Does the FTO gene interact with the socioeconomic status on the obesity development among young European children? Results from the IDEFICS study

Abstract

Background:

Various twin studies revealed that the influence of genetic factors on psychological diseases or behaviour is more expressed in socioeconomically advantaged environments. Other studies predominantly show an inverse association between socioeconomic status (SES) and childhood obesity in Western developed countries. The aim of this study is to investigate whether the fat mass and obesity-associated (FTO) gene interacts with the SES on childhood obesity in a subsample (N=4406) of the IDEFICS (Identification and prevention of Dietary- and lifestyle-induced health EFfects In Children and infantS) cohort.

Methods:

A structural equation model (SEM) is applied with the latent constructs obesity, dietary intakes, physical activity and fitness habits, and parental SES to estimate the main effects of the latter three variables and a FTO polymorphism on childhood obesity. Further, a multiple group SEM is used to explore whether an interaction effect exists between the single nucleotide polymorphism rs9939609 within the FTO gene and SES.

Results:

Significant main effects are shown for physical activity and fitness (standardised β̂s = −0.113), SES (β̂s = −0.057) and the FTO homozygous AA risk genotype (β̂s = −0.177). The explained variance of obesity is ~9%. According to the multiple group approach of SEM, we see an interaction between SES and FTO with respect to their effect on childhood obesity (Δχ2=7.3, df=2, P=0.03).

Conclusion:

Children carrying the protective FTO genotype TT seem to be more protected by a favourable social environment regarding the development of obesity than children carrying the AT or AA genotype.

Introduction

Childhood obesity is a complex disorder where lifestyle factors, socioeconomic status (SES) and genetic factors play an important—and interconnected—role. The steep increase of the obesity epidemic in the past two decades might be largely caused by changes in the living environment that promotes both: excessive food intake and sedentary lifestyles.1 Such changes may have an impact on the effect of the genetic predisposition of an individual since genetic factors within a given environment do not only influence an individual’s body weight and body composition, but also the susceptibility to unhealthy behaviours. The investigation of interactions between genes and social environment may hence help to find answers to public health questions as whether individuals with a specific genetic makeup are more susceptible to a particular social environment and hence more influenceable by a prevention strategy or therapeutic interventions.

Some authors have reported a more pronounced influence of genetic factors on psychological characteristics or behaviour in socioeconomically privileged rather than in a socioeconomically disadvantaged environment.2,3 Disadvantaged social groups may be more exposed to social and economic risk factors that might mask the genetic influence on certain phenotypes. Especially with respect to obesity, for instance, Pigeyre et al.4 revealed an interaction between a neuromedin B polymorphism and maternal education.

For sure, modelling obesity and its determinants is a highly complex task. There are many potentially influential determinants that have been reported to have an impact on obesity5; and many of them are interconnected. Moreover, not all of these determinants can be measured directly and are therefore considered as latent constructs. For instance, SES can only be assessed by measuring different facets such as income, occupation and education. Although obesity is commonly assessed by the body mass index (BMI (kgm−2)), other anthropometric measurements such as waist circumference, waist-to-height ratio or skinfold thickness are used in addition to assess obesity, in particular in children. In view of the assumed high correlations between these measurements, we propose here treating also obesity as latent construct.

In general, regression methods fail to capture the influence of a network of latent constructs on the development of obesity. Thus, we would like to exploit a statistical model that can mirror highly complex association structures and is able to model latent constructs. Here, the method of choice is a structural equation model (SEM) that combines a network of latent constructs with the measurements of their observed indicators. In addition, SEM allows for modelling the correlations between the determinants. The model is then assessed by a comparison of the observed variance–covariance structure with the one implied by the network structure.

The aim of the present study is to examine whether the single nucleotide polymorphism (SNP) rs9939609 in the fat mass and obesity-associated gene (FTO) interacts with the parental SES on obesity in children aged 2–9 years. For this purpose, we apply a multiple group approach (MGA) of SEM. This exploratory approach consists of multiple comparisons between several distinctive groups. In our case, we consider three groups based on the three genotypes in our database.

The paper is organised as follows: in the following section, we present the European study IDEFICS (Identification and prevention of Dietary- and lifestyle-induced health EFfects In Children and infantS) on childhood obesity on which our analysis is based. We then describe how the SEM is designed and introduce the MGA. Next, we present the results of our analysis in detail and critically discuss these results. The paper concludes with a summary of the main results and a brief sketch of implications for research and policy.

