Jump to ContentJump to Main Navigation
Obesity Epidemiology$

Frank Hu

Print publication date: 2008

Print ISBN-13: 9780195312911

Published to Oxford Scholarship Online: September 2009

DOI: 10.1093/acprof:oso/9780195312911.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2019. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.  Subscriber: null; date: 17 September 2019

Genetic Predictors of Obesity

Genetic Predictors of Obesity

Chapter:
(p.437) 21 Genetic Predictors of Obesity
Source:
Obesity Epidemiology
Author(s):

Frank B. Hu

Publisher:
Oxford University Press
DOI:10.1093/acprof:oso/9780195312911.003.0021

Abstract and Keywords

This chapter begins with a review of the genetic factors underlying monogenic and syndromic forms of obesity. It describes the genetics of common obesity, with a particular focus on results from genome-wide linkage and candidate gene association studies. It also discusses recent findings using the genome-wide association (GWA) approach. Finally, several methodological problems that commonly plague genetic association studies, especially the inability to replicate findings, are addressed.

Keywords:   obesity research, genetic factors, genetic association studies, genome-wide association

The search for human obesity genes began several decades ago, but efforts have intensified in recent years with the completion of the Human Genome Project and advances in molecular biology, genotyping technology, and genetic epidemiologic methods. Several genetic factors responsible for rare monogenic forms of obesity have been identified; however, genes for common forms of obesity remain largely elusive. Nonetheless, there is hope that rapid advances in genomics technology and genetic association studies of complex diseases will provide new tools and an impetus for progress in the identification of susceptibility genes for common forms of obesity.

This chapter begins with a review of the genetic factors underlying monogenic and syndromic forms of obesity. We then describe the genetics of common obesity, with a particular focus on results from genome-wide linkage and candidate gene association studies. We also discuss recent findings using the genome-wide association (GWA) approach. Finally, several methodological problems that commonly plague genetic association studies, especially the inability to replicate findings, are addressed. A detailed description of the physiological basis of weight regulation is beyond the scope of this chapter, but would nonetheless facilitate a better understanding of the role that genetics play in the development of obesity. Information on this topic can be found in other excellent reviews.1- 3

Genetics of Monogenic Forms of Obesity

Over the past several decades, animal models, human linkage studies, and detailed genotyping and phenotyping of severely obese patients have greatly enhanced understanding of single mutations that contribute to the development of monogenic obesity.4- 8 These rare forms of severe obesity, typically beginning in childhood, result from spontaneous mutations in single genes, and display a Mendelian pattern of inheritance. Several genetic mutations responsible for monogenic obesity have been identified, many of which alter the leptin and melanocortin pathways (Fig. 21.1).5

Leptin is an adipocyte-secreted hormone transported across the blood-brain barrier to bind receptors that transmit satiety signals to the hypothalamic centers, where a complex network of neuropeptides regulate long-term energy homeostasis and weight control.9 Leptin signals through catabolic and anabolic pathways, each consisting of distinct classes (p.438)

Figure 21.1 Leptin and melanocortin pathways. Lep-R, leptin receptor; POMC, proopiomelanocortin; α-MSH, α-melanocyte-stimulating hormone; AGRP, agouti-related protein; MC4R, melanocortin-4 receptor; PC1, proconvertase 1; →, location of mutations responsible for monogenic obesity in man; →, AGRP is a natural antagonist of MC4R; +, pathway activated; −, pathway inhibited. Reproduced with permission from Clement K. Genetics of human obesity. Proc Nutr Soc. 2005;64:133–142.5

of neurons.2 The catabolic pathway includes the anorexigenic peptides proopiomelanocortin (POMC) and cocaine- and amphetamine-related transcript (CART), which reduce appetite and food intake. Increased leptin secretion stimulates the production of POMC, which is converted to α-melanocortin-stimulating hormone (α-MSH) through proconvertase 1 (PC1). The actions of melanocortins are mediated by a family of melanocortin receptors. The melanocrotin 4 receptor (MC4R) is largely expressed in the brain and central nervous system; activation of MC4R inhibits appetite and increases energy expenditure. The anabolic pathway includes neuropeptide-Y (NPY) and agouti-related protein (AGRP). Activation of NPY/AGRP neurons promotes positive energy balance by increasing appetite and food intake and decreasing energy expenditure. Reduced leptin secretion activates NPY/AGRP signaling and reduces MC4R signaling, thus stimulating food intake and promoting weight gain. Ghrelin, a gastrointestinal peptide hormone produced mainly by the stomach, opposes the action of leptin through disinhibition of NPY/AGRP, thereby stimulating short-term food intake and decreasing energy expenditure.10 Rare genetic mutations on the leptin and melanocortin pathways can disrupt both production and function of catabolic and anabolic neuropeptidies, leading to severe early-onset obesity and a variety of neuroendocrine abnormalities. In the following sections, we briefly review the major genetic mutations within these pathways that contribute to monogenic obesity (Table 21.1). For further information, readers can refer to several comprehensive reviews.4- 7

LEP Gene Mutations

Leptin, a hormone produced by adipose tissue and the product of the obese (ob) gene, plays a key role in regulating food intake and energy homeostasis. Ob/ob mice with a (p.439)

Table 21.1 Mutations in Human Obesity Affecting the Leptin and the Melanocortin Pathways

Gene

Transmission

Obesity

Associated Phenotypes

Leptin (LEP)

Recessive

Severe, from the first days of life

Leptin deficiency, gonadotropic, thyrotropic, insufficiency

Leptin receptor (LEPR)

Recessive

Severe, from the first days of life

Gonadotropic, thyrotropic and somatotropic insufficiency, high leptin

Proopiomelanocortin (POMC)

Recessive

Severe, from the first month of life

Adrenocorticotropin insufficiency, mild hypothyroidism, ginger hairs

Proconvertase 1 (PC1)

Recessive

Severe, from the first month of life

Gonadotropic and corticotropic insufficiency, hyperinsulinemia, other dysfunctions of gut peptides

Melanocortin-4 receptor (MC4R)

Dominant

Early onset, variable severity, large size

No other phenotype

Adapted from Clement K. Genetics of human obesity. C R Biol. 2006;329:608–622.6

homozygous LEP gene mutation exhibit complete leptin deficiency and early-onset severe obesity and diabetes.9 In 1997, Montague et al.11 reported on two leptin-deficient children from a consanguineous family of Pakistani origin who presented with early-onset severe obesity and hyperphagia. Both patients were homozygous for a single-nucleotide deletion at position 398 of the LEP gene, resulting in a frameshift of the leptin-coding region after Gly132 and a premature termination of peptide synthesis. Consistent with an autosomal recessive inheritance of the disorder, other family members, who were heterozygous for this mutation, were not severely obese. In a subsequent study, daily subcutaneous injections of recombinant human leptin for up to 4 years dramatically reduced body weight in three morbidly obese children with congenital leptin deficiency.12

LEPR Gene Mutations

Leptin receptors exist in several isoforms and play a part in modulating the availability and biological function of leptin.13 , 14 Mutations in LEPR in db/db mice produce the same phenotype as in ob/ob mice. Rather than displaying leptin deficiency, however, db/db mice are characterized by leptin resistance. A mutation in the human LEPR was first reported by Clement et al.15 in three morbidly obese sisters (13 to 19 years of age) from a consanguineous family of Algerian origin. The patients were homozygous for a single nucleotide substitution at a splice site in exon 16 of the LEPR gene, resulting in leptin receptor deficiency, and consequently, elevated serum leptin levels. They developed hyperphagia and severe obesity within a few months of birth. In that heterozygous parents and siblings of these affected patients were not severely obese, the disorder was characterized as an autosomal recessive trait. In a recent study, Farooqi et al.8 reported that pathogenic LEPR mutations were present in up to 3% of individuals with severe early-onset obesity. They identified five nonsense and four missense mutations in eight probands. All of the missense mutations were shown to either impair or completely prevent leptin receptor signaling. Interestingly, serum leptin levels in these patients were not substantially elevated.

(p.440) PC1 Gene Mutations

Proconvertases are required for processing of POMC into its constituent peptides. Loss-of-function mutations in PC1 gene have been shown to cause obesity.3 Congenital deficiency of PC1 was first described by Jackson et al.16 in a case report of a middle-aged woman presenting with severe, early-onset of obesity, impaired glucose tolerance, hypogonadism, hypoadrenalism, and reactive hypoglycemia. This woman was compound heterozygous for two mutations in the PC1 gene: a Gly483Arg missense mutation in PC1 resulting from a G→A substitution in exon 13 and an A→C substitution in the intron 5 donor splice site, which led to reduced production of functional PC1. The affected woman’s four children were heterozygous for one of the two mutations but of normal weight. Subsequently, Jackson et al.17 reported a second case of human PC1 deficiency, also due to compound heterozygosity for the two mutations. That patient shared the obesity phenotype with the first one, but also suffered from severe small-intestinal absorptive dysfunction.

