J Pediatr Endocrinol Metab 2021; 34(2): 151–162

Review Article

Birsen Yılmaz* and Makbule Gezmen Karadağ

The current review of adolescent obesity: the role

of genetic factors
https://doi.org/10.1515/jpem-2020-0480 obesity” can be used instead of both childhood and
Received August 18, 2020; accepted October 20, 2020; adolescent obesity [3].
published online November 16, 2020 The direct measurement methods are expensive and not
accessible, so in the evaluation of the body fat amount, the
Abstract: Obesity, a complex, multi-factor and heteroge-
body mass index (BMI, kg/m2) calculation is generally used
neous condition, is thought to result from the interaction of
in the clinic. BMI is accepted as a standard measurement
environmental and genetic factors. Considering the result
method for two-year-olds and older. Thanks to the BMI,
of adolescence obesity in adulthood, the role of genetic
body fat amount and adiposity are estimated in a healthy
factors comes to the fore. Recently, many genome-wide
adolescent population. However, BMI value may cause false
association studies (GWAS) have been conducted and
predictions in short adolescents with high muscle mass
many loci associated with adiposity have been identified.
[4, 5]. In addition to BMI, waist circumference and waist-hip
In adolescents, the strongest association with obesity has
ratio are used to evaluate abdominal obesity, and skinfold
been found in single nucleotide polymorphisms (SNP) in
thickness is used as an indicator of adiposity [6–8].
the FTO gene. Besides FTO, GWAS showed consistent ef-
There is a common formulation of BMI for the classi-
fects between variants in MC4R, TMEM18, TNNI3K, SEC16B,
fication of obesity in adults. However, it is hard to develop
GNPDA2, POMC and obesity. However, these variants may
a common index in children and adolescents due to
not have similar effects for all ethnic groups. Although
physiological changes occurring with growth and devel-
recently genetic factors are considered to contribute to
opment [9]. For this reason, the World Health Organization
obesity, relatively little is known about the specific loci
has developed some methods depending on age. WHO is-
related to obesity and the mechanisms by which they cause
sued “Standard Growth Curves” for ages 0–5 in 2006 and
obesity.
“Growth Reference Values” for children and adolescents
Keywords: adiposity; adolescent; FTO; genetic; obesity; aged between 5 and 19 in 2007 [10]. Although these stan-
polymorphism. dards are employed worldwide, it is emphasized that
countries should work on their standards due to the sig-
nificant differences in late childhood [11].
The Centers for Disease Control and Prevention (CDC)
Introduction defines slightly overweight and obese according to the
percentiles determined for age and gender by WHO.
There is no currently internationally accepted definition
Overweight is defined as BMI ≥85 percentile and <95
of childhood and adolescent obesity [1]. However, the
percentile while obesity is defined as BMI 95 percentile and
World Health Organization (WHO) defines overweight and
above [12]. The classification made by the CDC expert
obesity as “abnormal or excessive fat accumulation that
committee for adolescents and children is given in Table 1
presents a risk to health” [2, 3]. In the literature, “childhood
[13]. The Disease Control and Prevention Center recom-
mends WHO’s standard growth curves for infants and
*Corresponding author: Birsen Yılmaz, Department of Nutrition and young children under two years, and the CDC/National
Dietetics, Research Assistant, Faculty of Health Sciences, Gazi Health Statistics Center growth references for children 2
University, Emek, Bişkek Cad., 6. Sok, Gazi Üniversitesi No: 2, 06490, years and older [13].
Çankaya, Ankara, Turkey; and Department of Nutrition and Dietetics, In recent years, the prevalence of obesity in children
Faculty of Health Sciences, Çukurova University, Sarıçam, Adana,
and adolescents has increased significantly in developed
Turkey, Fax: +90 (312) 216 26 36, E-mail: dytbirsen@gmail.com.
https://orcid.org/0000-0002-4866-2818
and developing countries and has become a global public
Makbule Gezmen Karadağ, Department of Nutrition and Dietetics, health issue. According to the current data of the CDC, the
Faculty of Health Sciences, Gazi University, Çankaya, Ankara, Turkey obesity rate in children and adolescents aged 2–19 has
152 Yılmaz and Gezmen Karadağ: Role of genetic factors in adolescent obesity

Table : Body weight classification in adolescents. increase in adolescent obesity rates should be seriously
taken into account [24]. Therefore, it is important to un-
Body weight category Percentile range derstand the etiology of obesity, which is becoming a major
Underweight < percentile public health problem day by day. Considering the limited
Normal or healthy weight ≥ percentile and < percentile genetic data on obesity during adolescence, this review
Overweight ≥ percentile and < percentile
attempts to explain the role and mechanisms of underlying
Obese ≥ percentile
genetic factors in adolescent obesity.