Subjects and methods

IDEFICS is a multi-centre population-based longitudinal study that explores health effects with focus on childhood overweight/obesity.6 A cohort of 16 228 children aged 2–9 years was enrolled in eight European countries (Belgium, Cyprus, Estonia, Germany, Hungary, Italy, Spain and Sweden) according to a standardised protocol to collect data on anthropometric and demographic characteristics, physical activity and fitness (PAF), and dietary intakes (DIET) among others. PAF was measured by accelerometry, self-administered questionnaires and a shuttle-run test. DIET was, among others, assessed using a 24-h dietary recall of one day. Genotyping of rs9939609 within the FTO gene (A<T, minor allele frequency=40.4%) was done in a subsample of 4500 children (for detailed information about genotyping see Lauria et al.7). The analysis sample was further reduced to 4406 children by exclusion of unsuccessfully genotyped participants.

Means (±s.d.) or proportions are calculated for baseline characteristics of the study population. An SEM8 is applied to investigate whether SNP rs9939609 interacts with SES on childhood obesity. SES, DIET, PAF and obesity are modelled as latent constructs with multiple causal indicators: for SES we consider the measured indicators ‘household income level’, ‘maximum parental level of education’ and ‘maximum parental level of occupational position’ where all country-specific answer categories of these indicators were recoded to international standardised classification systems to make them comparable across countries.9 The latent construct DIET is captured by the indicators ‘usual energy intake per day in kcal’, ‘usual intake per day of protein in gram’, ‘usual intake per day of fat in gram’ and ‘usual intake per day of water in gram’, which are corrected for within-person variability. The latent construct PAF is reflected by ‘percentage of time spent in moderate-to-vigorous activity’ (cut-points according to Evenson et al.10), ‘average activity counts per minute’ (both measured by accelerometers), ‘self-reported hours per week the child was physically active’ and ‘maximum oxygen uptake VO2max’ (estimated from a shuttle-run test). Obesity is modelled as latent construct involving ‘age- and sex-standardised BMI z-scores’,11,12 ‘waist-to-height ratio’, ‘percentage of body fat’ (calculated from a bioelectrical impedance analysis according to Tyrrell et al.13), and ‘subcutaneous skinfold thickness’ (sum of subscapular and triceps) as indicators. PAF, DIET, SES and FTO are modelled as determinants of obesity. Using linear regressions, the indicators of DIET and PAF are adjusted for age, sex and country and the indicators of obesity are adjusted for age and sex. The obtained residuals are then used as observed indicators for all further analyses. In addition, the endogenous latent construct obesity is adjusted for sex, age and country to allow for estimating the impact of FTO, SES, PAF and DIET on obesity.

The basic SEM is depicted in Figure 1: ovals reflect latent constructs, boxes reflect observed indicators; error and disturbance terms as well as the reference categories (that is, FTO-TT genotype and Germany) are not represented. Single-headed arrows between latent constructs symbolise postulated pathways and double-headed arrows symbolise covariances between latent constructs. A confirmatory factor analysis (CFA) is employed to examine the reliability and validity of the measurement model. This enables us to investigate whether the four latent constructs are well represented by the indicator variables. The subsequent SEM additionally includes the structural relations between the latent constructs.

Figure 1
figure1

Structural equation model estimating the main effects of SES, dietary intakes (DIET), physical activity and fitness (PAF) habits and the rs9939609 (FTO) polymorphism. Standardised parameter estimates are shown; bold lines indicate statistically significant parameter estimates and fixed factor loadings (α=0.05). Ovals reflect latent constructs, boxes reflect observed indicators; error and disturbance terms as well as the reference categories (i.e., FTO-TT genotype and Germany) are not represented. Arrows between latent constructs represent postulated pathways.

The basic SEM is fitted to estimate main effects on obesity. To test for an interaction between SES and FTO, the MGA is exploited. The objective of the MGA is to compare distinct sets of parameters for each genotype, some of them restricted by assuming that they are equal across genotypes. The MGA follows a step-up approach and compares different nested models. That is, each of these models is estimated under additional constraints (see Table 1) and then compared to the preceding model with respect to its model fit. This procedure continues as long as the models do not significantly differ. In the final step, the only unconstrained parameter reflects the path from SES to obesity. This parameter is then tested to be equal across the three FTO genotypes.14 If the effect of SES on obesity is not equal for all FTO genotypes, we conclude that there is an interaction between SES and FTO. We use robust weighted least square estimators to fit SEM and Δχ2 difference tests to conduct MGA.

