rs9939609 in the FTO Gene is Associated with Obesity but not with Several Biochemical Parameters in Sardinian Obese Children
Dr. Sandro Loche, Servizio di Endocrinologia Pediatrica, Ospedale Microcitemico – ASL Cagliari, Via Jenner, 09121 Cagliari, Italy. Tel: +39 070 6095536; Fax: +39 070 6095657; E-mail: email@example.com
Several studies have reported an association of the intronic single nucleotide polymorphism (SNP) rs9939609 of the fat mass and obesity-associated (FTO) gene with obesity and with a number of obesity-related features. We studied the association of rs9939609 with obesity in 912 obese children and adolescents (426 males and 486 females, mean ± SD age 10.5 ± 3.3 years) and in 543 normal weight subjects. A number of biochemical and clinical parameters was also evaluated in 700 of these patients. In the obese group, mean body mass index standard deviation score (BMI-SDS) was similar between the three genotypes. The A allele was present in 55% of the patients’ and in 43% of controls’ chromosomes. The distribution of heterozygotes was similar between patients and controls (47%), while the distribution of AA homozygotes was significantly higher in patients (31% vs. 20%). Logistic regression analysis on the genotypes yielded a χ2 of 35.5 with an odds ratio of 1.6 (CI = 1.3–1.8), P < 1 × 10−5. None of the clinical and metabolic parameters tested was associated with the genotype. In conclusion, we have confirmed the strong association between FTO and obesity, and shown that only AA homozygotes are predisposed to develop obesity while TT homozygotes might be protected. Finally, we found no association between rs9939609 and a number of obesity-related abnormalities.
With the advent of genome-wide association studies (GWAS), several susceptibility loci for complex traits have been identified. One of the most striking is probably represented by the fat mass and obesity-associated (FTO) gene, reported by several groups as a locus associated with weight gain and obesity. Following the first studies in 2007 (Dina et al., 2007; Fraylinget al., 2007; Scott et al., 2007; Scuteri et al., 2007), the association of polymorphisms in intron 1 of the FTO gene with increased BMI has been confirmed in all populations studied so far (Hinney et al., 2007; Pascoe et al., 2007; Andreasen et al., 2008; Cecil et al., 2008; Loos et al., 2008; Do et al., 2008; Freathy et al., 2008; Grant et al., 2008; Haupt et al., 2008; Hertel et al., 2008; Hotta et al., 2008; Hubacek et al., 2008; Hunt et al., 2008; Kring et al., 2008; Ng et al., 2008; Rampersaud et al., 2008; Tönjes et al., 2008; Villalobos-Comparán et al., 2008). Among these polymorphisms, the single nucleotide polymorphism (SNP) rs9939609 is one of the most extensively studied, and its A allele is found more frequently among overweight and obese individuals than in normal weight subjects. While this SNP appears consistently associated with the obesity phenotype, conflicting results have been reported on its association with a number of clinical and biochemical obesity-related traits (Al-Attar et al., 2008; Freathy et al., 2008; Hubacek et al., 2008; Kring et al., 2008; Müller et al., 2008; Sanghera et al., 2008; Doney et al., 2009; Hakanen et al., 2009; Lappalainen et al., 2010; Xi et al., 2010). Furthermore, most association studies have been performed in general populations or in groups of individuals including normal weight as well as severely obese subjects, and only few studies have been performed in homogeneous groups of phenotypically selected individuals (Müller et al., 2008; Sanghera et al., 2008). Moreover, due to the lack of phenotypic and/or genetic homogeneity of the studied cohorts, it is difficult to understand whether any eventual association with clinical and endocrine-metabolic traits is linked to the FTO genotype or represents independent findings.
The aim of our study was to test the association of rs9939609 with obesity in a selected cohort of obese children and adolescents. To this end we performed a case/control study using a control group of healthy blood donors. Both patients and controls were selected from the genetically homogeneous population of Sardinia. We think that the genetic homogeneity of our population, although having documented differences between subgroups (Zavattari et al., 2000), may play a role in facilitating genetic association studies. In fact, susceptibility and protective alleles could indeed be present in quite distinct and well preserved haplotype blocks. This would minimize the breakdown of linkage disequilibrium (quite typical in mixed populations) and the resulting high number of different haplotypes bearing one allelic variant rather than the other. This would generate a higher probability of revealing the association of the etiologic variant with obesity.
In the obese group, we have also studied possible associations between rs9939609 and a number of obesity-related biochemical parameters.