reached 18.5%, affecting a total of 13.7 million children and

adolescents [14].
Adolescent obesity and genetic
According to the nutrition research conducted with 43 factors
million children living in 144 different countries by the
WHO, the rate of slightly overweight and obese in pre- The prevalence of obesity and obesity-related disease has
school children increased from 4.2% in 1990 to 6.7% in risen rapidly in recent years and has become a major public
2010 [15]. Considering the prevalence of obesity in ado- health problem in many populations [25]. Although
lescents, it increased from 10.6% in 1988–1994 to 20.6% in adolescent obesity is explained as an energy balance
2013–2014 [16]. About 5.9% of children and adolescents disruption caused by increased energy intake and/or
aged 2–19 have been reported to be seriously obese (BMI 95 decreased energy expenditure, childhood obesity has an
percentile 120% or BMI ≥35 kg/m2) [17]. exceedingly complex etiology. It is a pathological condi-
To evaluate the BMI changes over the years and the tion resulting from the interaction of many factors. These
BMI trends in the coming years, 19.2 million people from factors are mainly environmental, genetic, behavioral and
200 different countries were investigated. The study ecological (family, school, etc.) [18].
covering the years 1975–2014 revealed that the probability The changes in the lifestyle of individuals in recent
of achieving the global obesity target would be almost years have also seriously affected the epidemic of obesity.
zero if trends continue after 2000. In 2025, the global Some researchers note that lifestyle changes are the
prevalence of obesity is expected to reach 18% in men and result of a change called “social globalization”. With this
exceed 21% in women, and the serious obesity rate is change, individuals’ access to foods with high energy
expected to exceed 6% in men and 9% in women. How- content has become easier, and a sedentary lifestyle has
ever, it is anticipated that the prevalence of low-weight started to become widespread [26]. However, populations
individuals will continue in poor regions of the world, sharing the same obesogenic environment have both
especially in South Asia [18]. The prevalence of obesity in obese and normal-weight individuals with different rates
adolescents in Turkey is reported to be 2–15% [19]. Ac- of obesity. This can be explained by the fact that some
cording to the Turkey Nutrition and Health Survey, the individuals are more sensitive to gaining weight than
obesity prevalence age of 6–18 has been reported to be others when exposed to the same obesogenic environ-
8.2% (9.1% male, female 7.3%) [20]. Alper et al. [21] ment. This difference is considered due to the individual
investigated the obesity prevalence trends in children and genetic profile [22].
adolescents in Turkey and reported that it increased 11.6 The polygenic/common obesity seen in many cases
times between 2011–2015 and 1990–1995 and rose from arises from the combination of environmental factors with
0.6 to 7.3%. many gene variants of minor effect, while genes have a
Three types of obesity can be considered based upon major effect in monogenic obesity [27, 28]. It is reported
genetic and phenotypic traits: monogenic syndromic obesity, that at least 5–7% of nonsyndromic early-onset obesity is
monogenic nonsyndromic obesity and polygenic/common monogenic. Monogenic obesity is explained by before age
obesity. The activity of a single gene (Mendelian/monogenic) 5 years and the BMI is over three standard deviations
or more than one gene (polygenic/common) may be reflected above normal [27]. The leptin–melanocortin system,
in a trait [22]. Nonsyndromic early-onset severe obesity is located in the hypothalamus, is one of the most significant
generally monogenic, while other forms of obesity are poly- pathways responsible for controlling body weight and
genic and take place because of the cumulative action of food intake. Most nonsyndromic forms of obesity result
multiple susceptibility genes [23]. from mutations in the genes which affect the leptin–
Since obesity and obesity-related noncommunicable melanocortin pathway [27, 29]. While the mutations in
diseases decrease the quality of life of adolescents, the LEP, LEPR, POMC, PCSK1 and TUB genes are autosomal
 Yılmaz and Gezmen Karadağ: Role of genetic factors in adolescent obesity 153

recessive, autosomal dominant mutations have been [37]. In a systematic review that examined the effects of
identified in BDNF, NTRK2, SIM1, MC4R, SH2B1, MRAP2 genetics and the environment in twins and adopted chil-
and LRP2 genes [27]. These mutations are caused by de- dren, it was indicated that genetic factors may have a much
fects in proteins that act in the hypothalamus to control stronger effect on BMI trends than environmental factors
food intake and energy expenditure. Children with these up to the age of 18 [38].
types of defects can develop severe hyperphagia, which
leads to rapid body weight gain in the first year of life. It is
to be acknowledged that the most common monogenic Thrifty genotype hypothesis
form of obesity is because of the mutations in MC4R gene
in humans. [30]. The Thrifty Genotype Hypothesis was first developed by
There are some suggestions in the literature that early the American geneticist James V. Neel, and it was revised
body weight-related problems may continue in adulthood later. This hypothesis, which emerged concerning gene-
[31]. Childhood obesity importantly increases the preva- environment interactions, includes the concept of “thrifty
lence of adult obesity and is related to a lot of comorbidities genes”. Although Neel reported that these genes increase
[32]. In an American cohort study conducted to determine susceptibility to type 2 diabetes in certain groups of people,
the effects of adolescent obesity, adolescents were fol- this hypothesis often sheds light on studies related to
lowed periodically in four waves at certain intervals (first obesity [39, 40]. According to this hypothesis, indigenous
Wave: 1994–1995, 20,745 individuals aged 11–20 years, people with a hunter-gatherer lifestyle did not always
average age 15.9 years; second Wave, 1996, 14,738 in- achieve sufficient food, and they had famine periods from
dividuals aged 12–21, average age 16.5 years; third Wave, time to time. As a result of repeated recurrence of feeding
2001–2002, 15,197 individuals aged 18–27, average age 22.3 and fasting periods, adaptation occurred in these in-
years; and fourth Wave, 2007–2009, 15,701 individuals dividuals and resulted in the natural selection of thrifty
aged 23–32, average age is 28.9 years. At the last stage, the genes. Thus, by storing excess energy in the form of fat
study was completed with 8,834 individuals). Obese ado- when they reached the food, they continued to survive by
lescents have a higher tendency to become seriously obese using this stored fat during famine periods. Therefore, in-
in adulthood compared to normal body weight/slightly dividuals having thrifty genes will more likely to survive
overweight adolescents. Therefore, obesity in the adoles- times of food shortage than the others. However, the fact
cent period is significantly associated with severe obesity that individuals have these thrifty genes that seem to be
in adulthood [31]. According to the predictions made in a advantageous can turn into a disadvantage when they
systematic review and meta-analysis study based on BMI, it are chronically exposed to the “obesogenic environment”
was stated that approximately 55% of obese children characterized by a high-fat diet and a sedentary lifestyle.
would continue to be obese in the adolescent period and This may lead to increased susceptibility especially to
approximately 80% of obese adolescents would still be obesity and diabetes in individuals [39–41].
obese during adulthood. It was also highlighted that obese The most researched and documented population of
children and adolescents were five times more likely to thrifty genes is Arizona Pima Indians. While the life and
become obese in adulthood than nonobese children [33]. nutrition style of this population consisted of only enough
In a study on the role of genetic factors in obesity eti- food for themselves, an increase in body weight and dia-
ology, it was concluded that in twins who were raised betes rates are observed as a result of a modern obesogenic
together or separately, genetic factors constitute 30–70% lifestyle they have today [42, 43]. Type 2 diabetes mellitus is
of the change in BMI among individuals [34]. The compli- a common disease among Pima Indians and is strongly
ance rates between monozygotic twins for fat mass are associated with obesity. Continuous supply of food to in-
reported as ∼80 and ∼40% for dizygotic twins [34]. Pigeyre dividuals who develop thrifty genes can lead to obesity,
et al. (2016) indicated the heritability degree of BMI insulin resistance and diabetes in these individuals [42].
(interindividual variation rate that can be attributed to There is no shared agreement in the scientific literature
genetic factors) as 40–70%, while in a study conducted in about the thrifty genotype hypothesis. However, regardless
1997, the variation rate inherited factors responsible for of its evolutionary mechanism, it is accepted by many re-
adiposity was reported to be 30–50% [35, 36]. In a study searchers that genes that increase susceptibility to obesity
with monozygotic and dizygotic twins in China, genetics are associated with positive energy balance, which leads to
has been shown to play an increasingly important role in an increase in body weight by interacting with known
explaining body weight, height and BMI variation, espe- and unknown environmental factors [41]. Since the human
cially in boys, from early childhood to late adolescence population is genetically diverse and heterogeneous with
154 Yılmaz and Gezmen Karadağ: Role of genetic factors in adolescent obesity