POMC Gene Mutations

POMC-derived peptides play a critical role in regulating energy homeostasis and body weight through their actions at melanocortin receptors in the hypothalamus.5 Krude et al.18 described mutations in the POMC gene in a 5-year-old girl and a 5-year-old boy from unrelated families who developed early-onset obesity with hyperphagia. Both had pale skin, red hair, and adrenocorticotropin (ACTH) deficiency during infancy. The girl was compound heterozygous for two mutations (G7013T, C7133delta) in exon 3 of POMC, which resulted in loss of ACTH and α-MSH production. The boy was homozygous for a C3804A substitution in the 5′ untranslated region of POMC that abolished POMC translation. Additional loss-of-function mutations in the POMC gene have been identified in other severely obese children presenting with POMC deficiency.19 In POMC deficiency, obesity results from insufficient POMC-derived hormones and neuropeptides, ligands for the melanocortin receptors that are critical for body weight regulation.20

MC4R Gene Mutations

The MC4R is found in hypothalamic nuclei that regulate body weight by decreasing food intake and increasing energy expenditure.21 Homozygous MC4R knockout mice exhibit multiple metabolic phenotypes, including obesity, hyperphagia, hyperinsulinaemia, and hyperglycemia, while heterozygous mice present an intermediate obesity phenotype.22 In 1998, two independent studies reported multiple heterozygous frameshift mutations in the human MC4R gene that were associated with dominantly inherited obesity.23 , 24 Since then, more than 90 different mutations in this gene have been reported in obese subjects from various ethnic groups.4 These include frameshift, inframe deletion, nonsense, and missense mutations across the gene. Most of these mutations follow a dominant pattern of inheritance and result in partial or complete loss of receptor function.25 , 26 The prevalence of MC4R mutations ranges from 0.5% to 6% in severe cases of early-onset obesity.4 The prevalence of MC4R mutations in the general population, however, is very low. After screening 528 subjects for MC4R mutations by direct sequencing, Jacobson et al.27 detected six missense and six silent variants, but none were significantly associated with obesity or related phenotypes.

(p.441) In a German population of 1003 severely obese adults, nonsynonymous MC4R mutations that cause impaired receptor function occurred in only 2 subjects.28 Among 769 adult patients with a body mass index (BMI) of at least 35 kg/m2, the prevalence of obesity-specific MC4R mutations was 2.6%.29 Ma et al.30 sequenced the coding region of the MC4R gene in 426 full heritage, non-first-degree-related adult Pima Indians. They detected only three coding variations as heterozygotes in 12 of the 300 severely obese subjects in this population. Taken together, these studies suggest that although many pathogenic MC4R mutations have been identified, their prevalence in the general population is very low, and therefore, they account for only a small fraction of obesity.

Genetic Syndromes of Obesity

In a recent review of monogenic obesity in humans, Farooqi and O’Rahilly7 described approximately 30 rare syndromes of obesity caused by genetic mutations or chromosomal abnormalities. These syndromes are characterized by severe obesity and frequently accompanied by mental retardation. Prader-Willi Syndrome (PWS) is the most common, with 1 in 25,000 births affected.31 PWS is an autosomal-dominant disorder typically caused by a paternally inherited deletion at the chromosomal region 15q11.2-q12. It is characterized by obesity, hyperphagia, short status, mental retardation, and hypogonadotropic hypogonadism. Patients with PWS also have elevated circulating ghrelin, which may contribute to increased hunger and hyperphagia.32

Bardet-Biedl syndrome (BBS) is a very rare autosomal recessive disorder characterized by central obesity, mental retardation, hypogonadism in males, renal abnormalities, and pigmentary retinopathy.7 Eight BBS genes have been identified in various pedigrees through positional cloning and candidate gene studies, but their molecular functions have not been completely elucidated.33 Recently, a study in a French population showed an age-dependent association between common obesity and variants in BBS2, BBS4, and BBS6.34

Genetics of Common Obesity

Heritability of Obesity

The degree of genetic contribution to a trait such as obesity can be quantified by narrow-sense heritability, which is defined as the percent of total phenotypic variation that can be attributed to additive genetic effects (h 2 = V G/V P, where V G is the additive genetic variance and V P is the phenotypic variance).35 Findings from family and twin studies suggest that obesity and obesity-related traits have a substantial heritable component. Studies comparing monozygotic (MZ) and dizygotic (DZ) twins have been especially informative. MZ twins share all genes, and DZ twins an average of half, making twin studies a useful way to estimate genetic heritability of obesity. Such analyses are based on the “equal environment” assumption that degree of environmental sharing among MZ co-twins is the same as that among DZ co-twins, and thus, any difference in shared phenotype between MZ and DZ twins is due to genetic factors.35 On the basis of data from more than 25,000 twin pairs and 50,000 biological and adoptive family members, Maes et al.36 estimated that the mean correlations for BMI were 0.74 for MZ twins, 0.32 for DZ twins, 0.25 for siblings, 0.19 for parent-offspring pairs, 0.06 for adoptive relatives, (p.442) and 0.12 for spouses. The stronger correlations among MZ twins than DZ twins, siblings, or parent-offspring pairs suggest a strong genetic influence on BMI.

In 1977, the National Heart, Lung, and Blood Institute (NHLBI) Twin Study demonstrated genetic heritability of obesity and other cardiovascular risk factors.37 Since then, numerous twin studies have produced heritability estimates ranging from 25% to 90% for BMI.38 Heritability estimates have ranged from 65% to 75% for fat mass and percentage of body fat;39-41 46% to 90% for waist circumference;42-44 48% to 69% for skinfolds (total, extremity, and trunk)44; and 38% to 73% for serum leptin.45-46 Similar heritability estimates ranging from 50% to 70% have been obtained for BMI in MZ twins reared apart.47 While the classic twin study requires the equal environment assumption, studies of twins reared apart have the advantage of co-twins raised in uncorrelated environments through random replacement.

While adoptive parents and their adopted offspring share only environmental sources of variance, adoptees and their biological parents share only genetic sources. This makes adoption studies another useful approach for separating genetic and environmental influences on obesity traits. Data from these studies suggest that genetic factors account for 20% to 60% of variation in BMI.35 In a study of over 3,500 subjects from the Danish Adoption Register, a strong relationship was observed between the BMI of adoptees and biological parents across a wide range of body fatness.48 In contrast, no significant relationship was observed between adoptees and adoptive parents.

In summary, common obesity is clearly a heritable trait, although the exact degree of genetic heritability is still debatable. Nonetheless, the strong genetic basis of obesity has spurred intensive efforts to identify obesity genes over the past two decades. Unlike monogenic obesity, common obesity is likely to be caused by many genes. Linkage analysis and candidate-gene associations have been the primary approaches to identifying common obesity susceptibility genes. A description of these two approaches, including their strengths and weaknesses, is presented later. Recent findings from GWA studies are also discussed.

Linkage Analysis and “Positional Cloning”

Linkage analyses map genetic loci using data from related individuals, including siblings, nuclear families, and large pedigrees.49 In genome-wide linkage analyses, a series of anonymous markers across the entire genome is used without a priori hypotheses to identify regions of the genome that may harbor disease susceptibility genes. Evidence for linkage is evaluated by a logarithm of the odds (LOD) score first proposed by Morton in 1955.50 Larger LOD scores indicate greater evidence for linkage. Significant linkage is commonly defined as an LOD score >3.6, while suggestive linkage is defined as a score >2.2.51 These criteria, however, are arbitrary and a modest LOD score (e.g., in the range of 1.5 to 2.2) does not necessarily exclude linkage. Nonetheless, applying the stringent criterion of at least 3.6 (equivalent to the genome-wide type 1 error rate of 0.05) is intended to minimize false-positive results.52

Although linkage analysis was initially used for mapping genes that underlie monogenic obesity, this method has also been widely applied to map common obesity genes. According to the Human Obesity Gene Map,53 253 quantitative trait loci (QTL) have been identified in 61 genome-wide scans performed in various populations, including Caucasians, African Americans, Mexican Americans, and Asians. Scans have also been performed in isolated populations, such as Pima Indians and the Old Order Amish. A wide variety of obesity measures or biomarkers have been evaluated, including BMI, fat mass; fat-free mass; skinfold thickness; intra-abdominal fat; waist circumference; (p.443) adipocyte size; percentage of body fat; respiratory quotient; and levels of insulin, leptin, glucose, and adiponectin.

Despite large numbers of linkage studies on common obesity, few obesity susceptibility genes have been identified and replicated. This limited success is likely due to multiple factors. First, the pathogenesis of common obesity, unlike monogenic obesity, is likely to involve a large number of genes, with each contributing only a modest effect. Most family-based linkage studies are underpowered to detect these effects.54 Second, common obesity is a complex and heterogeneous phenotype. A variety of obesity-related quantitative traits, such as BMI and measures of body composition, have been used in linkage analysis. Since obesity is significantly related to energy metabolism, some intermediate phenotypes (e.g., leptin levels, resting metabolic rate, and respiratory quotient) have also been used to search for obesity QTL.55 However, different genes may regulate different obesity-related traits, leading to heterogeneity of linkage regions across various studies. Third, obesity linkage studies have been conducted in populations that live in diverse obesogenic environments. Environmental factors may modify genetic influences (Chapter 22). However, most linkage analyses do not take gene-environment interactions into account, a factor that may also contribute to heterogeneous results. Finally, because genes that influence early-onset obesity may differ from those that contribute to weight gain in later life, linkage studies in children and adults may implicate different genomic regions.

Linkage analysis is only the first step in the gene discovery process. Once a genomic region is identified, the next step is to clone the gene through fine mapping, association studies, and functional analyses—the so-called positional cloning technique. This technique has recently led to the identification of several novel genes that may contribute to common obesity. In a 1998 genome-wide scan of 158 obese French Caucasian families, Hager et al.56 reported significant evidence for linkage of obesity to a region of chromosome 10p. Replication studies in other populations have confirmed the linkage.57 Subsequently, Boutin et al.58 conducted fine mapping of the chromosome 10p locus by assessing 16 polymorphic markers around the linkage peak in 620 individuals from 188 nuclear families. Further analysis narrowed the linkage signal to one marker located in intron 7 of the glutamate decarboxylase 2 (GAD2) gene. Association tests in two independent case-control studies suggested a relationship between several single-nucleotide polymorphisms (SNPs) in GAD2 gene (−243 A>G, +61450 C>A, and +83897 T>A) and risk of morbid obesity (BMI > 40 kg/m2). The −243 A>G SNP, which was associated with 30% increased obesity risk, was also associated with significantly higher hunger and disinhibition scores. In addition, functional data showed a sixfold increase in GAD2-promoter activity for the risk allele of the −243 A>G SNP. GAD2 encodes the glutamic acid decarboxylase enzyme, which catalyzes the formation of gamma-aminobutyric acid (GABA). GABA interacts with NPY to stimulate hunger and food intake.59 GAD2 is therefore considered a strong positional and biological candidate gene for obesity. Attempts to replicate these associations, however, have produced mixed results. A large family study and two independent case-control samples showed no associations between the −243 A>G SNP or two other GAD2 SNPs and morbid obesity.60 In contrast, a second study by Meyre et al.61 confirmed the association of the GAD2 −243 A>G SNP with early-onset severe obesity among French children [odds ratio (OR), 1.25; P = .04].

