Role of BMI-Associated Loci Identified in GWAS Meta-Analyses in the Context of Common Childhood Obesity in European Americans

Authors

  • Jianhua Zhao,

    1. Division of Human Genetics, The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, USA
    Search for more papers by this author
  • Jonathan P. Bradfield,

    1. Center for Applied Genomics, Abramson Research Center, The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, USA
    Search for more papers by this author
  • Haitao Zhang,

    1. Center for Applied Genomics, Abramson Research Center, The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, USA
    Search for more papers by this author
  • Patrick M. Sleiman,

    1. Center for Applied Genomics, Abramson Research Center, The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, USA
    Search for more papers by this author
  • Cecilia E. Kim,

    1. Center for Applied Genomics, Abramson Research Center, The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, USA
    Search for more papers by this author
  • Joseph T. Glessner,

    1. Center for Applied Genomics, Abramson Research Center, The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, USA
    Search for more papers by this author
  • Sandra Deliard,

    1. Division of Human Genetics, The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, USA
    Search for more papers by this author
  • Kelly A. Thomas,

    1. Center for Applied Genomics, Abramson Research Center, The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, USA
    Search for more papers by this author
  • Edward C. Frackelton,

    1. Center for Applied Genomics, Abramson Research Center, The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, USA
    Search for more papers by this author
  • Mingyao Li,

    1. Department of Biostatistics and Epidemiology, University of Pennsylvania, Philadelphia, Pennsylvania, USA
    Search for more papers by this author
  • Rosetta M. Chiavacci,

    1. Center for Applied Genomics, Abramson Research Center, The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, USA
    Search for more papers by this author
  • Robert I. Berkowitz,

    1. Behavioral Health Center and Department of Child and Adolescent Psychiatry, The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, USA
    2. Center for Weight and Eating Disorders, Department of Psychiatry, University of Pennsylvania, Philadelphia, Pennsylvania, USA
    3. Department of Pediatrics, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, USA
    Search for more papers by this author
  • Hakon Hakonarson,

    1. Division of Human Genetics, The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, USA
    2. Center for Applied Genomics, Abramson Research Center, The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, USA
    3. Department of Pediatrics, University of Pennsylvania, Philadelphia, Pennsylvania, USA
    Search for more papers by this author
  • Struan F.A. Grant

    Corresponding author
    1. Division of Human Genetics, The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, USA
    2. Center for Applied Genomics, Abramson Research Center, The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, USA
    3. Department of Pediatrics, University of Pennsylvania, Philadelphia, Pennsylvania, USA
    Search for more papers by this author

  • The first two authors contributed equally to the work.

(grants@chop.edu)

Abstract

Obesity is a serious health concern for children and adolescents, particularly in Western societies, where its incidence is now considered to have reached epidemic proportions. A number of genetic determinants of adult BMI have already been established through genome wide association studies (GWAS), most recently from the GIANT meta-analysis of such datasets combined. In this current study of European Americans, we examined the 32 loci detected in that GIANT study in the context of common childhood obesity within a cohort of 1,097 cases (defined as BMI ≥95th percentile), together with 2,760 lean controls (defined as BMI <50th percentile), aged between 2 and 18 years old. Nine of these single-nucleotide polymorphims (SNPs) yielded at least nominal evidence for association with common childhood obesity, namely at the FTO, TMEM18, NRXN3, MC4R, SEC16B, GNPDA2, TNNI3K, QPCTL, and BDNF loci. However, overall 28 of the 32 loci showed directionally consistent effects to that of the adult BMI meta-analysis. We conclude that among the 32 loci that have been reported to associate with adult BMI in the largest meta-analysis of BMI to date, at least nine also contribute to the determination of common obesity in childhood in European Americans, as demonstrated by their associations in our pediatric cohort.

Genome wide association studies (GWAS) have been successful in uncovering loci that are robustly associated with many complex traits. BMI is one such phenotype, with increasing numbers of variants being revealed with increasing sample size (1,2,3,4,5,6). Ultimately, researchers have combined their datasets to carry out meta-analyses, in order to elucidate additional genetic contributors to a trait. These latest signals yield relatively modest strengths of association compared to the initial “low-hanging fruit” variants in genes such as FTO (2).

Building on previous discoveries, 22 additional loci were reported by the GIANT consortium in the past year to be associated with BMI, primarily in adults, from a GWAS meta-analysis of 249,796 individuals (3). A total of 32 loci actually reached genome wide significance but ten were already known, namely FTO, TMEM18, MC4R, GNPDA2, BDNF, NEGR1, SH2B1, ETV5, MTCH2, and KCTD15. Of the novel loci, four had been previously reported to be associated with either weight and waist-hip ratio, namely SEC16B, TFAP2B, FAIM2, and NRXN3, while 18 had never been implicated in a GWAS of BMI-related traits previously, consisting of RBJ-ADCY3-POMC, GPRC5B-IQCK, MAP2K5-LBXCOR1, QPCTL-GIPR, TNNI3K, SLC39A8, FLJ35779-HMGCR, LRRN6C, TMEM160-ZC3H4, FANCL, CADM2, PRKD1, LRP1B, PTBP2, MTIF3-GTF3A, ZNF608, RPL27A-TUB, and NUDT3-HMGA1. The same study also tested these variants in the context of extreme obesity in children, where 30 of the 32 loci also showed directionally consistent effects to that of adult BMI.

