300539 was genotyped by Light-Cycler technology (Roche, Mannheim, Germany). Method was previously validated by calculation of the genotyping error rate from duplicate genotypes (6). Each genotyping plate was checked for Hardy–Weinberg equilibrium by χ2-test, after selection of one individual for each sib pair or sib group, when appropriate.

Statistical analysis

French and Italian children were pooled together after checking for genetic homogeneity at rs17300539 by Woolf's test (P > 0.05).

To take into account familial relationship within the French cohorts, we tested the association of rs17300539 with obesity binary trait and quantitative traits (BMI, serum adiponectin, plasma glucose, serum insulin, HOMAIR, triglycerides, high-density lipoprotein-cholesterol) by binomial and Gaussian models of generalized estimated equations, respectively, with a specific correction term for relatedness, using STATA software (18).

Two models were drawn for quantitative metabolic traits in the pooled population: the first adjusted for gender, age, and pubertal stage; the second adjusted also for obesity status, in order to dissect obesity-mediated SNP effects on quantitative traits, from obesity-independent ones. Stratified analyses in pooled control and pooled obese children separately were also performed when analyzing metabolic parameters, in order to discover possible obese or nonobese children—restricted roles of rs17300539 on metabolism.

A meta-analysis of the three cohorts analyzed solely was performed following the pooled analysis of rs17300539 effects on adiponectin levels and metabolic parameters, in order to check for possible heterogeneity and results inconsistency through the different populations. Library r Meta of R software, version 2.8.1 (http:www.r-project.org), was used to this purpose.

Taking into consideration the low prevalence of the minor allele and the statistical power reduction due to interaction analyses, we decided to investigate rs17300539 effects on quantitative traits under the dominant model (GG carriers vs. GA+AA carriers).

We preferred nonstandardized B-coefficients to standardized B-coefficients while describing quantitative effect sizes, because we aimed to easily quantify the variations of continuous dependent parameters for unitary changes of the independent variables.

All skewed variables (P < 0.05 at Kolmogorov–Smirnov test) were previously logarithmically transformed to obtain normal distribution before analyses.

We calculated the minimal detectable effect size of rs17300539-A variant on the metabolic parameters, with a statistical power >0.8 assumed, by the QUANTO 1.2 software (http:hydra.usc.eduGxE).

Statistical significance was fixed at P = 0.05.

Results

  1. Top of page
  2. Abstract
  3. Methods and Procedures
  4. Results
  5. Discussion
  6. SUPPLEMENTARY MATERIAL
  7. Disclosure
  8. REFERENCES
  9. Supporting Information

In the pooled child population, the allele A of rs17300539 was found to be nominally associated with obesity risk (OR= 1.33 (1.03–1.71), P = 0.023; OR = 1.35 (1.06–1.85), P = 0.015, under additive and dominant models, respectively) (Table 2).

Table 2.  rs17300539 G/A and risk for obesity
inline image

The GA+AA carriers presented higher circulating adiponectin levels (B = 1.15, P = 0.002) but also higher BMI, Z-BMI, FSI, and HOMAIR than wild-type children (B = 0.97, P = 0.015; B = 0.29, P = 0.031; B = 1.10, P = 0.040; B = 0.31, P = 0.020, respectively) (Table 3).

Table 3.  Role of rs17300539 of ADIPOQ in the modulation of common metabolic parameters in the whole pooled population and in the obesity-related strata
inline image

After adjustment for the obesity status, adiponectin levels were even more significantly different between genotypes, with a 1.6 µg/ml higher adiponectin levels in GA+AA carriers compared to GG carriers (B = 1.60, P = 6 × 10−8). In contrast, rs17300539 did not show significant association with FSI and HOMAIR anymore (Table 3).

The association between rs17300539 and circulating adiponectin, as well as the lack of association between this SNP and all other metabolic parameters, was confirmed in both pooled nonobese and pooled obese children analyzed separately, and no significant genotype × obesity status interaction was found on metabolic quantitative traits (Table 3).

The meta-analysis showed a consistent, homogeneous effect of rs17300539 on adiponectin levels and confirmed the lack of any independent SNP role on all other metabolic parameters across the three populations (Supplementary Table S1).

HOMAIR, high-density lipoprotein-cholesterol, and triglycerides were significantly associated with adiponectin levels (Table 4). However, for each of the three variables, expected improvement produced by 1.6 µg/ml adiponectin increase, i.e. by the SNP's effect, was so small to be not statistically detectable (Table 4).

Table 4.  Variation of metabolic parameters according to adiponectin levels; expected metabolic effects of rs17300539 A/X genotype; minimal statistically detectable metabolic effects of rs17300539 A/X genotype
inline image

To explain the lack of any favorable independent associations between rs17300539-A allele and metabolic parameters despite higher adiponectin levels, we investigated the existence of clinical associations between adiponectin levels and metabolic parameters.