In a genome-wide linkage scan of Finnish obese nuclear families, Ohman et al.62 reported linkage to chromosome Xq24. Suviolahti et al.63 further investigated this locus in 218 obese Finnish sibling pairs by genotyping 9 microsatellite markers and 36 SNPs for 11 candidate genes spanning the 15-Mb linkage region. This approach led to significant associations between SNPs in AGTR2, SLC6A14, and SLC25A5 and obesity. (p.444) A follow-up study of 117 cases and 182 controls from the Finnish population reported significant associations between obesity and SNPs 22510 C>G (rs20718772) and 20649 C>T (rs2011162) in SLC6A14. In a second study of Swedish-Finnish subjects, Tiwari et al.64 also reported an association between SNP 22150 C>G and obesity, but in the opposite direction. More recently, Durand et al.65 examined SNPs 20649 C>T and 22510 C>G in a French population of 1267 obese and 649 nonobese, normoglycemic subjects. Results confirmed the initial findings by Suviolahti et al.63 (OR: 1.23 for 20649T and 1.36 for 22510G). However, no relationship between either SNP and childhood obesity was observed. Potential reasons for the discrepant findings from these studies will be discussed later in the chapter.

Candidate Gene Association Studies

Candidate gene association studies test the relationships between polymorphic markers within selected candidate genes and the obesity phenotype. Candidate genes are typically selected on the basis of locations within genomic regions (positional candidates) implicated by linkage analysis to the obesity phenotypes or biological functions (functional candidates).66 GAD2 and SLC6A14 are good examples of positional candidate genes. Functional candidates can be derived from animal models of obesity, in vitro characteristics of gene variants related to energy metabolism, or genes that have been implicated in monogenic obesity (discussed earlier).

Once candidate genes are selected, the next step is to choose genetic markers within those genes. Genetic variation can occur in many forms, including SNPs, copy number variants, microsatellites, and deletions of entire genes or regions of a chromosome. SNPs are the most common form of genetic variation, accounting for more than 90% of the total variation in the human genome. Because SNPs are widespread across the genome (>10 million SNPs have been identified) and are easily genotyped by a number of genotyping platforms, these have been the most commonly used markers in association studies. Four criteria have been commonly used to choose SNPs for genotyping: (a) the prior probability of being functional (e.g., exonic SNPs are more likely to be functional than intronic SNPs); (b) the degree of linkage disequilibrium (LD) among the SNPs; (c) missense variants detected by sequencing; and (d) the availability of high-throughput and low-cost SNP arrays that cover the whole genome.66

Most of the candidate gene association studies discussed later have evaluated only one or few SNPs in a candidate gene. A comprehensive approach should genotype all common (>5% frequency) nonsynonymous coding SNPs as well as other candidate SNPs in the regulatory region and splicing-sites. Selections can be based on purported function or earlier reported association with obesity or an obesity-related phenotype. In addition, a small number of “tagging SNPs” can serve as efficient surrogates for most remaining common SNPs of unknown function.67 The choice of these surrogate SNPs has been facilitated by the completion of Phase II of the HapMap,68 a comprehensive survey of LD patterns in samples from three major continental populations: Africans, East Asians, and Europeans. A simple and effective algorithm can be used to choose a set of tagging SNPs to capture any other SNPs in the region that have high pairwise correlations with one of the tagging SNPs.69 , 70 This algorithm has been implemented in the program Haploview (http://www.broad.mit.edu/mpg/haploview/), which can also be used to visualize LD patterns among a set of SNPs.

Because of their simplicity, case-control studies of unrelated individuals are the most common type of association studies. Such studies compare the frequency of variant alleles of selected candidate genes in obese versus nonobese individuals and determine (p.445) whether there is an association between the alleles and obesity phenotype. Association studies can be also carried out in multiple families.71 For example, the Transmission Disequilibrium Test (TDT), using parent/affected offspring trios, assesses whether the transmission of an allele from heterozygous parents to affected children deviates from that expected by chance (50%).71 The main advantage of family-based association studies is that they are not affected by population stratification bias (discussed later), but they suffer from several other disadvantages that prevent their widespread use.66 Not only is it difficult to recruit family trios, but selectively recruiting subjects and their parents introduces potential bias towards early-onset disease. Moreover, the power of TDT is low because only the heterozygous parents are informative. In contrast, case-control studies of unrelated individuals are easier to conduct and more powerful. For these reasons, association studies of unrelated individuals have been the most popular method for genetic association studies.

To date, a large number of obesity candidate genes have been tested in association studies in various populations. Most of the genes were selected based on their potential functions related to appetite control, food intake, energy metabolism, and adipocyte differentiation. According to the Human Obesity Gene Map,53 426 findings of positive associations with 127 candidate genes have been reported from genetic association studies. The vast majority of these findings have not been replicated. Only 22 genes have been confirmed by at least five positive studies, with varying degrees of statistical significance. However, these positive studies were offset by an equal or even higher number of negative studies. Therefore, careful meta-analyses of all published genetic associations are often required to synthesize the evidence about reported associations. In the following sections, we briefly review obesity candidate genes that have been subject to meta-analyses. The results of these meta-analyses are summarized in Table 21.2.

β3-Adrenergic Receptor Gene W64R Polymorphism

β3-Adrenergic receptors (ADRB3) are mainly expressed in adipose tissue and play a key role in regulating lipolysis and thermogenesis.72 In 1995, several studies examined the association between obesity and a tryptophan (W) to arginine (R) substitution at amino acid position 64 in the ADRB3 gene.73-75 Kadowaki et al.74 reported a significantly higher BMI for Japanese with the RR genotype compared to those with the WW genotype (24.7 kg/m2 vs. 22.1 kg/m2), while Widen et al.75 found the R allele to be significantly associated with an elevated WHR in Finns. Subsequently, dozens of studies have been published on the relationship between this polymorphism and obesity or obesity-related traits. Three meta-analyses have been published with somewhat contradictory results. The first was conducted by Allison et al.76 in 1998 and included results from 23 studies. No significant association between the W64R polymorphism and BMI was observed. These findings contrasted with the meta-analysis by Fujisawa et al.77 published in the same year in which pooled results from 31 studies showed a significantly higher mean BMI (mean difference 0.30 kg/m2) among the R allele carriers than among noncarriers. In 2001, Kurokawa et al.78 performed a meta-analysis of 27 studies in Japanese populations and found a significant mean difference in BMI of 0.26 kg/m2 between R allele carriers and noncarriers. The frequency of the variant was higher in Japanese than in Caucasians, which may have improved the power to detect a small effect of the polymorphism on BMI when meta-analysis was restricted to studies of Japanese populations. However, subsequent studies conducted in other Japanese populations have produced mixed results,79 , 80 underscoring the need for further investigations.

(p.446)

Table 21.2 Summary of Meta-Analyses of Candidate Gene Variants Associated with BMI or Obesity-Related Phenotypes

Author (Year)

Gene, Variations

Number of Studies

Findings

Allison et al. (1998)76

β3-Adrenergic receptor (ADRB3), W64R

23 studies (n = 7399)

Not significantly associated with BMI

Fujisawa et al. (1998)77

ADRB3, W64R

31 studies (n = 9236)

The carriers had significantly higher BMI, with a mean difference of 0.30 (0.13–0.47)

Kurokawa et al. (2001)78

ADRB3, W64R

27 studies (n = 6582; all Japanese)

The carriers had significantly higher BMI, with a mean difference of 0.26 (0.18–0.42)

Heo et al. (2002)87

Leptin receptor (LEPR), K109R, Q223R, and K656N

9 studies (n = 3263)

None of the three variants was significantly associated with BMI or waist circumference

Masud et al. (2003)85

Peroxisome proliferator-activated receptor-γ (PPARG2), P12A

30 studies (n = 19 136)

A12 allele was significantly associated with greater BMI only among those with BMI ≥ 27, with a mean difference of 0.11 between the carriers and noncarriers

Geller et al. (2004)100

MC4R, V103I

14 studies (n = 7713)

Significantly associated with lower risk of obesity, OR = 0.69, 95% CI 0.50–0.96

Sookoian et al. (2005)94

Tumor Necrosis Factor-α (TNF), −308G>A

Obesity, 8 studies (n = 3562); BMI, 18 studies (n = 5009); WHR, 13 studies, (n = 3910); and leptin level, 4 studies (n = 845)

Associated with increased risk of obesity, OR = 1.23, 95% CI 1.04–1.45; associated with elevated BMI (P = .034) but not waist-to-hip ratio or leptin levels

Paracchini et al. (2005)86

Leptin receptor (LEPR), Q223R, K109R, and K656N; PPARG2, P12A

Q223R, 10 studies (n = 2972); K109R, 7 studies (n = 1696); and K656N, 7 studies (n = 2064); and PPARG2, 6 studies (n = 4022) (all healthy subjects)

None of the variants was significantly associated with obesity risk: Q223R, OR = 1.13, 95% CI 0.98–1.30; K109R, OR = 1.05, 95% CI 0.89–1.23; K656N, OR = 1.02, 95% CI 0.86–1.21; and P12A, OR = 1.13, 95% CI 0.98–1.29

Marti et al. (2006)91

Glucocorticoid receptor gene (GR or NR3C1), N363S

13 studies (n = 5909)

Carriers had slightly elevated BMI (0.18, 95% CI 0–0.35) than noncarriers. Not significantly associated with obesity risk, OR = 1.02, 95% CI 0.56–1.87

Qi et al. (2007)99

Interleukin 6 (IL6), −174G>C

19 studies (n = 26 944)

The genotypes were not significantly associated with BMI, waist circumference, or waist-to-hip ratio

(p.447) Peroxisome Proliferator Activated Receptor-γ Gene P12A Polymorphism

Peroxisome proliferator activated receptor-γ (PPARG) is an attractive obesity candidate gene because it regulates adipocyte differentiation, lipid metabolism, and insulin sensitivity.81 , 82 The most frequently studied PPARG variant is the proline (P) to alanine (A) substitution at amino acid 12 which reduces PPARG activity and improves insulin sensitivity.83 , 84 Masud et al.85 carried out a meta-analysis using 40 datasets from 30 independent studies to examine the effect of the P12A polymorphism on BMI. There was a negligible difference in mean BMI (0.07 kg/m2) between the A allele carriers and noncarriers. However, stratified analysis revealed significant differences only among obese subjects (mean difference of 0.11 kg/m2). Recently, Paracchini et al.86 summarized data from six case-control studies and reported a borderline significant increased risk of obesity associated with the A allele (OR: 1.13, 95% CI: 0.98 to 1.29).