more than 7.7 million single-nucleotide polymorphisms treatments [51]. Genome-wide association studies in the
(SNP), identifying potential gene–diet interactions related literature are regularly followed, updated and published.
to various diseases in humans, including obesity, is further According to the most recent publication, as of September
complicated by gene–gene interactions [3]. Gene–diet in- 2018, this database contains 5,687 GWAS, which includes
teractions need to be well understood to develop person- 71,673 variant-trait relationships from 3,567 publications
alized dietary recommendations. In a systematic review [52]. While these GWAS shed light on understanding dis-
examining diet and SNPs interactions, it has been reported ease susceptibility (through the identification of disease-
that there is no consistent evidence for total energy, car- causing genes and mechanisms), they have also led to new
bohydrate, and fat intake with SNPs [44]. Further clinical developments in clinical care developments (e.g. setting
studies are needed to understand the mechanisms under- new drug targets and disease markers) and in personalized
lying genetic variables and how they can interact with medicine (e.g. optimization of the sample, risk prediction
lifestyle and the environment. and genotype-based treatments) [53].
Recently, GWAS have identified a large number of
(more than 700) loci associated with adiposity [54]. The vast
Genome-wide association studies (GWAS) majority of these loci are defined for BMI (92% of loci) and
waist/hip ratio, which is adiposity-related markers [55].
The main purpose of human genetics is to identify genetic Yengo et al. (2018) determined 941 near-independent
risk factors for rare Mendelian diseases such as cystic genome-wide significant SNPs at 536 polygenic loci for
fibrosis as well as common diseases such as type 2 diabetes. BMI [56]. On the other hand, a more recent GWAS on 362,499
Many different technologies, study designs and analytical individuals from UK Biobank identified 98 independent loci,
tools are available to identify genetic risk factors [45]. It has 29 of which have not previously been associated with
long been known that the risk of complex disorders such as anthropometric traits [57]. These loci have been guiding in
cardiovascular diseases, type 2 diabetes or cancer is greatly defining the role of genetic factors in the pathophysiology of
influenced by the individual’s genetic history, but risk- human obesity [55]. Hundreds of genes were examined in
bearing genetic structures are not exactly known [46]. studies on candidate gene approaches before 2007, but few
With recently conducted genome-wide association of these genes were confirmed as genetic risk factors for
studies (GWAS), it is aimed to determine the genetic risk obesity. Four reports in 2007 (1-GWAS for anthropometric
factors for diseases common in populations, to measure features, 2-GWAS for early-onset severe obesity, 3-GWAS for
and analyze DNA sequence changes in the human genome type 2 diabetes, 4-fat mass and obesity-related gene-fat mass
[47]. Before the commencement of the GWAS, reference and obesity-associated gene study) explained that SNPs in
genome sequences were obtained through the Human the first intron of the fat mass and obesity-related gene FTO
Genome Project (1990–2003), one of the two major pro- were associated with obesity-related parameters [58–61].
jects. Although it was assumed that genetic information
related to diseases could be obtained easily in the first
years of this project, GWAS studies conducted in recent Genome-wide association studies in
years have offered a different perspective [48]. adolescent obesity
The primary purpose of GWAS is to use genetic risk
factors to make predictions about who is at risk, to develop There are some studies examining birth weight, BMI,
new prevention and treatment strategies, and to define the widespread obesity and severe early obesity in children,
biological basis of disease susceptibility [49]. Advocates of although most GWAS are performed in adults. In GWAS, in
GWAS studies argue that such studies will identify many which BMI and widespread obesity are examined in chil-
variants that contribute to common diseases, but the size of dren, many identical loci have been identified with GWAS
the data sets obtained will present significant analysis and in adults. Although it has some genetic similarities with
interpretation problems. Other researchers continue to early-onset severe obesity and birth weight BMI and
question the necessity and usefulness of this still expensive widespread obesity, it is largely guided by its unique loci
approach [50]. [62]. In children and adolescents, consistent effects have
The first GWA was conducted on age-related macular been demonstrated for variants in MC4R as well as FTO. In
degeneration, and the Complement Factor H gene was a study in which GWAS in children were analyzed, whether
identified as the major risk factor. This study played an 25 SNPs are associated with pediatric BMI in European-
important role in understanding the biological basis of Americans in 13 previously reported BMI locations (NEGR1,
genetic effects and developing new pharmacological SEC16B, TMEM18, INSIG2, SFRS10/ETV5/DGKG, GNPDA2,
 Yılmaz and Gezmen Karadağ: Role of genetic factors in adolescent obesity 155

Table : Single-nucleotide polymorphisms associated with obesity in adolescents.