Materials and Methods
The association study for BMI was conducted in 912 Sardinian obese children and adolescents (426 boys and 486 girls, mean ± SD age 10.5 ± 3.3 years) recruited from the Pediatric Endocrinology Unit at the Microcitemico Hospital in Cagliari. Among this group, 700 subjects were extensively evaluated for a number of biochemical parameters. Their main clinical and laboratory findings are summarized in Table 1. Obesity was defined in children with a BMI > 95th percentile according to the Italian BMI reference charts (Cacciari et al., 2006). The control group consisted of 543 Sardinian healthy blood donors (250 males and 293 females, mean ± SD age 34.2 ± 7.7 years) with normal BMI (<24 kg/m2). The control group was only used in the first analytical step, i.e., comparison of allelic and genotypic frequencies between obese patients and controls. For this purpose, the control group was randomly selected among healthy normal weight subjects. All subjects were biologically unrelated. The study was approved by the local Ethical Committee and informed consent was obtained from all subjects and/or their legal guardians.
Table 1. Clinical and biochemical characteristics of 700 obese children in relation to their genotype at the rs9939609 SNP.
|Age|| 12.7|| 12.7|| 12.7|| 12.9|
|BMI-SDS|| 2.65 (0.03)|| 2.67 (0.03)|| 2.59 (0.02)|| 2.76 (0.13)|
|SBP (mmHg)||105.57 (0.56)||104.79 (0.91)||106.01 (0.87)||105.74 (1.19)|
|DBP (mmHg)|| 61.76 (0.34)|| 61.51 (0.57)|| 61.71 (0.51)|| 62.25 (0.75)|
|Insulinemia (mU/L)|| 15.23 (0.34)|| 15.77 (0.65)|| 14.79 (0.45)|| 15.44 (0.84)|
|OGGT 0′ (mmol/L)|| 4.90 (0.01)|| 4.91 (0.03)|| 4.90 (0.02)|| 4.89 (0.03)|
|OGTT 120′ (mmol/L)|| 5.82 (0.04)|| 5.86 (0.07)|| 5.85 (0.05)|| 5.69 (0.08)|
|HOMA-IR|| 3.35 (0.08)|| 3.46 (0.15)|| 3.26 (0.10)|| 3.39 (0.19)|
|Uricemia (μmol/L)||305.13 (9.95)||315.24 (11.30)||299.18 (7.73)||301.56 (11.90)|
|Total cholesterol (mmol/L)|| 4.35 (0.03)|| 4.35 (0.06)|| 4.37 (0.04)|| 4.34 (0.07)|
|HDL cholesterol (mmol/L)|| 1.33 (0.01)|| 1.32 (0.02)|| 1.33 (0.02)|| 1.34 (0.02)|
|LDL cholesterol (mmol/L)|| 2.70 (0.03)|| 2.70 (0.05)|| 2.71 (0.04)|| 2.67 (0.06)|
|Tryglicerides (mmol/L)|| 0.71 (0.02)|| 0.73 (0.03)|| 0.71 (0.02)|| 0.68 (0.03)|
|AST (U/L)|| 24.34 (0.41)|| 24.80 (0.47)|| 24.49 (0.79)|| 23.28 (0.49)|
|ALT (U/L)|| 22.61 (0.42)|| 23.95 (0.83)|| 22.33 (0.59)|| 21.28 (0.82)|
|TSH (mU/L)|| 2.71 (0.14)|| 2.52 (0.16)|| 2.87 (0.27)|| 2.59 (0.16)|
|Ft3 (pmol/L)|| 7.04 (0.08)|| 6.99 (0.09)|| 7.22 (0.15)|| 6.73 (0.14)|
|Ft4 (pmol/L)|| 17.63 (0.26)|| 17.63 (0.26)|| 15.50 (0.39)|| 17.76 (0.26)|
DNA was extracted from peripheral whole blood by the salting-out method. Polymerase chain reaction (PCR) and genotyping were performed using an Applied Biosystems (ABI, Foster City, CA) TaqMan 7000 system. SNP primers and probes for rs9939609 polymorphism were provided by ABI Assay-on-demand. Allelic frequencies did not deviate from the expected Hardy–Weinberg equilibrium, in both obese patients and controls (P= 0.14 and P= 0.29, respectively). Double genotyping of 10% of the cohort yielded a concordance rate of 100%.