LEPR Gene Polymorphisms

In addition to rare mutations in LEPR causing monogenic obesity, several common SNPs on this gene may be relevant to the common form of obesity. Three SNPs resulting in amino acid substitutions, including Q223R, K109R, and K656N, have been extensively examined with respect to obesity. The R223 and R109 variants occur more frequently among Asians than other ethnic groups, while the N656 variant is more frequent among Caucasians.86 In an earlier meta-analysis, Heo et al.87 summarized data from 9 studies yielding a total of 3263 related and unrelated subjects from diverse ethnic background. They found no significant relationships between the three LEPR alleles and BMI or waist circumference in the overall population or subgroups defined by age, sex, and ethnicity. A more recent meta-analysis of case-control studies yielded similar results.86 The pooled ORs for obesity were 1.13 (95% CI: 0.98 to 1.30) for Q223R (10 studies), 1.05 (95% CI: 0.89 to 1.23) for K109R (7 studies), and 1.02 (95% CI: 0.86 to 1.21) for K656N (7 studies).

Glucocorticoid Receptor Gene N365S Polymorphism

Increased cortisol production has been implicated in the development of visceral obesity (see Chapter 18). The glucocorticoid receptor belongs to a nuclear receptor subfamily and is involved in the regulation of the transcription of glucocorticoid-responsive genes.88 The glucocorticoid receptor gene (GRL) is located on chromosome 5q31.3 and contains a common asparagine (N) to serine (S) substitution at codon 363 of exon 2. The S variant increases the transactivating capacity and has been shown to be associated with an increased sensitivity to glucocorticoids.89 , 90 This variant has been associated with increased BMI, but the results have been inconsistent. Marti et al.91 conducted a meta-analysis to assess the association between the N363S polymorphism and obesity risk. The analysis, including 5909 subjects from 12 published and 3 unpublished studies, found that carriers of the S allele had a modest but significantly higher BMI than noncarriers (mean difference of 0.18 kg/m2). However, the association between this variant and obesity risk was not statistically significant.

Tumor Necrosis Factor-α Gene −308G>A Polymorphism

Tumor necrosis factor-α (TNFA) is an inflammatory cytokine that stimulates production of other cytokines and regulates glucose and lipid metabolism and insulin resistance.92 Adipose tissue is a major source of endogenous TNFA production, and elevated levels of TNFA are associated with increased adiposity and insulin resistance in humans. The G to A substitution at (p.448) position −308 in the promoter region of the TNFA gene has been shown in vitro to enhance nuclear factor binding, resulting in increased transcriptional activity.93 Sookoian et al.94 summarized association studies on this polymorphism in relation to obesity, insulin resistance, and hypertension (n = 3,562). There was an increased risk of obesity associated with the combined GA and AA genotypes compared with the GG genotype (OR: 1.23). Mean BMI and WHR, however, were not significantly different between the two genotype groups.

Interleukin-6 Gene −174G>C Polymorphism

Interleukin-6 (IL6) is a proinflammatory cytokine secreted by adipose tissue, immune cells, and muscles. Circulating levels of IL6 are elevated in obesity and predict development of both insulin resistance and type 2 diabetes.95 , 96 A −174G>C polymorphism within the IL6 promoter region has been associated with plasma IL6 levels, fasting insulin levels, measures of insulin sensitivity, and glucose homeostasis.97 , 98 However, a recent meta-analysis of 26,944 individuals from 19 studies found no significant association between this SNP and measures of adiposity (BMI, WHR, or waist circumference).99

MC4R Gene V103I Polymorphism

Genetic variability in MC4R has not only been implicated in monogenic obesity but also the common form of obesity. The V103I polymorphism on this gene has been extensively studied in regard to obesity risk. In a meta-analysis of 7,713 individuals from 14 studies, the pooled OR of obesity for the I allele carriers was 0.69 (95% CI: 0.50 to 0.96).100 A more recent study of 7,937 participants also reported a significantly inverse association between this variant and obesity risk.101

Taken together, meta-analyses suggest modest, if any, associations between widely studied polymorphisms and various measures of obesity. There appears to be more consistent evidence to support associations with genetic variations in GRL, MC4R, and TNFA, but the evidence is far from conclusive. While this review has focused on the main effects of polymorphisms, whether these candidate loci interact with the environment to modulate risk of obesity must also be considered. The topic of gene-environment interactions is further discussed in Chapter 22.

Genome-Wide Association Studies

Although the candidate gene approach has had some successes in identifying susceptible genes for common diseases, it has been hampered by the modest contribution of each SNP to overall heritability, the limited scope in surveying the large number of SNPs in the whole genome, and variability in criteria used for selecting candidate genes and SNPs.102 The candidate gene approach is also limited by our incomplete understanding of biological mechanisms of the disease. Instead of relying on selecting the correct genes, the GWA approach surveys the entire genome for causal genetic variants in a comprehensive and unbiased manner. Recent advances in genotyping technology have made this approach feasible.103 Commercially available products commonly used in GWA studies to simultaneously assay hundreds of thousands of loci include the Affymetrix and Illumina SNP chips. The SNPs available on these chip sets are selected either at random across the genome (earlier Affymetrix products) or based on LD from the HapMap (Illumina products and more recent Affymetrix products). These high-density SNP arrays can capture >80% of common variations (minor allele frequency >10%) in the human genome.104

(p.449) Multistage approaches have been commonly used to screen and replicate promising leads from GWA scans.105 In a two-stage design, a subset of available subjects are genotyped on a genome-wide SNP panel, then a much smaller subset of the most significant markers are genotyped on the remaining subjects. Such a study, if designed and analyzed appropriately, can have nearly as much power at a much lower cost than a single-stage design in which all subjects are genotyped on the genome-wide panel.106 , 107 Recently, the GWA approach has identified several common SNPs associated with chronic diseases in unexpected genes. The first successful example was the association between complement factor H variants and age-related macular degeneration.108 Subsequent GWA studies have identified several novel loci for type 2 diabetes,109-112 coronary heart disease,113-115 and other conditions.116 These findings demonstrate the potential of GWA analyses to identify new susceptibility genes for complex diseases.

The first GWA study specifically on obesity was conducted by Herbert et al.117 After genotyping 694 participants from the Framingham Heart Study offspring cohort for 116,204 SNPs, the authors found that only SNP rs7566605 G>C near the insulin-induced gene 2 (INSIG2) was significantly associated with obesity. The rs7566605 CC genotype was associated with obesity in three replication studies of family-based samples, as well as three of four case-control studies of unrelated individuals. A meta-analysis of all the case-control samples showed that the CC genotype was significantly associated with obesity under a recessive model, with an OR of 1.22. Because INSIG2 is involved in fatty acid and cholesterol synthesis,118 it is a plausible obesity candidate gene. However, subsequent replications have been inconsistent.119- 122

More recently, a GWA scan identified a common variant in the FTO gene that was associated with type 2 diabetes risk; this association was entirely mediated through its association with obesity.123 The association between SNP rs9939609 T>A and BMI, initially discovered in the GWA scan, was replicated in 13 cohorts with 38,759 participants. Adults with the AA genotype (16%) weighed about 3 kg more and had 67% increased risk of obesity compared to those with the TT genotype. Further analysis of two large birth cohorts suggested that the FTO SNP was not associated with changes in fetal growth, but was associated with childhood adiposity. The association between FTO variants and obesity has been confirmed in several additional studies.124 , 125 So far, these variants represent the most replicated genetic markers for common forms of obesity, although biological function of the FTO gene remains unknown.126

Methodological Problems in Obesity Association Studies

Association studies are commonly used to identify genetic variants that affect polygenic traits such as obesity. Such designs have had some success but have been plagued by lack of reproducibility. In a systematic review, Hirschhorn et al.127 found that of 166 putative associations studied three or more times, only six were reproduced at least 75% of the time. There are many potential reasons for the lack of reproducibility and these have been discussed extensively.66 , 128- 130 In the following section we briefly discuss a few of these.