Sample characteristics Major variables studied BMI, kg/m Main findings Author/
year

, overweight/obese, PP-TMEM (rs), ROPNL Individuals with *Polymorphisms in the Meng et al.,
 with normal body weight (rs), CDH (rs), MFAP- normal body weight: FTO (rs and 
Chinese individuals, between  GALNT (rs), MFAP-GALNT .–. rs) have a
and  years old (rs), FTO (rs and Individuals over- nominal effect on
rs), FERL (rs) weight/obese: BMI/obesity.
.–. *rs, rs,
rs and
rs have cumu-
lative effects on the risk
of obesity/overweight
European-American adolescent  SNPs related to BMI and the relation- In the European *The interaction Young
cohort, ship of these SNPs with each other were study; between PRKD-FTO et al., 
In the European study; , boys examined; Boys: . genes was found signif-
(mean age . years), , girls ADCY (rs), BDNF (rs), Girls: . icant.
(mean age . years), CADM (rs), ETV (rs), In the American *This study suggests
In the American study; , boys FAIM (rs), FANCL (rs), study; possible epistatic
(mean age . years), , girls FLJ (rs), FTO (rs), Boys: . effects on BMI in the
(mean age . years) GNPDA (rs), GPRCB Girls: . adolescent period. This
(rs), KCTD (rs), LMXB suggests potential
(rs), LRPB (rs), LRRNC interactions between
(rs), LZTR (rs), MAPK genes in biological
(rs), MCR (rs), MTCH pathways that are
(rs), MTIF (rs), important in obesity.
NCR_BAT (rs), NEGR
(rs), NRXN (rs), NUDT
(rs), POMC (rs), PRKD
(rs), PTBP (rs), QPCTL
(rs), RPLA (rs), SHB
(rs), SLCA (rs),
TFAPB (rs), TMEM (rs),
TNNIK (rs), ZCH (rs)
The study group included   SNPs associated with obesity at early The study group: Variants in the FTO gene Frayling
obese children and adolescents age; . have been reported to et al., 
(mean age . years); the control FTO (rs), FTO (rs), FTO The control group: contribute strongly to
group included  healthy in- (rs), FTO (rs), Corf . early-onset obesity.
dividuals in (mean age . years) (rs), SNP_A Among the  SNPs
(rs), FTO (rs), TSHR studied, the positive
(rs), BC (rs), FTO findings of FTO SNPs
(rs), rs, rs, represent.
PCSK (rs), HLA-DQA
(rs), none (rs)
, children (mean age .  SNPs, which are thought to increase Children; Nine of the variants den Hoed
years) and  adolescents (mean susceptibility to obesity, were studied. Boys: .±. studied were associated et al., 
age . years) rs, rs, rs, Girls: .±. with BMI (p<.), while
Individuals were randomly selected rs, rs, rs, Adolescents; the strongest relation-
from two different countries rs, rs, rs, Boys: .±. ship was found between
(Denmark and Estonia). rs, rs, rs, Girls: .±. the variant close to the
rs, rs, rs, TMEM gene and BMI
rs and rs (NEGR, (p=. × −).
SECB, LYPLAL, TMEM, ETV,
GNPDA, TFAPB, MSRA, BDNF, MTCH,
BCDIND, NRXN, SHB, FTO, MCR, and
KCTD loci, respectively)
156 Yılmaz and Gezmen Karadağ: Role of genetic factors in adolescent obesity

Table : (continued)

Sample characteristics Major variables studied BMI, kg/m Main findings Author/
year

 Caucasian adolescents aged PAX (rs), MRPS (rs), ND *PAX (p=. × −) Melka et al.,
between  and  years MCR (rs), MTCH (rs), was associated with to- 
FTO (rs), GPRCB (rs), tal body fat, while
MTIF (rs), NRXN (rs), MRPS (p=. × −),
LRPB (rs), SECB MCR (p=. × −),
(rs), SLCA (rs), FTO (. × −) and
ZNF (rs), NEGR (rs), MTCH (. × −) were
RPLA (rs), CADM found to be related to
(rs), ETV/DGKG (rs) BMI.
*PAX, MRPS and FTO
were also associated
with higher blood
pressure.
 ( boys,  girls) mexican NEGR (rs), SECB-RASAL Obese individuals: Polymorphisms in the Jimenez-
children and adolescent in- (rs), TMEM (rs), . (.–.) TMEM gene Osorio
dividuals aged – years TMEM (rs), IL-beta Nonobese in- (rs; et al., 
(rs), ADIPOQ (rs), dividuals: . rs) were asso-
GNPDA (rs), TNF-α (.–.) ciated with obesity
(rs), IL- (rs), IL- (p=.).
(rs), LEP (rs), MTCH
(rs), LGR-LINC-BDNF
(rs), BCDIND-FAIM (rs),
FTO (rs), MCR (rs),
MCR (rs), KCTD (rs)
, European adolescence and FTO (rs), TMEM (rs), Individuals with *This study identified Graff et al.,
young adults between ages – MCR (rs), TNNIK (rs), extreme values for seven independent loci 
years SECB (rs), GNPDA BMI are excluded. near FTO, TMEM,
(rs), POMC (rs), PRKD MCR, TNNIK, SECB,
(rs), CADM (rs), GNPDA and POMC.
SHB (rs) *Four loci (near PRKD,
TNNIK, SECB, and
CADM) had larger ef-
fects, and one locus
(near SHB) had a
smaller effect on BMI
during adolescence and
young adulthood
compared with older
adults.

ND, no data.

MTCH2, BDNF, FAIM2, SH2B1, FTO, MC4R and KCTD15) associated with BMI variation in Chinese children [64].
were examined. The relationship between BMI and 15 Some SNPs, especially FTO (rs1558902) MC4R (rs571312),
different variants, with the strongest association in the FTO TMEM18 (rs2867125), GNPDA2 (rs10938397), BDNF
gene, has been reported [63]. In another study, the effect of (rs10767664) and FAIM2 (rs7138803), are associated with
GWAS loci of all type 2 diabetes in Chinese children on BMI, obesity and body weight for both adults and children.
pediatric BMI was evaluated. Researchers have argued that On the other hand, different variants of the same genes
the HHEX-IDE gene is associated with an increase in BMI. (FTO, MC4R and BNDF) have been reported in both wide-
This gene has been emphasized since the increase in BMI in spread obesity and nonsyndromic obesity in children [41].
childhood affects the risk of type 2 diabetes in older ages. Obesity risk allele carriers are thought to have a greater
Besides, variants occurring in FAIM2, NPC1, FTO, MC4R, rate of change in BMI increase during adolescence and early
BDNF and GNPDA2 genes have been reported to be adulthood, and as the age increases some variables may
 Yılmaz and Gezmen Karadağ: Role of genetic factors in adolescent obesity 157

Table : Potential candidate genes and variants related to obesity in both adolescents and adults.