It is important to determine which variants previously uncovered in GWAS analyses of adult traits are exerting an effect early on in life to shed light on mechanisms of action. In this study, we aimed at examining the BMI loci reported from the GIANT meta-analysis in the context of a large cohort of subjects presenting with common childhood obesity, where we specifically excluded extreme cases in order to determine the relative impact of these variants in the pathogenesis of this pediatric trait. To achieve this goal, we leveraged genotyping data generated on the Human Hap550 BeadChip (Illumina, San Diego, CA) in our pediatric cohort.

Our case-control design consisted of 1,097 European American children with common obesity (defined as BMI ≥95th percentile) and 2,760 lean controls (defined as BMI <50th percentile) of the same ethnicity (based on principal components analysis). The age-range was restricted to subjects between 2 and 18 years of age due to the age-range limit of the BMI reference range. All individuals were drawn from a conservatively defined cohort, where outliers were excluded to avoid the consequences of potential measurement error or Mendelian causes of extreme obesity.

We examined the 32 single nucleotide polymorphisms (SNPs) corresponding to the loci uncovered in the GIANT BMI meta-analysis in the context of common childhood obesity. Nine of these SNPs yielded at least nominal evidence for association to childhood obesity (P < 0.05) (Table 1).

Table 1.  Case-control analysis results for all 32 BMI-associated loci (1,097 cases and 2,760 controls) reported in the GIANT meta-analysis in the context of childhood obesity, sorted by chromosome
inline image

Variation at the loci harboring FTO, TMEM18, NRXN3, and MC4R yielded the strongest associations and survived correction for the number of tests applied, namely rs1558902 (P = 1.42 × 10−8; odds ratio (OR) = 1.34), rs2867125 (P = 2.03 × 10−4; OR = 0.78), rs10150332 (P = 7.97 × 10−4; OR = 1.23), and rs571312 (P = 0.001; OR = 1.21), respectively.

Next, with a lower magnitude of association, came SEC16B, GNPDA2, and TNNI3K, with rs543874 yielding an OR = 1.21 (P = 0.0038), rs10938397 yielding an OR = 1.14 (P = 0.0097), and rs1514175 yielding an OR = 1.14 (P = 0.011), respectively. Lastly was the QPCTL and BDNF loci, with rs2287019 yielding an OR = 1.17 (P = 0.015) and rs10767664 yielding an OR = 0.87 (P = 0.03), respectively.

SNPs residing at the RBJ, SDCCAG8, ZNF608, MTIF3, KCTD15, TFAP2B, FAIM2, NPC1, TNKS/MSRA, PTBP2, LRP1B, NUDT3, PTER, CADM2, RPL27A, SH2B1, SLC39A8, TMEM160, MTCH2, MAF, MAP2K5, FANCL, GPRC5B, FLJ35779, LRRN6C, PRKD1, NEGR1, and ETV5 loci did not reveal statistically significant evidence of association in our childhood obesity cohort. However, overall 28 of the 32 loci yielded directionally consistent effects to that of the adult BMI meta-analysis (Table 1), with only TMEM160 being inconsistent with the extreme childhood obesity investigation in the same study. In addition, there were no significant differences between genders (Supplementary Table S1 online).

Two loci were previously described in a French/German GWAS of extreme childhood obesity (7) and were specifically tested in the GIANT meta-analysis (3). Although not statistically significant, we observed evidence for consistency of direction of association with variants at the TNKS/MSRA and SDCCAG8 loci. Similarly, we see comparable directions of effect for variants at two of the three loci previously reported to be associated with early-onset obesity (8), namely PTER, and MAF but not NPC1.

From this analysis of genotype data generated in our cohort of common pediatric obesity in European Americans, it is clear that a number of loci detected in the GIANT meta-analysis of adult BMI also play a role in our phenotype of interest.