Table 1 Description of models and model fit indices for the basic SEM, the basic multiple group model MG and multiple group models MG 1 through MG 4 using data from the IDEFICS study (N=4406)

Additional power calculations15 for this exploratory approach revealed that the sample size of our study is sufficient to detect a significant difference in the model fits for each of the Δχ2 difference tests conducted (see Table 1 for the considered degrees of freedom and Table 2 for the resulting power).

Table 2 Δχ2 difference tests for nested and constrained models; the power of each test has been calculated based on the degrees of freedom of the respective χ2 test given in Table 1, assuming α=0.05 and RMSEA values of 0.045 for both models under the alternative hypothesis that both models are not equal (N=4406)

We report the root mean square error of approximation (RMSEA) and the comparative fit index (CFI) as fit indices. RMSEA values <0.05 and CFI values >0.95 indicate good fit.16 We also report the χ2 statistic although we do not use it as fit index because of its drawback being sensitive to large sample sizes.17 Standardised parameter estimates are reported where standardised estimated structural regression parameters indicate changes in units of standard deviations. The residual variances for usual energy intake and for moderate-to-vigorous activity are fixed to 0.033 and 0.059, respectively, based on the results of two exploratory factor analyses to ensure that all estimated variances are positive.

Results

In the following, we first present basic descriptive results. Then, we give the overall model fits of the measurement model and of the basic SEM before we discuss the estimated basic SEM. Finally, we present the results of the MGA. For this purpose, we describe the results of the first model of this approach (model MG; see Table 1) which is estimated without any constraints on the model parameters in some more detail before we report the results of the Δχ2 difference test on interaction between FTO and SES.

Basic demographic characteristics are shown in Table 3. Mean age (±s.d.) is 6.0 years (1.8) in boys and 6.1 years (1.8) in girls. The most frequent FTO genotype is AT (47.3%), whereas the homozygous AA genotype only occurs in 16.7%. At least one parent has the highest educational level in 40.1% and the highest occupational level in 30.3%; however, the most frequent parental income level belongs to the medium category (28.2%).

Table 3 Demographic characteristics of 4406 children included in the analysis

The overall model fit of the CFA model is very good (RMSEA=0.036; CFI=0.96) which supports the postulated measurement model before introducing the path structure into the SEM.

Table 1 presents the model fit indices of all SEMs leading to an inconsistent assessment of the model fit (for example, basic SEM: RMSEA=0.05; CFI=0.79). According to RMSEA the fitted models are acceptable; CFI values indicate, however, poor model fits.

Table 4 shows estimated paths coefficients, covariances and variances for the basic SEM. Standardised main effects are also shown in Figure 1. Non-standardised estimates are interpreted as in ordinary least squares (OLS) regression and are used to compare equal paths between genotypes. Standardised estimates are interpreted in units of standard deviations and should be used to compare different paths within one genotype. The rs9939609 homozygous risk genotype AA is a statistically significant positive predictor for obesity (β̂ = 0.154 (P<0.001), standardised β̂s = 0.177). The standardised coefficient implies that the AA genotype increases obesity by 0.177 s.d. compared to the reference TT. PAF (β̂ = −0.101 (P<0.001), β̂s = 0.113) and SES (β̂ = 0.079 (P=0.002), β̂s = −0.057) have a statistically significant negative main effect on obesity. An increase of the latent construct PAF or SES by 1 s.d. decreases obesity by 0.113 and 0.057 s.d., respectively. There is no statistically significant main effect for DIET (β̂ = − 0.005 (P=0.758), β̂s = − 0.006). The only statistically significant association within the endogenous variables is between PAF and DIET (covariance: 0.056 (P=0.037)). The basic SEM shows country-specific differences of obesity. The total explained variance of obesity is ~9%.