Variants’ names Potential candidate genes Risk Effect estimates p-Values Author
(BMI‐associated loci) alleles

Children and Per allele change in BMI β (s.e.m.) Speliotes

adults et al.,
rs FTO A . (.) . × − 
rs TMEM C . (.) . × −
rs MCR A . (.) . × −
rs GNPDA G . (.) . × −
rs BDNF A . (.) . × −
rs NEGR A . (.) . × −
rs SHB T . (.) . × −
rs ETV T . (.) . × −
rs MTCH T . (.) . × −
rs KCTD G . (.) . × −
rs RBJ C . (.) . × −
rs GPRCB C . (.) . × −
rs MAPK G . (.) . × −
rs QPCTL C . (.) . × −
rs TNNIK A . (.) . × −
rs SLCA T . (.) . × −
rs FLJ T . (.) . × −
rs LRRNC G . (.) . × −
rs TMEM A . (.) . × −
rs FANCL T . (.) . × −
rs CADM G . (.) . × −
rs PRKD T . (.) . × −
rs LRPB C . (.) . × −
rs PTBP C . (.) . × −
rs MTIF G . (.) . × −
rs ZNF A . (.) . × −
rs RPLA C . (.) . × −
rs NUDT G . (.) . × −
Children Effect allele frequency Graff
rs FTO A/T . . et al.,
rs MCR A/C . . 
rs SECB, LZTR G/A . .
rs TMEM C/T . .
rs TFAPB G/A . .
rs MTCH T/C . .
Childhooda Adultsa Childhood Adults Scherag
rs SDCCAG T . (.;.) . (.;.) . . et al.,
rs SDCCAG T . (.;.) . (.;.) . . 
rs SDCCAG C . (.;.) . (.;.) . .
rs TMEM T . (.;.) . (.;.) . × − . × −
rs TNKS/MSRA A . (.;.) . (.;.) . .
rs TNKS/MSRA A . (.;.) . (.;.) . .
rs TNKS/MSRA T . (.;.) . (.;.) . .
rs FTO A . (.;.) . (.;.) . × − . × −
rs FTO A . (.;.) . (.;.) . × − . × −
rs MCR A . (.;.) . (.;.) . × − .
158 Yılmaz and Gezmen Karadağ: Role of genetic factors in adolescent obesity

Table : (continued)

Variants’ names Potential candidate genes Risk Effect estimates p-Values Author
(BMI‐associated loci) alleles

Childhoodb Adultsb Childhood Adults Meyre

rs FTO C . (., . (., . × − . × − et al.,
.) .) 
rs MCR C . (., . (., . × − . × −
.) .)
rs NPC A −. (−., −. (−., . × − . × −
−.) −.)
rs MAF A . (., . (−., . × − . × −
.) .)
rs PTER C . (−., . (−., . × − . × −
.) .)
rs PRL A . (−., −. (−., . × − . × −
.) −.)
Childhood Adults Childhood Adults León-Mim-
rs ADIPOQ G . (.) . (.) . . ila
rs UCP T . (.) −. (.) . . et al., 
rs FTO A . (.) . (.) . .
rs TMEM C . (.) . (.) . .
rs INSIG G . (.) . (.) . .
rs FAIM/ BCDIN A . (.) . (.) . .
rs GNPDA G . (.) . (.) . .
rs MCR C . (.) −. (.) . .

Adolescents ND Zhao
rs FTO A . × − et al.,
rs MCR A . 
rs TNNIK A .
rs SECB G .
rs TMEM T . × −
rs GNPDA G .
rs BDNF T .
rs NRXN C , × −
rs QPCTL T .

BMI, body mass index; ND, no data. aOdds ratio for effect allele of (% CI). bPer allele change in BMI (% CI).

have stronger predictive effects during adolescence and In a GWA study conducted in 2013, seven independent
young adulthood due to increasing environmental effects, loci (p<5.0 × 10−8) were identified in adolescent and early
which may mask genetic effects. As the age and number of adulthood individuals; FTO (p=3.72 × 10−23), TMEM18
exposure to environmental effects will increase as you get (p=3.24 × 10−17), MC4R (p=4.41 × 10−17), TNNI3K
older, studies conducted in the early stages can give clearer (p=4.32 × 10−11), SEC16B (p=6.24 × 10−9), GNPDA2
results [65, 66]. Additionally, it is suggested that adolescent (p=1.11 × 10−8) and POMC (p=4.94 × 10−8) [67]. Besides, the
and early adulthood is a high-risk period in terms of body genetic effect of BMI loci changes in a complex way over time.
weight gain [67]. In a study on this subject, a positive cor- Therefore, it is important to investigate loci that affect obesity
relation was found between the variants in/near FTO, risk throughout life [68]. Table 2 summarizes the studies on
MC4R, MTCH2, TFAP2B, SEC16B and TMEM18 and the obesity and genetics in adolescents. Similar to adult studies,
change in BMI during adolescence and young adulthood the strongest relationship with obesity in adolescents was
[65]. Similarly, changes in BMI from childhood to adulthood found in SNPs in the FTO gene. The dominant SNPs may differ
and FTO and MC4R (two variants defining susceptibility in when ethnic groups change [60, 69–72]. Potential candidate
BMI in the early period) were investigated, and changes in genes and variants related to obesity in both adolescents and
BMI during adolescence and early adulthood were reported adults are shown in Table 3 [68, 73–77]. Even though similar
to offer stronger evidence than in late adulthood [66]. genes are associated with BMI, loci may vary according to
 Yılmaz and Gezmen Karadağ: Role of genetic factors in adolescent obesity 159