The neuronal control on various aspect of energy consumption and energy balance, in particular, the hypothalamus has been well recognized (9); indeed the knockout mouse for FTO supports this fact (10). NRXN3, the most strongly associated novel locus of these reported regions (i.e., the GIANT meta-analysis was the first to implicate it in BMI) in our cohort, encodes a cell adhesion molecule and receptor which is highly expressed in the central nervous system (11). Prior studies on NRNX3 function have suggested it plays an important role in alcohol dependence, cocaine addiction, and illegal substance abuse (12,13). The association of NRNX3 with BMI has lead to suggestions that obesity results partly as a consequence of food addiction controlled by the central nervous system (11). The association of this locus in our pediatric cohort, further refines the association of NRNX3 locus with respect to obesity-related traits and suggests that it exerts its influence in early life. Indeed, the same conclusion of an early age effect can be drawn for the loci harboring FTO, TMEM18, NRXN3, MC4R, SEC16B, GNPDA2, TNNI3K, QPCTL, and BDNF. Interestingly, NRNX3, GNPDA2, and QPCTL were not significantly associated with extreme obesity in children in the GIANT study (3).

We have followed up older GWAS reports for BMI in this same cohort (14), albeit when our collection was substantially smaller in scale. However, with relatively modest novel signals derived form the most recent meta-analysis, we elected to use a leaner control set in this instance to enhance the contrast between patients and controls in order to increase sensitivity for the newest loci in this pediatric setting.

For the loci that did not yield significant evidence for association could be as a consequence of power issues, but could also suggest that these particular loci play a smaller role in the pathogenesis of common childhood obesity.

We conclude that among 32 loci that have been reported to associate with adult BMI in a recent meta-analysis of GWAS data, at least nine also contribute to the pathogenesis of common childhood obesity in European Americans.

Methods and Procedures

Research subjects

Childhood BMI cohort from Philadelphia. All subjects were consecutively recruited from the Greater Philadelphia area from 2006 to 2010 at the Children's Hospital of Philadelphia. Our study cohort consisted of 1,097 obese children (BMI ≥95th percentile) and 2,760 controls (BMI <50th percentile). All of these participants had their blood drawn into a 6 ml EDTA blood collection tube and were subsequently DNA extracted for genotyping. All subjects were biologically unrelated and were aged between 2 and 18 years old due to the age-range limit of the BMI reference range. The recruitment was based on systematic recruitment across the hospital network, of which those with no underlying conditions were selected for this study. This study was approved by the Institutional Review Board of the Children's Hospital of Philadelphia. Parental informed consent was given for each study participant for both the blood collection and subsequent genotyping.

Genotyping

Illumina Infinium assay. We performed high throughput genome-wide SNP genotyping, using the Illumina Infinium™ II HumanHap550 BeadChip technology (Illumina), at the Center for Applied Genomics at Children's Hospital of Philadelphia, as described previously (15). The resources available for this project included the Illumina technology platform itself plus nine pipetting robotic systems (Tecan, San Jose, CA), eight scanners, a laboratory information management system, and automated allele-calling software. The workflow is robotic-based for automatic sample processing and includes algorithms for quality control of genotypes, including call rates above 98%.

Genotypes were obtained using high-density SNP arrays, and then imputed based on release 22 of the HapMap Project. All individuals and SNPs in this study passed our filter for quality control (genome wide 19,462 markers excluded based on Hardy Weinberg P ≤ = 1 × 10−6, 4,172 SNPs failed genotype missingness >0.05 and 20,136 SNPs had a minor allele frequency <0.01).

Analysis

Normalization of BMI data. From our database of heights and weights for our multidimensional scaling determined whites. As BMI values vary heavily across pediatric age groups, each BMI value was adjusted for age and sex then expressed as a z-score. Subjects with BMI z-scores greater than 3 or less than −3 were excluded as outliers to avoid the consequences of potential measurement error or Mendelian causes of extreme obesity.

Association. We queried the data for the indicated SNPs in our pediatric samples. We used principal components analysis (16) in order to minimize the potential impact of population stratification in our case/control sample sets. Eigenstrat 3.0 was employed to remove outliers and to subsequently calculate the principal components. The principal components were then used as covariates in a logistic regression, using the software mach2dat, to compute the P values, ORs, and standard errors.

SUPPLEMENTARY MATERIAL

Supplementary material is linked to the online version of the paper at http:www.nature.comoby

ACKNOWLEDGEMENT

We would like to thank all participating subjects and families. Elvira Dabaghyan, Hope Thomas, Kisha Harden, Andrew Hill, Kenya Fain, Crystal Johnson-Honesty, Cynthia Drummond, Shanell Harrison and Sarah Wildrick provided expert assistance with genotyping or data collection and management. We would also like to thank Smari Kristinsson, Larus Arni Hermannsson and Asbjörn Krisbjörnsson of Raförninn ehf for their extensive software design and contribution. This research was financially supported by the Children's Hospital of Philadelphia. We want to thank the network of primary care clinicians, their patients and families for their contribution to this project and clinical research facilitated through the Pediatric Research Consortium (PeRC) at The Children's Hospital of Philadelphia. The study is supported by an Institute Development Award from The Children's Hospital of Philadelphia, a Research Development Award from the Cotswold Foundation and NIH grant 1R01HD056465-01A1.

DISCLOSURE

The authors declared no conflict of interest.

Ancillary