Discussion

  1. Top of page
  2. Abstract
  3. Methods and Procedures
  4. Results
  5. Discussion
  6. SUPPLEMENTARY MATERIAL
  7. Disclosure
  8. REFERENCES
  9. Supporting Information

Our study first confirms the association between rs17300539-A allele of ADIPOQ and higher levels of adiponectin in children, which was previously reported in smaller European samples (6,7). This association was not significantly affected by obesity status. Moreover, this association was consistent through the three separate cohorts at meta-analysis despite significant variation of participant's age and type of recruitment among cohorts, suggesting a homogeneous effect during childhood and adolescence. Carrying at least one minor allele A, increased adiponectin level of 1.60 µg/ml (14% of mean adiponectin concentration in wild-type children), which is remarkably similar to what was reported in adult populations (+1.64 µg/ml) (5). This supports a constant effect of this ADIPOQ SNP on adiponectin levels over life time (5). “In vitro” evidence of enhanced transcriptional activity of A allele in respect to the common allele G, brings a reliable functional explication for such a consistent association (6).

The rs17300539-A variant may contribute to the risk of early obesity through chronic hyperadiponectinemia, which probably favors lipid storage both indirectly, by increased insulin sensitivity, and directly, by autocrine action on preadipocytes and mature adipocytes (6,12,13). High-fat diet fed transgenic ob/ob mice with increased adiponectin levels within physiological range, exhibit markedly higher age-dependent weight gains than their ob/ob counterparts, suggesting a causal link between baseline higher adiponectin and increased risk of obesity over time (13). Adiponectin has been also suggested to increase food intake by a direct action at the brain level: its trimers and hexamers have been shown to enter the cerebrospinal fluid in mice and have an orexygenic function through AdipoR1 receptor activation in the arcuate hypothalamus (19).

Our study also provides by pooled analysis the first evidence of positive association between rs17300539-A variant and childhood obesity.

Despite its association with higher adiponectin levels, ADIPOQ rs17300539-A variant showed an obesity-mediated detrimental association with higher FSI and HOMAIR in the pooled population and, after adjusting and stratification for obesity status, this variant did not show any favorable independent association with metabolic parameters, either in the pooled population or in obese or nonobese children analyzed solely. The meta-analysis showed a consistent lack of any favorable association between rs17300539-A allele and the metabolic parameters through the three cohorts.

This result is in agreement with data issued from the meta-analysis of >9,000 adult subjects and longitudinal data from the Framingham Offspring Study, which did not show any association between rs17300539-A allele and protection against type 2 diabetes (5,8).

As a matter of facts, the 1.6 µg/ml increase in adiponectin concentration due to the ADIPOQ rs17300539-A variant, is associated with so modest improvements of the metabolic parameters we tested, that they were not statistically detectable, despite an adequate sample size (Table 4). This supports the previous hypothesis that the lack of association of ADIPOQ variation with metabolic end points may be due to both cumulative modest genotypic effects on circulating adiponectin levels, and equally modest impact of a small change in total adiponectin levels on metabolism (8). As further confirmation, the ADIPOQ transgenic ob/ob mice, which display a more favorable metabolic profile than their nontransgenic counterparts, actually have two- to threefold higher adiponectin levels, similar to levels obtained upon exposure to peroxisome proliferator–activated receptor-γ agonists (13).

Therefore, the increase in adiponectin levels observed in GA+AA carriers seems not to be large enough to imply any appreciable metabolic advantage in general population and among obese children, where it fails to counterbalance the metabolic damages of obesity. It seems then possible to rule out the hypothesis that rs17300539 may be associated with the so-called (20) “metabolically healthy obesity”.

This study has some limitations. First of all, the association between rs17300539-A allele and early obesity has modest significance and needs further replication. Moreover, the cross-sectional design of the study does not permit to exploit the robust association between rs17300539 and adiponectin levels in order to provide, by Mendelian randomization, a genetic unconfounded assessment of cause–effect link between higher adiponectin and obesity risk, due to the well-expected inverse association between the hormone and obesity; in contrast, a prospective design, with genotype and adiponectin at baseline and obesity as clinical outcome, may be more suitable to this purpose. Finally, the possible effect of rs17300539 on the distribution of the three different complexes of adiponectin should be investigated.

In conclusion, ADIPOQ rs17300539-A variant, though definitely increasing circulating adiponectin levels, does not exert any appreciable protective metabolic effect in children. In contrast, it may contribute to the increased risk for childhood obesity and related insulin resistance.

REFERENCES

  1. Top of page
  2. Abstract
  3. Methods and Procedures
  4. Results
  5. Discussion
  6. SUPPLEMENTARY MATERIAL
  7. Disclosure
  8. REFERENCES
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Methods and Procedures
  4. Results
  5. Discussion
  6. SUPPLEMENTARY MATERIAL
  7. Disclosure
  8. REFERENCES
  9. Supporting Information

supporting Information

FilenameFormatSizeDescription
oby_1810_sm_1.doc37KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.