Table 4 Parameter estimates for the basic SEM using IDEFICS data (N=4406)

The multiple group model with complete heterogeneity (model MG) reveals that the path from PAF to obesity is statistically significant in all three genotype groups (TT: β̂ = − 0.109 (P<0.001), AT: β̂ = −0.103 (P<0.001), AA: β̂ = −0.086 (P=0.050)). SES is a statistically significant inverse predictor of obesity for the TT genotype (TT: β̂ = −0.170 (P<0.001)), but not statistically significant for the other genotypes (AT: β̂ = −0.030 (P=0.424), AA: β̂ = −0.051 (P=0.354); results are not shown).

The MGA reveals significant differences between the regression parameters for the latent constructs of the rs9939609 genotype groups (Table 2). In the first three steps, factor loadings, intercepts, thresholds for categorical variables and path coefficients besides the path between SES and obesity are shown to be equal across all three FTO genotypes. In the last step, the Δχ2 difference test yields a statistically significant difference between the model fits of the model with freely estimated regression parameters for SES on obesity (MG 3) in comparison with the model MG 4 where all regression parameters are assumed to be equal across genotypes (Δχ2=7.3, df=2, P=0.03).

Discussion

Owing to our results, parental SES may interact with the polymorphism rs9939609 (FTO) in its influence on childhood obesity. The results of the MGA implied that the advantage of favourable socioeconomic conditions in which the child grows up is especially apparent for children carrying the protective FTO genotype TT. We found a strong positive association of the AA risk genotype with obesity scaled to units of BMI z-score. The basic SEM showed in addition that an increase of 1 s.d. of PAF reduces obesity around twice as much as a 1 s.d. increase on the parental SES scale. Furthermore, the model suggested that if carriers of the AA genotype increase their PAF by around 1.5 s.d. they may compensate for their genetic predisposition which can be seen from the following equation: β̂sAA + 1.5 · β̂sPAF = 0.177 − 1.5 · 0.113 ≈ 0. The only statistically significant association within the endogenous variables is between PAF and DIET. We were unable to detect an association between SES and both: DIET or PAF.

There are some other studies that investigated whether socioeconomic factors modify the effects of genetic variations on health outcomes18, 19, 20, 21, 22 but only a few examined the FTO gene in this context. Our finding is supported by Corella et al.23 who reported an interaction between education and FTO rs9939609 regarding their influence on BMI in adults. However, other investigators24 could not reveal an interaction between two other FTO polymorphisms with education and income on BMI in adults. Besides the interaction which is of interest, that is, between SES and FTO, interactions between the environmental variable physical activity and FTO single nucleotide polymorphisms are reported in the literature.25,26

FTO has been long considered ‘a gene of unknown function in an unknown pathway’27 that has frequently been associated with fat mass and predisposes to childhood and adult obesity.28, 29, 30, 31 Human FTO presents high homology with the murine Fto, located on mouse chromosome 8.32 In recent years, several papers shed light on its physiological role but a complete understanding of the ‘true cellular function of FTO remains a puzzle’, as reviewed by Larder et al.33 However, it is frequently reported that the FTO protein is expressed in multiple tissues with particularly high expression levels in the brain and the hypothalamus, which is a key location for regulation of energy balance and the regulation of appetite.27,34, 35, 36 According to Way and Lieberman,37 especially genes affecting brain function appear to influence adaptive behaviours and the degree to which a person is emotionally responsive under favourable or unfavourable social environments.

In accordance to other studies, the polymorphism rs9939609, SES and PAF have significant direct effects on obesity with inverse associations of SES and PAF.7,27,34,38, 39, 40 However, our results did not show any evidence for an association between DIET and obesity. A possible explanation may be that the diet indicators used for the present analysis only capture one dimension of diet and may miss important information. Moreover, misreporting and measurement errors, that are a special problem when measuring diet in children, may affect or even mask associations between diet and obesity.40

We applied SEMs because they allow to model and test a complex association network incorporating several indicator variables simultaneously. Thus, it is possible to assess significance and importance of relationships in the context of the overall structure, even between predictor variables, which might lead to more valid conclusions than several single regression analyses. However, contradictory overall model fit values indicate a weakness of our results. While RMSEA attests a good fit, CFI values lower 0.95 might indicate that the postulated network does not match the true structure or that the correlations between the selected indicators are too weak. This inconsistent assessment of the model fit may also be due to reverse causation which cannot completely be ruled out because of the cross-sectional study design which does not allow conclusions on causal associations.