Clinical and Laboratory Parameters
Systolic blood pressure (SBP), diastolic blood pressure (DBP), serum concentration of insulin, uric acid, total, high density lipoprotein (HDL), and low density lipoprotein (LDL) cholesterol, tryglicerides, aspartate transferase (AST), alanine transferase (ALT), thyroid-stimulating hormone (TSH), free triiodothyronine (Ft3), and free thyroxine (Ft4) were evaluated in all obese subjects. Oral glucose tolerance test (OGTT) was also performed in all patients according to clinical recommendations for children. After an overnight fast, 1.75 grams of glucose per kilogram of body weight were given orally, up to a maximum of 75 g. Serum glucose was measured in blood samples taken at time 0 and after 120 min. Homeostasis model assessment for insulin resistance (HOMA-IR) was calculated according to the formula: HOMA-IR (mmol/L ×μU/mL) = fasting glucose (mmol/L) × fasting insulin (μU/mL)/22.5. Glucose was measured using the glucose oxidase method (Autoanalyzer, Beckman Coulter, USA).
Ft3, Ft4, and TSH were determined by immunochemiluminescent assay (Immulite 2000, Siemens Healthcare Diagnostics, Deerfield, USA). Sensitivity of the assays was 1.5 pmol/L (Ft3), 4.0 pmol/L (Ft4), and 0.03 mU/L (TSH), and intra- and inter-assay coefficients of variation were 3.2% and 5% (Ft3), 3.3% and 4.1% (Ft4), and 2.1% and 3.1% (TSH), respectively. Blood glucose levels were measured using the glucose-oxidase method. Serum insulin levels were measured using a commercial radioimmunoassay (Aldatis, Rome, Italy). Sensitivity was 0.3 μU/mL with intra- and inter-assay coefficient of variation of 2.3% and 3.5%, respectively. All reagents were provided by Medical Systems Corporation (Genoa, Italy). Uric acid, LDL, HDL and total cholesterol, tryglicerides, AST, and ALT were determined by conventional methods.
Body mass index standard deviation score (BMI-SDS), derived from the Italian reference data (Cacciari et al., 2006), was used in the analyses to compare BMI in children of different ages. BMI-SDS was calculated by the least mean squares algorithms (LMS) method comparing the calculated BMI with the distribution of BMI in a standard population with the same age and gender in order to adjust BMI between different children for growth and development for comparison. The case/control association analysis was evaluated by conducting a logistic regression analysis using the STATA Data Analysis and Statistical Software, without adjusting for age and sex. Given the size of our sample sets, we calculated a statistical power at 75.0% (P < 0.05) to detect a genetic effect with an OR = 1.2, at 96.0% to find an OR = 1.3 and 100% to find an OR ≥ 1.4, for variants with minor allele frequency equal to 0.43, such as the rs9939609 variant. Nonparametric one-way analysis of variance was used to test differences among the three genotypes for each clinical and biochemical parameters. Normality of the data was evaluated using the Kolmogorov–Smirnov test.
Mean BMI-SDS in the obese group was similar between the three genotypes (Table 1). The distribution of rs9939609 allele (A and T) frequencies between patients and controls is shown in Table 2. The predisposing A allele was present in 55% of the patients’ and in 43% of controls’ chromosomes. The distribution of heterozygotes was similar between patients and controls (47%), while the distribution of homozygotes for the susceptibility allele A, was significantly higher in patients (31% vs. 20%, respectively) (Table 2). Logistic regression analysis on the genotypes yielded a χ2 of 35.5 with an odds ratio of 1.6 (CI = 1.3–1.8), P < 1 × 10−5. None of the clinical and endocrine-metabolic parameters tested was associated with the genotype (Table 1).
Table 2. Comparison of allelic and genotypic frequencies between patients and controls at the rs9939609 SNP.