ethnicity and age groups. As the genetic data expand in the today. Although the etiology of obesity involves many
literature studies, it has been shown that, in addition to factors, the interaction of genetic factors with the envi-
basically known genes, new loci may also be associated with ronment has been one of the most striking issues recently.
obesity. On the other hand, the majority of BMI-associated Comorbid diseases associated with obesity affect adoles-
loci are shared between adolescents/childhood and adult cents both physically and mentally. Therefore, studies are
cohorts, but the index SNPs may slightly vary. The over- being conducted to attempt to develop new methods for the
lapped obesity susceptibility loci show a consistent direction prevention and treatment of obesity. Although there are
of effects on BMI. Given the current knowledge evidencing the significant improvements regarding the factors that cause
involvement of BMI-related loci in the regulation of energy obesity, there is not yet sufficient data on the etiology,
balance through the central nervous system, one may genetics and other factors influencing obesity. Genetic
postulate that similar mechanisms may be in play in early factors that play a role in the etiology of obesity and the
childhood obesity. Graff et al. (2017) investigated the obesity interaction of these factors with biological, environmental,
susceptibility variants in a cohort study. They found that and ecological factors may shed light on new de-
SNPs in/near FTO, MC4R, MTCH2, TFAP2B, SEC16B and velopments in obesity treatment. Understanding the
TMEM18 were significantly associated (p<0.0015≈0.05/34) mechanisms that explain obesity will be a significant
with BMI change in European Americans, while there are no step in the development of new therapeutic approaches.
significant SNPs after Bonferroni correction in African Although the FTO locus has been the most associated with
Americans and Hispanic Americans [68]. In a GWA study obesity among the loci identified so far, this locus can
among European children, four new loci associated with se- explain a very small part of the interindividual variations.
vere early-onset obesity were shown (rs564343 in phospho- Therefore, to identify related loci in adolescent obesity,
furin acidic cluster sorting protein 1; PACS1, rs11109072 in there is a need for large-scale research on a community
rhabdomyosarcoma 2 associated transcript; RMST, rs1957894 basis, which includes different ethnic groups due to the
in protein kinase C eta; PRKCH and rs11208659 in LEPR). The different variants among the populations as well as the
loci in the PACS1, RMST and PRKCH genes were studied in important role of age stratification in future genetic
Chinese children, and only the rs564343 in PACS1 gene was studies, like GWAS, should be considered.
found to be associated with a severe risk of obesity [78]. In a
review investigating the differences in the effect and Asians, Research funding: None declared.
but less important among those of African descent. However, Author contributions: All authors have accepted
in Asian children, this relationship has been found to be responsibility for the entire content of this manuscript
either the same or stronger than European children [79]. and approved its submission.
Genome-wide association studies have made significant Competing interests: Authors state no conflict of interest.
progress in identifying SNPs associated with BMI in children
and adolescents. It was shown that BMI is highly heritable,
especially in populations with higher average BMI, in coun- References
tries with a higher gross domestic product, and families of
lower socioeconomic status [80]. Known SNPs explain ∼6% 1. Martos-Moreno GÁ, Barrios V, Chowen JA, Argente J. Adipokines in
childhood obesity. Vitam Horm 2013;91:107–42.
of the variance of BMI, despite >50% evidence of inheritance
2. World Health Organization (WHO). Obesity. Available from:
from twin and family studies. This phenomenon is known as https://www.who.int/topics/obesity/en/ [Accessed 20 Jul 2020].
“missing heritability” [56, 81]. In a study using a novel 3. Garver WS. Gene-diet interactions in childhood obesity. Curr
method, genome-wide complex trait analysis, it was found Genom 2011;12:180–9.
that this analysis explains a significant proportion of the 4. Deurenberg P, Weststrate JA, Seidell JC. Body mass index as a
genetic impact defined as “missing heritability” [81]. How- measure of body fatness: age- and sex-specific prediction
formulas. Br J Nutr 1991;65:105–14.
ever, to better understand the effects of genetic factors in
5. Kumar S, Kelly AS. Review of childhood obesity: from
childhood obesity, there is a need for large-scale studies epidemiology, etiology, and comorbidities to clinical assessment
considering age stratification in different ethnic groups. and treatment. Mayo Clin Proc 2017;92:251–65.
6. World Health Organization Expert Committee Committe. Physical
status: the use of and interpretation of anthropometry. Geneva: World

Conclusion Health Organization; 1995. p. 854. WHO Technical Report Series.

7. Fernandez JR, Redden DT, Pietrobelli A, Allison DB. Waist
circumference percentiles in nationally representative samples of
Adolescent obesity has become a major public health African-American, European-American, and Mexican-American
problem in both developed and developing countries children and adolescents. J Pediatr 2004;145:439–44.
160 Yılmaz and Gezmen Karadağ: Role of genetic factors in adolescent obesity