Strengths of our study are, among others, the heterogeneity of the study sample reflecting various European cultures, the highly standardised examination programme, and the objective assessment of lifestyle factors.

In summary, our study suggested an interaction between the FTO polymorphism rs9939609 and SES on childhood obesity, which reflects the sensitivity of the FTO gene to the social environment. More insights into the biology of FTO are needed to understand if and how it regulates gene expression under different socioeconomic conditions. Despite this limitation, our analysis showed that an individual genetic susceptibility to obesity could be compensated by adopting and maintaining a lifestyle in families that supports physical active behaviour. However, as long as the complex underlying biology of FTO is not yet understood, the interpretation of the moderating effect is speculative. Further research in how genes and social environment can moderate each other will be needed to fully understand this complex relationship and to use it as a robust evidence-base for health policy. This holds particularly true against the current intense discussion of how patterns of fat metabolism that are influenced by genetic architecture represent aSn adaptive response to psychosocial environment.41

References

  1. 1

    Hill JO, Peters JC . Environmental contributions to the obesity epidemic. Science 1998; 280: 1371–1374.

    CAS  Article  Google Scholar 

  2. 2

    Turkheimer E, Haley A, Waldron M, d’Onofrio B, Gottesman II . Socioeconomic status modifies heritability of IQ in young children. Psychol Sci 2003; 14: 623–628.

    Article  Google Scholar 

  3. 3

    Tuvblad C, Grann M, Lichtenstein P . Heritability for adolescent antisocial behavior differs with socioeconomic status: gene–environment interaction. J Child Psychol Psychiatry 2006; 47: 734–743.

    Article  Google Scholar 

  4. 4

    Pigeyre M, Bokor S, Romon M, Gottrand F, Gilbert C, Valtuena J et al. Influence of maternal educational level on the association between the rs3809508 neuromedin B gene polymorphism and the risk of obesity in the HELENA study. Int J Obes 2010; 34: 478–486.

    CAS  Article  Google Scholar 

  5. 5

    Moreno LA, Pigeot I, Ahrens W . Epidemiology of Obesity in Children and Adolescents – Prevalence and Etiology, chapter 26. Springer, 2011, pp 483–492.

    Google Scholar 

  6. 6

    Ahrens W, Bammann K, Siani A, Buchecker K, De Henauw S, Iacoviello L et al. IDEFICS Consortium. The IDEFICS cohort: design, characteristics and participation in the baseline survey. Int J Obes 2011; 35 (Suppl 1): S3–S15.

    Article  Google Scholar 

  7. 7

    Lauria F, Siani A, Bammann K, Foraita R, Huybrechts I, Iacoviello L et al. IDEFICS Consortium. Prospective analysis of the association of a common variant of FTO (rs9939609) with adiposity in children: results of the IDEFICS study. PLoS One 2012; 7: e48876.

    CAS  Article  Google Scholar 

  8. 8

    Bollen KA . Structural equation models with latent variables. John Wiley & Sons: New York, 1989.

    Google Scholar 

  9. 9

    Bammann K, Gwozdz W, Lanfer A, Barba G, De Henauw S, Eiben G et al. IDEFICS Consortium Socioeconomic factors and childhood overweight in Europe: results from the multi-centre IDEFICS study. Pediatr Obes 2013; 8: 1–12.

    CAS  Article  Google Scholar 

  10. 10

    Evenson KR, Catellier DJ, Gill K, Ondrak KS, McMurray RG . Calibration of two objective measures of physical activity for children. J Sports Sci 2008; 26: 1557–1565.

    Article  Google Scholar 

  11. 11

    Cole TJ, Freeman JV, Preece MA . Body mass index reference curves for the UK, 1990. Arch Dis Child 1995; 73: 25–29.

    CAS  Article  Google Scholar 

  12. 12

    Cole TJ, Freeman JV, Preece MA . British 1990 growth reference centiles for weight, height, body mass index and head circumference fitted by maximum penalized likelihood. Stat Med 1998; 17: 407–429.

    CAS  Article  Google Scholar 

  13. 13

    Tyrrell V, Richards G, Hofman P, Gillies G, Robinson E, Cutfield W . Foot-to-foot bioelectrical impedance analysis: a valuable tool for the measurement of body composition in children. Int J Obes Relat Metab Disord 2001; 25: 273–278.