False-Positive Findings and the Winner’s Curse

False-positives (type 1 errors) can arise from a number of sources, including chance findings or statistical fluctuation, multiple testing, and publication bias. Most nonreplicated (p.450) findings in the literature were initially positive, but could not be reproduced in subsequent studies. In most situations, the association in the first positive report exceeded the genetic effect estimated by meta-analysis in subsequent studies, a phenomenon referred to as the winner’s curse.131 , 132 Most of the genetic associations for common obesity reviewed in this chapter follow the winner’s curse pattern. Thus, the genetic associations identified in the first positive study cannot typically be used to estimate the overall or true genetic effect.131

Multiple Testing

With advances in genotyping technology, it is now feasible to assess a large number of SNPs simultaneously, which can increase the risk for false-positives due to multiple testing. This is becoming a growing concern, particularly with the advent of GWA, where hundreds of thousands of markers are assessed simultaneously. For these analyses, the standard significance threshold of α = 0.05 (producing one false-positive result for every 20 independent tests) is considered too liberal. Conversely, procedures that maintain strong control of the family-wide error rate (i.e., the probability of any false-positives), such as the Bonferroni correction, which is roughly equivalent to 0.05 divided by the total number of markers tests, are likely to be too conservative, and can increase the number of false-negative results. They are also inappropriate for GWA studies since the extensive LD in the genome ensures that many SNPs are correlated.133 An appropriate strategy to correct for multiple testing should, therefore, balance the risk of false-positives and false-negatives. Permutation testing is a nonparametric resampling approach that is used to control the family-wide type 1 error rate.134 Because this approach retains the correlations among SNPs present in the actual data, it is less conservative than the Bonferroni correction. In addition, the false discovery rate (FDR) method has been increasingly used to address the multiple testing issue; it controls the expected proportion of false-positives among all positive results, instead of controlling any chance of false-positive findings (as Bonferroni correction does).135 This procedure can reduce false-positive results while attaining greater power to detect true discoveries. Ultimately, replication of genetic associations across different populations is the best protection against false-positive findings resulting from multiple testing or other sources.

Genotyping Errors

Though typically low in most modern genotyping platforms, genotyping errors are inevitable in large association studies. They can lead not only to reduced power but also false-positive results.136 There are many causes of genotyping errors, including DNA contamination, calling of inappropriate alleles, and nonspecificity of experimental assays. The accuracy of genotyping is critical in association studies of complex diseases because a small genetic effect can be easily masked or exaggerated by even a small amount of genotyping error.137 By genotyping case and control samples together and blinding researchers and technicians to case-control status, systematic genotyping errors can be minimized. Because deviance from Hardy-Weinberg equilibrium (HWE) usually hints at genotyping errors, HWE tests should be performed for each SNP in the control samples before conducting genetic association tests.138 In GWA studies, stringent quality control procedures are necessary to minimize sample handling errors and remove poor-quality DNAs and SNPs before conducting the association analyses.110- 112

(p.451) Population Stratification

Population stratification can arise from disproportionate selection of cases and controls from genetically mixed populations.139 This particular form of confounding occurs when ethnicity or ancestry distort the relationship between a genetic marker and disease risk.140 One classic example of population stratification is a strong negative association between the Gm haplotype Gm3;5,13,14 and type 2 diabetes. Although initially observed in a sample of 4920 Pima Indians, it disappeared after adjustment for European ancestry.141 Because Gm3;5,13,14 is a marker for Caucasian admixture, overrepresentation of European ancestry in the controls led to an artificial inverse association between the marker and diabetes. In the literature, such clear examples of large biases created by population stratification are rare,142 and several simulation and empirical studies have found little evidence of bias due to stratification in carefully matched case-control designs.140 , 143 Potential bias from population stratification can be minimized by selecting cases and controls from an ethnically and racially homogeneous population and controlling for ancestry in the analyses. Prospective cohorts with cases and controls selected from a clearly defined source population are less susceptible to population stratification bias than retrospective studies.

Although large population stratification biases rarely occur between cases and controls, small biases due to subtle differences in genetic background are still of potential concern in association studies, even those with European-derived populations.144 Devlin and Roeder145 proposed a genomic control (GC) approach to control for population stratification in association studies. On the basis of the assumption that population stratification often leads to an inflated χ2 test that too often rejects the null hypothesis, the inflation factor lambda estimated from a set of randomly selected null loci are used to statistically correct for observed genetic associations. A drawback of this approach is that it may lead to overcorrection for markers that do not differ in frequency across subpopulations.

More recently, Price et al.146 proposed a principal component analysis-based method to adjust for population stratification. In this approach, several principal components are derived from genome-wide genotype data to capture population structure. These components are then included in the regression model as covariates to adjust for population structure. Typically, the first few components that capture most ancestry or ethnic differences between cases and controls are included as covariates to adjust for potential population stratification. There is some evidence that this approach is more powerful and provides better control of the type 1 error rate than the GC method.146

False-Negative Findings (Type 2 Errors)

False-negative results, usually arising from small underpowered studies, can also contribute to lack of replication in genetic association studies. Because most genetic variants have low penetrance and only modest effects on complex traits such as obesity, a large sample size (often in the range of thousands of cases and controls) is required for power to achieve even nominal significance.66 One good example is the association between the PPARG2 P12A variant and risk of type 2 diabetes. An initial study found a strong effect with an OR of 4.35 (P = .028) for the PP genotype,147 but four of five subsequent studies failed to confirm the association. A meta-analysis of more than 3,000 individuals found a modest (1.25-fold) but significant (P = .002) increase in diabetes risk associated with the more common P allele.84 The nonreplication in earlier individual studies is likely due (p.452) to small sample size and insufficient power. Indeed, recent association studies with much larger sample sizes have further replicated the association.148 , 149 This example illustrates that meta-analysis can substantially improve the power of genetic analyses and help reconcile divergent findings from multiple association studies. However, meta-analysis is not a substitute for large and well-designed association studies.

The frequency of the polymorphism is also an important determinant of the power of an association study. Most studies are designed to test the hypothesis that the genetic risk for a complex trait is due to disease predisposing alleles or haplotypes with relatively high frequencies (>5%) on the basis of the common disease-common variant hypothesis.131 A meta-analysis of 25 different reported associations has provided some support for this hypothesis,131 and several recent studies150 , 151 have suggested that a limited number of common haplotypes account for most of the variation in a candidate gene. However, there is also evidence that rare alleles or haplotypes contribute to complex traits. For example, the GRL N365S and MC4R V103I polymorphisms associated with obesity (discussed earlier) have <5% frequency of the variant allele. Testing the alternative hypothesis that rare alleles or haplotypes are responsible for common obesity will require much larger samples and genotyping efforts than most of the published studies.

Phenotype and/or genotype measurement errors are another consideration in power estimation in association studies152 because such errors may significantly reduce the power to identify a genetic association. Commonly used measures of obesity, such as BMI and waist circumference, are imperfect measures of adiposity and body fat distribution. In addition, self-reported measures are susceptible to differential misreporting according to actual obesity status (see Chapter 5). These errors, although relatively modest, can lead to diminished power and inconsistent associations in the literature, especially when the true genetic association is small.

Genuine Heterogeneity

Genuine heterogeneity in genetic associations may exist across different studies, although it is often difficult to distinguish it from nonreplications due to biases or inadequate power. As previously mentioned, epidemiologic studies have used BMI, waist circumference, percent body fat, weight change, and plasma leptin concentrations to assess adiposity. Although these measures are highly correlated, they reflect different aspects of fatness that may not be regulated by the same genetic mechanism. Moreover, while some studies have used moderate overweight as the phenotypic outcome, others have focused on morbid obesity. It is possible that the genetic loci contributing to morbid obesity differ from those for mild obesity.

Reduced exposure to environmental pressure may mean that genetics play a larger part in childhood obesity than in late-onset obesity.4 Thus, obese children are promising target populations for both linkage scans and association studies. In some studies, the magnitude of genetic associations for obesity appears to differ between samples of children and adults.65 These divergent results may be due to chance, but they may also represent real heterogeneity in the genetics of early-onset and late-onset obesity. Different genetic or environmental backgrounds in diverse populations are another source of genuine heterogeneity. To date, most genetic association studies of obesity and other complex traits have been conducted in white populations. For individual studies and meta-analyses, it is desirable to have ethnically homogenous samples that reduce potential bias due to population stratification. However, different genetic architecture and allele frequencies in different ethnic groups make it important to replicate these associations in other racial (p.453) and ethnic groups. The evidence for causality is strengthened if the same variant is found to be associated with the disease in multiple ethnic groups. However, lack of replication in other ethnic or racial groups does not necessarily invalidate the observed genetic associations because some genetic risk may be ethnic-specific.

Gene-environment interactions can also contribute to genuine heterogeneity in genetic associations. The premise of such interactions is that the genetic association is contingent on the environmental context of the populations. Thus, a genetic effect may manifest in one population but not the other, depending on background dietary and lifestyle factors. Although gene-environment interactions are widely believed to explain many of the inconsistent results in the literature, the search for such interactions has been challenging both conceptually and methodologically. In the next chapter, we will discuss the role of gene-environmental interactions in the development of obesity.

Summary

Efforts to map monogenic forms of obesity have been met with great success. Most monogenic obesity cases discovered so far appear to be caused by genetic alterations of the leptin and melanocortin pathways, including rare mutations on the LEP, LEPR, PC1, POMC, and MC4R genes. These genes play critical roles in appetite control, food intake, and energy homeostasis. Disruption of the functions of these genes causes childhood onset of severe forms of obesity. However, the number of obesity cases caused by single-gene mutations is extremely small. To date, less than 200 human obesity cases caused by single-gene mutations in 11 different genes have been reported in the literature,53 but these mutations do not appear to contribute to common forms of obesity.

Mapping genes for common forms of obesity has proven more difficult than initially anticipated. Despite a strong hereditary component of common obesity, the contributions of individual genes have not been clearly elucidated and most genetic associations have not been reproduced. Meta-analyses have been conducted for several commonly studied polymorphisms (Table 21.2). Overall, there is suggestive evidence to support the associations for the GRL N365S, MC4R V103I, and TNF −308 G>A polymorphisms, but the genetic effects are small and additional confirmation in large samples is clearly needed.