8. Moreno LA, Rodríguez G, Guillén J, Rabanaque MJ, León JF, Ariño 25. Agha M, Agha R. The rising prevalence of obesity: part A: impact
A. Anthropometric measurements in both sides of the body in the on public health. Int J Surg Oncol 2017;2:e17.
assessment of nutritional status in prepubertal children. Eur J 26. Goryakin Y, Lobstein T, James WP, Suhrcke M. The impact of
Clin Nutr 2002;56:1208–15. economic, political and social globalization on overweight and
9. World Health Organization (WHO). Global strategy on diet, obesity in the 56 low and middle income countries. Soc Sci Med
physical activity and health. Available from: https://www.who. 2015;133:67–76.
int/dietphysicalactivity/childhood_what/en/ [Accessed 25 Jul 27. da Fonseca ACP, Mastronardi C, Johar A, Arcos-Burgos M,
2020]. Paz-Filho G. Genetics of non-syndromic childhood obesity
10. World Health Organization (WHO). Available from: https://www. and the use of high-throughput DNA sequencing
who.int/news-room/fact-sheets/detail/obesity-and-overweight technologies. J Diabetes Complications 2017;10:1549–61.
[Accessed 25 Jul 2020]. 28. Ranadive SA, Vaisse C. Lessons from extreme human obesity:
11. Bundak R, Furman A, Gunoz H, Darendeliler F, Bas F, Neyzi O. monogenic disorders. Endocrinol Metab Clin North Am 2008;3:
Body mass index references for Turkish children. Acta Paediatr 733-x.
2006;95:194–8. 29. Chung WK. An overview of mongenic and syndromic obesities in
12. Centers for Disease Control and Prevention (CDC). Defining humans. Pediatr Blood Cancer 2012;58:122–8.
childhood obesity. Available from: https://www.cdc.gov/ 30. Mason K, Page L, Balikcioglu PG. Screening for hormonal,
obesity/childhood/defining.html [Accessed 26 Jul 2020]. monogenic, and syndromic disorders in obese infants and
13. Kuczmarski RJ, Ogden CL, Guo SS, Grummer-Strawn LM, children. Pediatr Ann 2014;43:e218–24.
Flegal KM, Mei Z, et al. 2000 CDC Growth Charts for the United 31. The NS, Suchindran C, North KE, Popkin BM, Gordon-Larsen P.
States: methods and development. Vital Health Stat 2002;11: Association of adolescent obesity with risk of severe obesity in
1–190. adulthood. J Am Med Assoc 2010;304:2042–7.
14. Centers for Disease Control and Prevention (CDC). Childhood 32. Shoemaker AH, Chung ST, Fleischman A, Pediatric Endocrine
obesity facts. Available from: https://www.cdc.gov/obesity/ Society Obesity Special Interest Group. Trends in pediatric
data/childhood.html [Accessed 28 Jul 2020]. obesity management, a survey from the Pediatric Endocrine
15. de Onis M, Blössner M, Borghi E. Global prevalence and trends of Society Obesity Committee. J Pediatr Endocrinol Metab 2020;33:
overweight and obesity among preschool children. Am J Clin Nutr 469–72.
2010;92:1257–64. 33. Simmonds M, Llewellyn A, Owen CG, Woolacott N. Predicting
16. Ogden CL, Carroll MD, Lawman HG, Fryar CD, Kruszon-Moran D, adult obesity from childhood obesity: a systematic review and
Kit BK, et al. Trends in obesity prevalence among children and meta-analysis. Obes Rev 2016;17:95–107.
adolescents in the United States, 1988-1994 through 2013-2014. 34. Min J, Chiu DT, Wang Y. Variation in the heritability of body mass
J Am Med Assoc 2016;315:2292–9. index based on diverse twin studies: a systematic review. Obes
17. Skinner AC, Skelton JA. Prevalence and trends in obesity and Rev 2013;14:871–82.
severe obesity among children in the United States, 1999-2012. 35. Bouchard C. Genetic determinants of regional fat distribution.
JAMA Pediatr 2014;168:561–6. Hum Reprod 1997;12:1–5.
18. NCD-RisC. Trends in adult body-mass index in 200 countries from 36. Pigeyre M, Yazdi FT, Kaur Y, Meyre D. Recent progress in genetics,
1975 to 2014: a pooled analysis of 1698 population-based epigenetics and metagenomics unveils the pathophysiology of
measurement studies with 19·2 million participants. Lancet 2016; human obesity. Clin Sci 2016;130:943–86.
387:1377–96. 37. Liu Q, Yu C, Gao W, Cao W, Lyu J, Wang S, et al. Genetic and
19. Aktaş D, Öztürk FN, Kapan Y. Determination of the obesity environmental effects on weight, height, and BMI under 18 Years
prevalence and affecting risk factors, of eating habits among in a Chinese population-based twin sample. Twin Res Hum Genet
adolescents. TAF Prev Med Bull 2015;14:406–12. 2015;18:571–80.
20. Turkey Nutrition and Health Survey Report. Ankara/Turkey: 38. Silventoinen K, Rokholm B, Kaprio J, Sorensen TI. The genetic
Hacettepe University Faculty of Health Sciences Department of and environmental influences on childhood obesity: a
Nutrition and Dietetics; 2014. systematic review of twin and adoption studies. Int J Obes 2010;
21. Alper Z, Ercan İ, Uncu YA. Meta-analysis and an evaluation of 34:29–40.
trends in obesity prevalence among children and adolescents in 39. Neel JV. Diabetes mellitus: a “thrifty” genotype rendered
Turkey: 1990 through 2015. J Clin Res Pediatr Endocrinol 2018;10: detrimental by “progress”?. Am J Hum Genet 1962;14:353–62.
59–67. 40. Neel JV. The “thrifty genotype” in 1998. Nutr Rev 1999;57:S2–9.
22. Albuquerque D, Stice E, Rodriguez-Lopez R, Manco L, Nóbrega C. 41. Garver WS, Newman SB, Gonzales-Pacheco DM, Castillo JJ,
Current review of genetics of human obesity: from molecular Jelinek D, Heidenreich RA, et al. The genetics of childhood obesity
mechanisms to an evolutionary perspective. Mol Genet Genomics and interaction with dietary macronutrients. Genes Nutr 2013;8:
2015;290:1191–221. 271–87.
23. Akıncı A, Türkkahraman D, Tekedereli İ, Özer L, Evren B, Şahin İ, 42. Knowler WC, Pettitt DJ, Bennett PH, Williams RC. Diabetes
et al. Novel mutations in obesity-related genes in Turkish children mellitus in the Pima Indians: genetic and evolutionary
with non-syndromic early onset severe obesity: a multicentre considerations. Am J Phys Anthropol 1983;62:107–14.
study. J Clin Res Pediatr Endocrinol 2019;11:341–9. 43. Knowler WC, Pettitt DJ, Saad MF, Charles MA, Nelson RG, Howard
24. Hoedjes M, Makkes S, Halberstadt J, Noordam H, Renders CM, BV, et al. Obesity in the Pima Indians: its magnitude and
Bosmans JE, et al. Health-related quality of life in children and relationship with diabetes. Am J Clin Nutr 1991;53:1543S–51S.
adolescents with severe obesity after intensive lifestyle 44. Drabsch T, Gatzemeier J, Pfadenhauer L, Hauner H, Holzapfel C.
treatment and at 1-year follow-up. Obes Facts 2018;11:116–28. Associations between single nucleotide polymorphisms and total
 Yılmaz and Gezmen Karadağ: Role of genetic factors in adolescent obesity 161