We have confirmed a strong association between the FTO intronic rs9939609 polymorphism alleles and obesity. In our case–control association study on a large sample of Sardinian obese children and adolescents, the A allele was significantly more frequent in patients’ (55%) than in controls’ chromosomes (43%). These results are consistent with previous reports (Scuteri et al., 2007; Frayling et al., 2007; Scott et al., 2007; Dina et al., 2007; Hinney et al., 2007; Pascoe et al., 2007; Andreasen et al., 2008; Cecil et al., 2008; Loos et al., 2008; Do et al., 2008; Freathy et al., 2008; Grant et al., 2008; Haupt et al., 2008; Hertel et al., 2008; Hotta et al., 2008; Hubacek et al., 2008; Hunt et al., 2008; Kring et al., 2008; Ng et al., 2008; Rampersaud et al., 2008; Tönjes et al., 2008; Villalobos-Comparán et al., 2008; Xi et al., 2010), and clearly show that a group of individuals carefully selected for the phenotype is much more informative than a general population. In fact, in our cohort, we found an OR of 1.6, supported by a statistical significance stronger than that previously reported in most association studies performed in case/control groups of comparable size (Hotta et al., 2008; Kring et al., 2008; Xi et al., 2010). Calculation of statistical power demonstrated the ability of the dataset used for this study to reveal genetic effects with an OR even lower than that actually found. It is worth noting that the distribution of heterozygotes is almost superimposable between patients and controls. In previous reports, data were analyzed either by pooling heterozygotes with AA (Cecil et al., 2008; Zimmermann et al., 2009), or with TT homozygotes (Kring et al., 2008; Hakanen et al., 2009), indicating that the mode of transmission was not clearly defined. Interestingly, whereas homozygous subjects for the A allele are predisposed to develop obesity, subjects homozygous for the T allele might be protected from gaining weight, as suggested by the fact that in our cohort the frequency of TT homozygotes is 34% in lean subjects compared to 22% among obese individuals. In fact, although a number of functional studies have been reported on the expression of the FTO gene, it is difficult to determine whether one of the two alleles considered here determines a gain or loss of function or whether both have two different functions, i.e., one predisposing, and the other protective. For this reason, we want to highlight the fact that although the majority of the association studies point to the predisposing effect of the A allele as well as of other predisposing alleles and haplotypes, the protective role of the TT genotype should also be considered, a hypothesis already put forward by others (Zimmermann et al., 2009; Wardle et al., 2009).
The lack of association between rs9939609 genotype and BMI-SDS was not surprising, since in the obese group studied the BMI variability was clearly reduced compared to the general population. However, we felt it was worth looking for associations with other clinical and biochemical features to investigate for a possible contribution of the rs9939609 genotype in influencing these parameters, controlling for BMI. We found no association between the SNP and any of the clinical and biochemical parameters studied. These observations are in partial agreement with some previous reports (Freathy et al., 2008; Hubacek et al., 2008; Kring et al., 2008; Müller et al., 2008; Sanghera et al., 2008; Xi et al., 2010) but contrast with others (Doney et al., 2009; Hakanen et al., 2009; Lappalainen et al., 2010). This is also not surprising since most of these studies have been performed in general populations or in samples not selected for the obese phenotype (Freathy et al., 2008; Hubacek et al., 2008; Doney et al., 2009; Hakanen et al., 2009; Lappalainen et al., 2010), or were case–control studies (Kring et al., 2008; Sanghera et al., 2008; Xi et al., 2010), while we analyzed the possible association of the genotype with clinical and endocrine-metabolic parameters. Although the reported results are very variable, they seem to indicate that it is obesity itself and not the genotype which influences the association with a given parameter. In fact, it has been shown that these associations are no longer found when adjusted for BMI (Freathy et al., 2008; Kring et al., 2008), suggesting that in a selected cohort of obese subjects, such as in our study, clinical and biochemical parameters may not be associated with the FTO genotype. Our data are also consistent with the results of Müller et al. (2008) who found a positive association between SNP rs9939609 and obesity in a case–control study of German overweight/obese children and adolescents. However, a number of metabolic parameters including blood glucose, triglycerides, HDL cholesterol, and LDL cholesterol were not associated with the genotype. In this regard it is also noteworthy that, in our study, mean BMI-SDS was similar between the three different genotypes. Thus, abnormalities of endocrine-metabolic parameters typically found in obese subjects appear to be linked to the obesity state, independently of the genotype. Our finding does not contradict those obtained from large GWAS, since in the latter studies a great number of individuals with different body weights ranging from lean to severely obese are pooled together.
As previously reported (Zavattari et al., 2010), our study population has the advantage of being large and genetically homogeneous. Thus, we are confident that the lack of association between the FTO genotype and the clinical and endocrine-metabolic parameters in our study, truly reflects the absence of any direct involvement of FTO in endocrine-metabolic homeostasis.
In conclusion, our results confirm the strong association between FTO and obesity. We have also shown that only AA homozygous subjects are predisposed to develop obesity, and that individuals carrying the TT genotype might be protected. Finally, we found no association between rs9939609 and a number of obesity-related clinical and biochemical parameters.
This work was supported by a Grant from Regione Autonoma della Sardegna to SL. We are gratefully indebted to our nursing staff (Donatella Arghittu, Valentina Bianco, Patrizia Sanna), and our lab technicians (Maria Grazia Contini, Danilo Mosino, Teresa Trogu) for their continuous and skilful help in patients’ care and sample processing. We wish to thank Elena Bacchelli for her skilful contribution to the statistical analysis and critical discussion of the data.
The authors have nothing to disclose.