GWA scans have recently emerged as a comprehensive and powerful approach to identify genetic variants related to complex diseases. Using this approach, several common genetic variants associated with chronic disease have been uncovered in unexpected genes. The association between FTO variants and obesity identified through a diabetes GWA scan has been replicated in multiple populations. Because many GWA studies on chronic diseases have been conducted or are underway, and virtually all have collected anthropometric information, the data generated by these studies will provide a tremendous resource for identifying obesity susceptibility genes. Pooled analyses of these GWA studies are necessary to improve power and reduce false negative results.

Similar to other complex diseases, the puzzle of common obesity genetics cannot be solved through a single approach. Future studies will need to harness the resources from large, well-powered population-based studies for initial discovery, replication, and mining of gene-gene and gene-environment interactions for common types of obesity. Although GWA studies will become the mainstay of genetic epidemiology, functional and positional candidate genes will continue to be investigated in genetic association (p.454) studies of obesity. The identified associations need to be replicated in different ethnic and racial groups. Furthermore, fine mapping and functional studies are required to identify the causal variants. Animal models of obesity, gene expression studies, and advances in genomics technology will continue to provide new insights into biological mechanisms of obesity as well as new tools for genetic epidemiologic research.

References

Bibliography references:

1. Hofbauer KG. Molecular pathways to obesity. Int J Obes Relat Metab Disord. 2002;26(Suppl 2):S18–S27.

2. Cummings DE, Schwartz MW. Genetics and pathophysiology of human obesity. Annu Rev Med. 2003;54:453–471.

3. Clement K, Ferre P. Genetics and the pathophysiology of obesity. Pediatr Res. 2003;53:721–725.

4. Bell CG, Walley AJ, Froguel P. The genetics of human obesity. Nat Rev Genet. 2005;6:221–234.

5. Clement K. Genetics of human obesity. Proc Nutr Soc. 2005;64:133–142.

6. Clement K. Genetics of human obesity. C R Biol. 2006;329:608–622.

7. Farooqi IS, O’Rahilly S. Monogenic obesity in humans. Annu Rev Med. 2005;56:443–458.

8. Farooqi IS, Wangensteen T, Collins S, et al. Clinical and molecular genetic spectrum of congenital deficiency of the leptin receptor. N Engl J Med. 2007;18:237–247.

9. Zhang Y, Proenca R, Maffei M, Barone M, Leopold L, Friedman JM. Positional cloning of the mouse obese gene and its human homologue. Nature. 1994;372:425–432.

10. Horvath TL, Diano S, Sotonyi P, Heiman M, Tschop M. Minireview: ghrelin and the regulation of energy balance—a hypothalamic perspective. Endocrinology. 2001;142:4163–4169.

11. Montague CT, Farooqi IS, Whitehead JP, et al. Congenital leptin deficiency is associated with severe early-onset obesity in humans. Nature. 1997;387:903–908.

12. Farooqi IS, Matarese G, Lord GM, et al. Beneficial effects of leptin on obesity, T cell hyporesponsiveness, and neuroendocrine/metabolic dysfunction of human congenital leptin deficiency. J Clin Invest. 2002;110:1093–1103.

13. Chen H, Charlat O, Tartaglia LA, et al. Evidence that the diabetes gene encodes the leptin receptor: identification of a mutation in the leptin receptor gene in db/db mice. Cell. 1996;84:491–495.

14. Chua SC Jr, Chung WK, Wu-Peng XS, et al. Phenotypes of mouse diabetes and rat fatty due to mutations in the OB (leptin) receptor. Science. 1996;271:994–996.

15. Clement K, Vaisse C, Lahlou N, et al. A mutation in the human leptin receptor gene causes obesity and pituitary dysfunction. Nature. 1998;392:398–401.

16. Jackson RS, Creemers JW, Ohagi S, et al. Obesity and impaired prohormone processing associated with mutations in the human prohormone convertase 1 gene. Nat Genet. 1997;16:303–306.

17. Jackson RS, Creemers JW, Farooqi IS, et al. Small-intestinal dysfunction accompanies the complex endocrinopathy of human proprotein convertase 1 deficiency. J Clin Invest. 2003;112:1550–1560.

18. Krude H, Biebermann H, Luck W, Horn R, Brabant G, Gruters A. Severe early-onset obesity, adrenal insufficiency and red hair pigmentation caused by POMC mutations in humans. Nat Genet. 1998;19:155–157.

19. Krude H, Biebermann H, Schnabel D, et al. Obesity due to proopiomelanocortin deficiency: three new cases and treatment trials with thyroid hormone and ACTH4-10. J Clin Endocrinol Metab. 2003;88:4633–4640.

20. Krude H, Biebermann H, Gruters A. Mutations in the human proopiomelanocortin gene. Ann NY Acad Sci. 2003;994:233–239.

(p.455) 21. Mountjoy KG, Mortrud MT, Low MJ, Simerly RB, Cone RD. Localization of the melanocortin-4 receptor (MC4-R) in neuroendocrine and autonomic control circuits in the brain. Mol Endocrinol. 1994;8:1298–1308.

22. Huszar D, Lynch CA, Fairchild-Huntress V, et al. Targeted disruption of the melanocortin-4 receptor results in obesity in mice. Cell. 1997;88:131–141.

23. Yeo GS, Farooqi IS, Aminian S, Halsall DJ, Stanhope RG, O’Rahilly S. A frameshift mutation in MC4R associated with dominantly inherited human obesity. Nat Genet. 1998;111:2.

24. Vaisse C, Clement K, Guy-Grand B, Froguel P. A frameshift mutation in human MC4R is associated with a dominant form of obesity. Nat Gene. 1998;20:113–114.

25. Farooqi IS, Yeo GS, Keogh JM, et al. Dominant and recessive inheritance of morbid obesity associated with melanocortin 4 receptor deficiency. J Clin Invest. 2000;106:271–279.

26. Tao YX. Molecular mechanisms of the neural melanocortin receptor dysfunction in severe early onset obesity. Mol Cell Endocrinol. 2005;239:1–14.

27. Jacobson P, Ukkola O, Rankinen T, et al. Melanocortin 4 receptor sequence variations are seldom a cause of human obesity: the Swedish Obese Subjects, the HERITAGE Family Study, and a Memphis cohort. J Clin Endocrinol Metab. 2002;87:4442–4446.

28. Hinney A, Bettecken T, Tarnow P, et al. Prevalence, spectrum, and functional characterization of melanocortin-4 receptor gene mutations in a representative population-based sample and obese adults from Germany. J Clin Endocrinol Metab. 2006;91:1761–1769.

29. Lubrano-Berthelier C, Dubern B, Lacorte JM, et al. Melanocortin 4 receptor mutations in a large cohort of severely obese adults: prevalence, functional classification, genotype-phenotype relationship, and lack of association with binge eating. J Clin Endocrinol Metab. 2006;91:1811–1818.

30. Ma L, Tataranni PA, Bogardus C, Baier LJ. Melanocortin 4 receptor gene variation is associated with severe obesity in Pima Indians. Diabetes. 2004;53:2696–2699.

31. Goldstone AP. Prader-Willi syndrome: advances in genetics, pathophysiology and treatment. Trends Endocrinol Metab. 2004;15:12–20.

32. DelParigi A, Tschop M, Heiman ML, et al. High circulating ghrelin: a potential cause for hyperphagia and obesity in Prader-Willi syndrome. J Clin Endocrinol Metab. 2002;87:5461–5464.

33. Sheffield VC. Use of isolated populations in the study of a human obesity syndrome, the Bardet-Biedl syndrome. Pediatr Res. 2004;55:908–911.

34. Benzinou M, Walley A, Lobbens S, et al. Bardet-Biedl syndrome gene variants are associated with both childhood and adult common obesity in French Caucasians. Diabetes. 2006;55:2876–2882.

35. Bouchard C, Perusse L, Rice T, Rao DC. Chapter 9: Genetics of human obesity. In: Bray GA, Bouchard C, eds. Handbook of Obesity: Etiology and Pathophysiology. 2nd ed. New York: Marcel Dekker; 2004.

36. Maes HH, Neale MC, Eaves LJ. Genetic and environmental factors in relative body weight and human adiposity. Behav Genet. 1997;27:325–351.

37. Feinleib M, Garrison RJ, Fabsitz R, et al. The NHLBI twin study of cardiovascular disease risk factors: methodology and summary of results. Am J Epidemiol. 1977;106:284–285.

38. Bouchard C, Perusse L. Genetic aspects of obesity. Ann NY Acad Sci. 1993;699:26–35.

39. Hanisch D, Dittmar M, Hohler T, Alt KW. Contribution of genetic and environmental factors to variation in body compartments—a twin study in adults. Anthropol Anz. 2004;62:51–60.

40. Nguyen TV, Howard GM, Kelly PJ, Eisman JA. Bone mass, lean mass, and fat mass: same genes or same environments? Am J Epidemiol. 1998;147:3–16.

41. Faith MS, Pietrobelli A, Nunez C, Heo M, Heymsfield SB, Allison DB. Evidence for independent genetic influences on fat mass and body mass index in a pediatric twin sample. Pediatrics. 1999;104(1 Pt):61–67.

(p.456) 42. Rose KM, Newman B, Mayer-Davis EJ, Selby JV. Genetic and behavioral determinants of waist-hip ratio and waist circumference in women twins. Obes Res. 1998;6:383–392.

43. Selby JV, Newman B, Quesenberry CP Jr, et al. Genetic and behavioral influences on body fat distribution. Int J Obes. 1990;14:593–602.

44. Schousboe K, Visscher PM, Erbas B, et al. Twin study of genetic and environmental influences on adult body size, shape, and composition. Int J Obes Relat Metab Disord. 2004;28:39–48.