    CAS  Article  Google Scholar 

  14. 14

    Brown TA . Confirmatory factor analysis. Guilford: New York, 2006.

    Google Scholar 

  15. 15

    MacCallum RC, Browne MW, Cai L . Testing differences between nested covariance structure models: power analysis and null hypotheses. Psychol Methods 2006; 11: 19–35.

    Article  Google Scholar 

  16. 16

    Hu Lt Bentler PM . Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Struct Equ Modeling 1999; 6: 1–55.

    Article  Google Scholar 

  17. 17

    Fan X, Thompson B, Wang L . Effects of sample size, estimation methods, and model specification on structural equation modeling fit indexes. Struct Equ Modeling 1999; 6: 56–83.

    Article  Google Scholar 

  18. 18

    Silventoinen K, Sarlio-Lähteenkorva S, Koskenvuo M, Lahelma E, Kaprio J . Effect of environmental and genetic factors on education-associated disparities in weight and weight gain: a study of Finnish adult twins. Am J Clin Nutr 2004; 80: 815–822.

    CAS  Article  Google Scholar 

  19. 19

    Miller G, Chen E . Unfavorable socioeconomic conditions in early life presage expression of proinflammatory phenotype in adolescence. Psychosom Med 2007; 69: 402–409.

    CAS  Article  Google Scholar 

  20. 20

    Cole SW, Arevalo JM, Takahashi R, Sloan EK, Lutgendorf SK, Sood AK et al. Computational identification of gene–social environment interaction at the human IL6 locus. Proc Natl Acad Sci 2010; 107: 5681–5686.

    CAS  Article  Google Scholar 

  21. 21

    Johnson W, Kyvik KO, Mortensen EL, Skytthe A, Batty GD, Deary IJ . Education reduces the effects of genetic susceptibilities to poor physical health. Int J Epidemiol 2010; 39: 406–414.

    Article  Google Scholar 

  22. 22

    Lagou V, Liu G, Zhu H, Stallmann-Jorgensen IS, Gutin B, Dong Y et al. Lifestyle and socioeconomic-status modify the effects of ADRB2 and NOS3 on adiposity in European-American and African-American adolescents. Obesity 2011; 19: 595–603.

    CAS  Article  Google Scholar 

  23. 23

    Corella D, Carrasco P, Sorl JV, Coltell O, Ortega-Azorn C, Guillen M et al. Éducation modulates the association of the FTO rs9939609 polymorphism with body mass index and obesity risk in the Mediterranean population. Nutr Metab Cardiovasc Dis 2012; 22: 651–658.

    CAS  Article  Google Scholar 

  24. 24

    Holzapfel C, Grallert H, Baumert J, Thorand B, Döring A, Wichmann HE et al. First investigation of two obesity-related loci (TMEM18, FTO) concerning their association with educational level as well as income: the MONICA/KORA study. J Epidemiol Community Health 2011; 65: 174–176.

    Article  Google Scholar 

  25. 25

    Andreasen CH, Stender-Petersen KL, Mogensen MS, Torekov SS, Wegner L, Andersen G et al. Low physical activity accentuates the effect of the FTO rs9939609 polymorphism on body fat accumulation. Diabetes 2008; 57: 95–101.

    CAS  Article  Google Scholar 

  26. 26

    Vimaleswaran KS, Li S, Zhao JH, Luan J, Bingham SA, Khaw KT et al. Physical activity attenuates the body mass index–increasing influence of genetic variation in the FTO gene. Am J Clin Nutr 2009; 90: 425–428.

    CAS  Article  Google Scholar 

  27. 27

    Frayling TM, Timpson NJ, Weedon MN, Zeggini E, Freathy RM, Lindgren CM et al. A common variant in the FTO gene is associated with body mass index and predisposes to childhood and adult obesity. Science 2007; 316: 889–894.

    CAS  Article  Google Scholar 

  28. 28

    Scuteri A, Sanna S, Chen WM, Uda M, Albai G, Strait J et al. Genome-wide association scan shows genetic variants in the FTO gene are associated with obesity-related traits. PLoS Genet 2007; 3: e115.

    Article  Google Scholar 

  29. 29

    Cecil JE, Tavendale R, Watt P, Hetherington MM, Palmer CN . An obesity-associated FTO gene variant and increased energy intake in children. N Engl J Med 2008; 359: 2558–2566.