45. Li HJ, Ji CY, Wang W, Hu YH. A twin study for serum leptin, soluble leptin receptor, and free insulin-like growth factor-I in pubertal females. J Clin Endocrinol Metab. 2005;90:3659–3664.

46. Narkiewicz K, Szczech R, Winnicki M, et al. Heritability of plasma leptin levels: a twin study. J Hypertens. 1999;17:27–31.

47. Allison DB, Kaprio J, Korkeila M, Koskenvuo M, Neale MC, Hayakawa K. The heritability of body mass index among an international sample of monozygotic twins reared apart. Int J Obes Relat Metab Disord. 1996;20:501–506.

48. Stunkard AJ, Sorensen TI, Hanis C, et al. An adoption study of human obesity. N Engl J Med. 1986;314:193–198.

49. Dawn Teare M, Barrett JH. Genetic linkage studies. Lancet. 2005;366:1036–1044.

50. Morton NE. Sequential tests for the detection of linkage. Am J Hum Genet. 1955;7:277–318.

51. Lander E, Kruglyak L. Genetic dissection of complex traits: guidelines for interpreting and reporting linkage results. Nat Genet. 1995;11:241–247.

52. Khoury MJ, Beaty TH, Cohen BH. Fundamentals of Genetic Epidemiology. New York: Oxford University Press; 1993.

53. Rankinen T, Zuberi A, Chagnon YC, et al. The human obesity gene map: the 2005 update. Obesity (Silver Spring). 2006;14:529–644.

54. Risch NJ. Searching for genetic determinants in the new millennium. Nature. 2000;405:847–856.

55. Comuzzie AG, Allison DB. The search for human obesity genes. Science. 1998;280:1374–1377.

56. Hager J, Dina C, Francke S, et al. A genome-wide scan for human obesity genes reveals a major susceptibility locus on chromosome 10. Nat Genet. 1998;20:304–308.

57. Saar K, Geller F, Ruschendorf F, et al. Genome scan for childhood and adolescent obesity in German families. Pediatrics. 2003;111:321–327.

58. Boutin P, Dina C, Vasseur F, et al. GAD2 on chromosome 10p12 is a candidate gene for human obesity. PLoS Biol. 2003;1:E68.

59. Zheng H, Corkern M, Stoyanova I, Patterson LM, Tian R, Berthoud HR. Peptides that regulate food intake: appetite-inducing accumbens manipulation activates hypothalamic orexin neurons and inhibits POMC neurons. Am J Physiol Regul Integr Comp Physiol. 2003;284:R1436–R1444.

60. Swarbrick MM, Waldenmaier B, Pennacchio LA, et al. Lack of support for the association between GAD2 polymorphisms and severe human obesity. PLoS Biol. 2005;3:e315.

61. Meyre D, Boutin P, Tounian A, et al. Is glutamate decarboxylase 2 (GAD2) a genetic link between low birth weight and subsequent development of obesity in children? J Clin Endocrinol Metab. 2005;90:2384–2390.

62. Ohman M, Oksanen L, Kaprio J, et al. Genome-wide scan of obesity in Finnish sibpairs reveals linkage to chromosome Xq24. J Clin Endocrinol Metab. 2000;85:3183–3190.

63. Suviolahti E, Oksanen LJ, Ohman M, et al. The SLC6A14 gene shows evidence of association with obesity. J Clin Invest. 2003;112:1762–1772.

64. Tiwari HK, Allison DB. Do allelic variants of SLC6A14 predispose to obesity? J Clin Invest. 2003;112:1633–1636.

(p.457) 65. Durand E, Boutin P, Meyre D, et al. Polymorphisms in the amino acid transporter solute carrier family 6 (neurotransmitter transporter) member 14 gene contribute to polygenic obesity in French Caucasians. Diabetes. 2004;53:2483–2486.

66. Newton-Cheh C, Hirschhorn JN. Genetic association studies of complex traits: design and analysis issues. Mutat Res. 2005;573:54–69.

67. Neale BM, Sham PC. The future of association studies: gene-based analysis and replication. Am J Hum Genet. 2004;75:353–362.

68. International HapMap Consortium. A haplotype map of the human genome. Nature. 2005;437:1299–1320.

69. de Bakker PI, Yelensky R, Pe’er I, Gabriel SB, Daly MJ, Altshuler D. Efficiency and power in genetic association studies. Nat Genet. 2005;37:1217–1223.

70. Carlson CS, Eberle MA, Rieder MJ, Yi Q, Kruglyak L, Nickerson DA. Selecting a maximally informative set of single-nucleotide polymorphisms for association analyses using linkage disequilibrium. Am J Hum Genet. 2004;74:106–120.

71. Spielman RS, McGinnis RE, Ewens WJ. Transmission test for linkage disequilibrium: the insulin gene region and insulin-dependent diabetes mellitus (IDDM). Am J Hum Genet. 1993;52:506–516.

72. Krief S, Lonnqvist F, Raimbault S, et al. Tissue distribution of beta 3-adrenergic receptor mRNA in man. J Clin Invest. 1993;91:344–349.

73. Clement K, Vaisse C, Manning, B St. J., et al. Genetic variation in the beta3-adrenergic receptor and an increased capacity to gain weight in patients with morbid obesity. N Engl J Med. 1995;333:352–354.

74. Kadowaki H, Yasuda K, Iwamoto K, et al. A mutation in the beta 3-adrenergic receptor gene is associated with obesity and hyperinsulinemia in Japanese subjects. Biochem Biophys Res Commun. 1995;215:555–560.

75. Widen E, Lehto M, Kanninen T, Walston J, Shuldiner AR, Groop LC. Association of a polymorphism in the beta 3-adrenergic-receptor gene with features of the insulin resistance syndrome in Finns. N Engl J Med. 1995;10:348–351.

76. Allison DB, Heo M, Faith MS, Pietrobelli A. Meta-analysis of the association of the Trp64Arg polymorphism in the beta3 adrenergic receptor with body mass index. Int J Obes Relat Metab Disord. 1998;22:559–566.

77. Fujisawa T, Ikegami H, Kawaguchi Y, Ogihara T. Meta-analysis of the association of Trp64Arg polymorphism of beta 3-adrenergic receptor gene with body mass index. J Clin Endocrinol Metab. 1998;83:2441–2444.

78. Kurokawa N, Nakai K, Kameo S, Liu ZM, Satoh H. Association of BMI with the beta3-adrenergic receptor gene polymorphism in Japanese: meta-analysis. Obes Res. 2001;9:741–745.

79. Oizumi T, Daimon M, Saitoh T, et al. Genotype Arg/Arg, but not Trp/Arg, of the Trp64Arg polymorphism of the beta(3)-adrenergic receptor is associated with type 2 diabetes and obesity in a large Japanese sample. Diabetes Care. 2001;24:1579–1583.

80. Matsushita Y, Yokoyama T, Yoshiike N, et al. The Trp(64)Arg polymorphism of the beta(3)-adrenergic receptor gene is not associated with body weight or body mass index in Japanese: a longitudinal analysis. J Clin Endocrinol Metab. 2003;88:5914–5920.

81. Berger J, Moller DE. The mechanisms of action of PPARs. Annu Rev Med. 2002;53:409–435.

82. Ristow M, Muller-Wieland D, Pfeiffer A, Krone W, Kahn CR. Obesity associated with a mutation in a genetic regulator of adipocyte differentiation. N Engl J Med. 1998;339:953–959.

83. Stumvoll M, Haring H. The peroxisome proliferator-activated receptor-gamma2 Pro12Ala polymorphism. Diabetes. 2002;51:2341–2347.

84. Altshuler D, Hirschhorn JN, Klannemark M, et al. The common PPARgamma Pro12Ala polymorphism is associated with decreased risk of type 2 diabetes. Nat Genet. 2000;26:76–80.

(p.458) 85. Masud S, Ye S, SAS Group. Effect of the peroxisome proliferator activated receptor-gamma gene Pro12Ala variant on body mass index: a meta-analysis. J Med Genet. 2003;40:773–780.

86. Paracchini V, Pedotti P, Taioli E. Genetics of leptin and obesity: a HuGE review. Am J Epidemiol. 2005;162:101–114.

87. Heo M, Leibel RL, Fontaine KR, et al. A meta-analytic investigation of linkage and association of common leptin receptor (LEPR) polymorphisms with body mass index and waist circumference. Int J Obes Relat Metab Disord. 2002;26:640–646.

88. van Rossum EF, Russcher H, Lamberts SW. Genetic polymorphisms and multifactorial diseases: facts and fallacies revealed by the glucocorticoid receptor gene. Trends Endocrinol Metab. 2005;16:445–450.

89. Huizenga NA, Koper JW, De Lange P, et al. A polymorphism in the glucocorticoid receptor gene may be associated with and increased sensitivity to glucocorticoids in vivo. J Clin Endocrinol Metab. 1998;83:144–151.

90. Lin RC, Wang WY, Morris BJ. High penetrance, overweight, and glucocorticoid receptor variant: case-control study. BMJ. 1999;319:1337–1338.

91. Marti A, Ochoa MC, Sanchez-Villegas A, et al. Meta-analysis on the effect of the N363S polymorphism of the glucocorticoid receptor gene (GRL) on human obesity. BMC Med Genet. 2006;7:50.

92. Hotamisligil GS, Spiegelman BM. Tumor necrosis factor alpha: a key component of the obesity-diabetes link. Diabetes. 1994;43:1271–1278.

93. Kroeger KM, Carville KS, Abraham LJ. The -308 tumor necrosis factor-alpha promoter polymorphism effects transcription. Mol Immunol. 1997;34:391–399.

94. Sookoian SC, Gonzalez C, Pirola CJ. Meta-analysis on the G-308A tumor necrosis factor alpha gene variant and phenotypes associated with the metabolic syndrome. Obes Res. 2005;13:2122–2131.

95. Pickup JC, Mattock MB, Chusney GD, Burt D. NIDDM as a disease of the innate immune system: association of acute-phase reactants and interleukin-6 with metabolic syndrome X. Diabetologia. 1997;40:1286–1292.

96. Hu FB, Meigs JB, Li TY, Rifai N, Manson JE. Inflammatory markers and risk of developing type 2 diabetes in women. Diabetes. 2004;53:693–700.

97. Fernandez-Real JM, Broch M, Vendrell J, et al. Interleukin-6 gene polymorphism and insulin sensitivity. Diabetes. 2000;49:517–520.

98. Kubaszek A, Pihlajamaki J, Punnonen K, Karhapaa P, Vauhkonen I, Laakso M. The C-174G promoter polymorphism of the IL-6 gene affects energy expenditure and insulin sensitivity. Diabetes. 2003;52:558–561.

99. Qi L, Zhang C, van Dam RM, Hu FB. Interleukin-6 genetic variability and adiposity: associations in two prospective cohorts and systematic review in 26,944 individuals. J Clin Endocrinol Metab. 2007;92:3618–3625.

100. Geller F, Reichwald K, Dempfle A, et al. Melanocortin-4 receptor gene variant I103 is negatively associated with obesity. Am J Hum Genet. 2004;74:572–581.

101. Heid IM, Vollmert C, Hinney A, et al. Association of the 103I MC4R allele with decreased body mass in 7937 participants of two population based surveys. J Med Genet. 2005;42:e21.

102. Hirschhorn JN, Daly MJ. Genome-wide association studies for common diseases and complex traits. Nat Rev Genet. 2005;6:95–108.

103. Altshuler D, Daly M. Guilt beyond a reasonable doubt. Nat Genet. 2007;39(7):813–815.

104. Pe’er I, de Bakker PI, Maller J, Yelensky R, Altshuler D, Daly MJ. Evaluating and improving power in whole-genome association studies using fixed marker sets. Nat Genet. 2006;38:663–667.

105. Kraft P. Efficient two-stage genome-wide association designs based on false positive report probabilities. Pac Symp Biocomput. 2006:523–534.

106. Wang H, Thomas DC, Pe’er I, Stram DO. Optimal two-stage genotyping designs for genome-wide association scans. Genet Epidemiol. 2006;30:356–368.

(p.459) 107. Skol AD, Scott LJ, Abecasis GR, Boehnke M. Joint analysis is more efficient than replication-based analysis for two-stage genome-wide association studies. Nat Genet. 2006;38:209–213.

108. Klein RJ, Zeiss C, Chew EY, et al. Complement factor H polymorphism in age-related macular degeneration. Science. 2005;308:385–389.

109. Sladek R, Rocheleau G, Rung J, et al. A genome-wide association study identifies novel risk loci for type 2 diabetes. Nature. 2007;445:881–885.

110. Zeggini E, Weedon MN, Lindgren CM, et al. Replication of genome-wide association signals in UK samples reveals risk loci for type 2 diabetes. Science. 2007;316:1336–1341.

111. Diabetes Genetics Initiative of Broad Institute of Harvard and MIT, Lund University and Novartis Institutes of BioMedical Research, Saxena R, et al. Genome-wide association analysis identifies loci for type 2 diabetes and triglyceride levels. Science. 2007;316:1331–1336.

112. Scott LJ, Mohlke KL, Bonnycastle LL, et al. A genome-wide association study of type 2 diabetes in Finns detects multiple susceptibility variants. Science. 2007;316:1341–1345.

113. McPherson R, Pertsemlidis A, Kavaslar N, et al. A common allele on chromosome 9 associated with coronary heart disease. Science. 2007;316:1488–1491.

114. Helgadottir A, Thorleifsson G, Manolescu A, et al. A common variant on chromosome 9p21 affects the risk of myocardial infarction. Science. 2007;316:1491–1493.

115. Samani NJ, Erdmann J, Hall AS, et al. Genomewide association analysis of coronary artery disease. N Engl J Med. 2007;357:443–453.

116. Wellcome Trust Case Control Consortium. Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls. Nature. 2007;447:661–678.

117. Herbert A, Gerry NP, McQueen MB, et al. A common genetic variant is associated with adult and childhood obesity. Science. 2006;312:279–283.

118. Yabe D, Brown MS, Goldstein JL. Insig-2, a second endoplasmic reticulum protein that binds SCAP and blocks export of sterol regulatory element-binding proteins. Proc Natl Acad Sci USA. 2002;99:12753–12758.

119. Dina C, Meyre D, Samson C, et al. Comment on “A common genetic variant is associated with adult and childhood obesity.” Science. 2007;315:187.

120. Loos RJ, Barroso I, O’rahilly S, Wareham NJ. Comment on “A common genetic variant is associated with adult and childhood obesity.” Science. 2007;315:187.

121. Rosskopf D, Bornhorst A, Rimmbach C, et al. Comment on “A common genetic variant is associated with adult and childhood obesity.” Science. 2007;315:187.

122. Lyon HN, Emilsson V, Hinney A, et al. The association of a SNP upstream of INSIG2 with body mass index is reproduced in several but not all cohorts.PLoS Genet. 2007;3(4):e61.

123. Frayling TM, Timpson NJ, Weedon MN, 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.

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

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

126. Groop L. From fused toes in mice to human obesity. Nat Genet. 2007;39(6):706–707.

127. Hirschhorn JN, Lohmueller K, Byrne E, Hirschhorn K. A comprehensive review of genetic association studies. Genet Med. 2002;4:45–61.

128. Hattersley AT, McCarthy MI. What makes a good genetic association study? Lancet. 2005;366:1315–1323.

129. Cordell HJ, Clayton DG. Genetic association studies. Lancet. 2005;366:1121–1131.

130. Kathiresan S, Newton-Cheh C, Gerszten RE. On the interpretation of genetic association studies. Eur Heart J. 2004;25:1378–1381.

(p.460) 131. Lohmueller KE, Pearce CL, Pike M, Lander ES, Hirschhorn JN. Meta-analysis of genetic association studies supports a contribution of common variants to susceptibility to common disease. Nat Genet. 2003;33:177–182.

132. Ioannidis JP, Ntzani EE, Trikalinos TA, Contopoulos-Ioannidis DG. Replication validity of genetic association studies. Nat Genet. 2001;29:306–309.

133. Nyholt DR. A simple correction for multiple testing for single-nucleotide polymorphisms in linkage disequilibrium with each other. Am J Hum Genet. 2004;74:765–769.

134. Dudbridge F. A note on permutation tests in multistage association scans. Am J Hum Genet. 2006;78(6):1094–1095.

135. Benjamini Y, Drai D, Elmer G, Kafkafi N, Golani I. Controlling the false discovery rate in behavior genetics research. Behav Brain Res. 2001;125:279–284.

136. Hao K, Li C, Rosenow C, Hung Wong W. Estimation of genotype error rate using samples with pedigree information—an application on the GeneChip Mapping 10K array. Genomics. 2004;84:623–630.

137. Xu J, Turner A, Little J, Bleecker ER, Meyers DA. Positive results in association studies are associated with departure from Hardy-Weinberg equilibrium: hint for genotyping error? Hum Genet. 2002;111:573–574.

138. Hosking L, Lumsden S, Lewis K, et al. Detection of genotyping errors by Hardy-Weinberg equilibrium testing. Eur J Hum Genet. 2004;12:395–399.

139. Redden DT, Allison DB. Nonreplication in genetic association studies of obesity and diabetes research. J Nutr. 2003;133:3323–3326.

140. Wacholder S, Rothman N, Caporaso N. Population stratification in epidemiologic studies of common genetic variants and cancer: quantification of bias. J Natl Cancer Inst. 2000;92:1151–1158.

141. Knowler WC, Williams RC, Pettitt DJ, Steinberg AG. Gm3;5,13,14 and type 2 diabetes mellitus: an association in American Indians with genetic admixture. Am J Hum Genet. 1988;43:520–526.

142. Cardon LR, Palmer LJ. Population stratification and spurious allelic association. Lancet. 2003;361:598–604.

143. Ardlie KG, Lunetta KL, Seielstad M. Testing for population subdivision and association in four case-control studies. Am J Hum Genet. 2002;71:304–311.

144. Campbell CD, Ogburn EL, Lunetta KL, et al. Demonstrating stratification in a European American population. Nat Genet. 2005;37:868–872.

145. Devlin B, Roeder K. Genomic control for association studies. Biometrics. 1999;55:997–1004.

146. Price AL, Patterson NJ, Plenge RM, Weinblatt ME, Shadick NA, Reich D. Principal components analysis corrects for stratification in genome-wide association studies. Nat Genet. 2006;38:904–909.

147. Deeb SS, Fajas L, Nemoto M, et al. A Pro12Ala substitution in PPARgamma2 associated with decreased receptor activity, lower body mass index and improved insulin sensitivity. Nat Genet. 1998;20:284–287.

148. Florez JC, Burtt N, de Bakker PI, et al. Haplotype structure and genotype-phenotype correlations of the sulfonylurea receptor and the islet ATP-sensitive potassium channel gene region. Diabetes. 2004;53:1360–1368.

149. Weedon MN, McCarthy MI, Hitman G, et al. Combining information from common type 2 diabetes risk polymorphisms improves disease prediction. PLoS Med. 2006;3:e374.

150. Johnson GC, Esposito L, Barratt BJ, et al. Haplotype tagging for the identification of common disease genes. Nat Genet. 2001;29:233–237.

151. Daly MJ, Rioux JD, Schaffner SF, Hudson TJ, Lander ES. High-resolution haplotype structure in the human genome. Nat Genet. 2001;29:229–232.

152. Gordon D, Finch SJ. Factors affecting statistical power in the detection of genetic association. J Clin Invest. 2005;115:1408.