    CAS  Article  Google Scholar 

  30. 30

    Haupt A, Thamer C, Machann J, Kirchhoff K, Stefan N, Tschritter O et al. Impact of variation in the FTO gene on whole body fat distribution, ectopic fat, and weight loss. Obesity 2008; 16: 1969–1972.

    Article  Google Scholar 

  31. 31

    Bauer F, Elbers CC, Adan RA, Loos RJ, Onland-Moret NC, Grobbee DE et al. Obesity genes identified in genome-wide association studies are associated with adiposity measures and potentially with nutrient-specific food preference. Am J Clin Nutr 2009; 90: 951–959.

    CAS  Article  Google Scholar 

  32. 32

    Peters T, Ausmeier K, Dildrop R, Rüther U . The mouse Fused toes (Ft) mutation is the result of a 1.6-Mb deletion including the entire Iroquois B gene cluster. Mamm Genome 2002; 13: 186–188.

    CAS  Article  Google Scholar 

  33. 33

    Larder R, Cheung M, Tung Y, Yeo GS, Coll AP . Where to go with FTO? Trends Endocrinol Metab 2011; 22: 53–59.

    CAS  Article  Google Scholar 

  34. 34

    Dina C, Meyre D, Gallina S, Durand E, Körner A, Jacobson P et al. Variation in FTO contributes to childhood obesity and severe adult obesity. Nat Genet 2007; 39: 724–726.

    CAS  Article  Google Scholar 

  35. 35

    Fredriksson R, Hägglund M, Olszewski PK, Stephansson O, Jacobsson JA, Olszewska AM et al. The obesity gene, FTO, is of ancient origin, up-regulated during food deprivation and expressed in neurons of feeding-related nuclei of the brain. Endocrinology 2008; 149: 2062–2071.

    CAS  Article  Google Scholar 

  36. 36

    Tung YCL, Yeo GS . From GWAS to biology: lessons from FTO. Ann NY Acad Sci 2011; 1220: 162–171.

    Article  Google Scholar 

  37. 37

    Way BM, Lieberman MD . Is there a genetic contribution to cultural differences? Collectivism, individualism and genetic markers of social sensitivity. Soc Cogn Affect Neurosci 2010; 5: 203–211.

    Article  Google Scholar 

  38. 38

    Wang Y, Lim H . The global childhood obesity epidemic and the association between socio-economic status and childhood obesity. Int Rev Psychiatry 2012; 24: 176–188.

    Article  Google Scholar 

  39. 39

    Crespo CJ, Smit E, Troiano RP, Bartlett SJ, Macera CA, Andersen RE . Television watching, energy intake, and obesity in US children: results from the third National Health and Nutrition Examination Survey, 1988-1994. Arch Pediatr Adolesc Med 2001; 155: 360–365.

    CAS  Article  Google Scholar 

  40. 40

    Börnhorst C, Huybrechts I, Hebestreit A, Vanaelst B, Molnár D, Bel-Serrat S et al. IDEFICS Consortium Diet–obesity associations in children: approaches to counteract attenuation caused by misreporting. Public Health Nutr 2012; 16: 256–266.

    Article  Google Scholar 

  41. 41

    Wells JC . The evolution of human adiposity and obesity: where did it all go wrong? Dis Model Mech 2012; 5: 595–607.

    Article  Google Scholar 

Download references

Acknowledgements

This work was done as part of the IDEFICS Study (www.idefics.eu). We gratefully acknowledge the financial support of the European Community within the Sixth RTD Framework Program Contract No. 016181 (FOOD). We thank the IDEFICS children and their parents for taking the time to participate in this extensive examination programme. We are grateful for the support provided by school boards, headmasters, teachers, school staff and communities, and for the effort of all study nurses and our data managers, especially Claudia Brünings-Kuppe and Birgit Reineke.

Author information

Affiliations

Authors

Consortia

Corresponding author

Correspondence to R Foraita.

Ethics declarations

Competing interests

The authors declare no conflict of interest.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Foraita, R., Günther, F., Gwozdz, W. et al. Does the FTO gene interact with the socioeconomic status on the obesity development among young European children? Results from the IDEFICS study. Int J Obes 39, 1–6 (2015). https://doi.org/10.1038/ijo.2014.156

Download citation

Further reading

Search