energy, carbohydrate, and fat intakes: a systematic review. Adv 64. Zhao J, Bradfield JP, Zhang H, Annaiah K, Wang K, Kim CE, et al.
Nutr 2018;9:425–53. Examination of all type 2 diabetes GWAS loci reveals HHEX-IDE as
45. Bush WS, Moore JH. Chapter 11: genome-wide association a locus influencing pediatric BMI. Diabetes 2010;59:751–5.
studies. PLoS Comput Biol 2012;8:e1002822. 65. Hallman DM, Friedel VC, Eissa MA, Boerwinkle E, Huber JC Jr,
46. Dehghan A. Genome-wide association studies. Methods Mol Biol Harrist RB, et al. The association of variants in the FTO gene with
2018;1793:37–49. longitudinal body mass index profiles in non-Hispanic white
47. Gallagher MD, Chen-Plotkin AS. The post-GWAS era: from children and adolescents. Int J Obes 2012;36:61–8.
association to function. Am J Hum Genet 2018;102:717–30. 66. Hardy R, Wills AK, Wong A, Elks CE, Wareham NJ, Loos RJF, et al.
48. Gibson G. Hints of hidden heritability in GWAS. Nat Genet 2010; Life course variations in the associations between FTO and MC4R
42:558–60. gene variants and body size. Hum Mol Genet 2010;19:545–52.
49. Visscher PM, Wray NR, Zhang Q, Sklar P, McCarthy M, Brown MA, 67. Graff M, Ngwa JS, Workalemahu T, Homuth G, Schipf S, Teumer A,
et al. 10 Years of GWAS discovery: biology, function, and et al. Genome-wide analysis of BMI in adolescents and young
translation. Am J Hum Genet 2017;101:5–22. adults reveals additional insight into the effects of genetic loci
50. Hirschhorn JN, Daly MJ. Genome-wide association studies for over the life course. Hum Mol Gen 2013:3597–607. https://doi.
common diseases and complex traits. Nat Rev Genet 2005;6: org/10.1093/hmg/ddt205.
95–108. 68. Graff M, North KE, Richardson AS, Young KL, Mazul AL, Highland
51. Klein RJ, Zeiss C, Chew EY, Tsai JY, Sackler RS, Haynes C, et al. HM, et al. BMI loci and longitudinal BMI from adolescence to
Complement factor H polymorphism in age-related macular young adulthood in an ethnically diverse cohort. Int J Obes 2017;
degeneration. Science 2005;308:385–9. 41:759–68.
52. Buniello A, MacArthur JA L, Cerezo M, Harris LW, Hayhurst J, 69. Meng XR, Song JY, Ma J, Liu FH, Shang XR, Guo XJ, et al.
Malangone C, et al. The NHGRI-EBI GWAS Catalog of published Association study of childhood obesity with eight genetic
genome-wide association studies, targeted arrays and summary variants recently identified by genome-wide association studies.
statistics 2019. Nucleic Acids Res 2018;47:D1005–12. Pediatr Res 2014;76:310–5.
53. Tam V, Patel N, Turcotte M, Bossé Y, Paré G, Meyre D. Benefits and 70. den Hoed M, Ekelund U, Brage S, Grontved A, Zhao JH, Sharp SJ,
limitations of genome-wide association studies. Nat Rev Genet et al. Genetic susceptibility to obesity and related traits in
2019;20:467–84. childhood and adolescence: influence of loci identified by
54. Peters T, Nüllig L, Antel J, Naaresh R, Laabs BH, Tegeler L, et al. genome-wide association studies. Diabetes 2010;59:2980–8.
The role of genetic variation of BMI, body composition, and fat 71. Melka MG, Bernard M, Mahboubi A, Abrahamowicz M, Paterson
distribution for mental traits and disorders: a look-up and AD, Syme C, et al. Genome-wide scan for loci of adolescent
Mendelian randomization study. Front Genet 2020;11:373. obesity and their relationship with blood pressure. J Clin
55. Loos RJ. The genetics of adiposity. Curr Opin Genet Dev 2018;50: Endocrinol Metab 2012;97:E145–50.
86–95. 72. Young KL, Graff M, North KE, Richardson AS, Bradfield JP,
56. Yengo L, Sidorenko J, Kemper KE, Zheng Z, Wood AR, Weedon MN, Grant SFA, et al. Influence of SNP*SNP interaction on BMI in
et al. Meta-analysis of genome-wide association studies for European American adolescents: findings from the National
height and body mass index in ∼700000 individuals of European Longitudinal Study of Adolescent Health. Pediatr Obes 2016;
ancestry. Hum Mol Genet 2018;27:3641–9. 11:95–101.
57. Rask-Andersen M, Karlsson T, Ek WE, Johansson A. Genome- 73. León-Mimila P, Villamil-Ramírez H, Villalobos-Comparán M,
wide association study of body fat distribution identifies Villarreal-Molina T, Romero-Hidalgo S, López-Contreras B, et al.
adiposity loci and sex-specific genetic effects. Nat Commun Contribution of common genetic variants to obesity and obesity-
2019;10:339. related traits in mexican children and adults. PloS One 2013;8:
58. Dina C, Meyre D, Gallina S, Durand E, Körner A, Jacobson P, et al. e70640.
Variation in FTO contributes to childhood obesity and severe 74. Meyre D, Delplanque J, Chèvre JC, Lecoeur C, Lobbens S, Gallina
adult obesity. Nat Genet 2007;39:724–6. S, et al. Genome-wide association study for early-onset and
59. Frayling TM, Timpson NJ, Weedon MN, Zeggini E, Freathy RM, morbid adult obesity identifies three new risk loci in European
Lindgren CM, et al. A common variant in the FTO gene is populations. Nat Genet 2009;41:157–9.
associated with body mass index and predisposes to childhood 75. Scherag A, Dina C, Hinney A, Vatin V, Scherag S, Vogel CI, et al.
and adult obesity. Science 2007;316:889–94. Two new Loci for body-weight regulation identified in a joint
60. Hinney A, Nguyen TT, Scherag A, Friedel S, Brönner G, Müller TD, analysis of genome-wide association studies for early-onset
et al. Genome wide association study for early onset extreme extreme obesity in French and German study groups. PLoS Genet
obesity supports the role of fat mass and obesity associated gene 2010;6:e1000916.
(FTO) variants. PloS One 2007;2:e1361. 76. Speliotes EK, Willer CJ, Berndt SI, Monda KL, Thorleifsson G,
61. Scuteri A, Sanna S, Chen WM, Uda M, Albai G, Strait J, et al. Jackson AU, et al. Association analyses of 249,796 individuals
Genome-wide association scan shows genetic variants in the FTO reveal 18 new loci associated with body mass index. Nat Genet
gene are associated with obesity-related traits. PLoS Genet 2007; 2010;42:937–48.
3:e115. 77. Zhao J, Bradfield JP, Zhang H, Sleiman PM, Kim CE, Glessner JT,
62. Goodarzi MO. Genetics of obesity: what genetic association et al. Role of BMI-associated loci identified in GWAS meta-
studies have taught us about the biology of obesity and its analyses in the context of common childhood obesity in
complications. Lancet Diabetes Endocrinol 2018;6:223–36. European Americans. Obesity 2011;19:2436–9.
63. Zhao J, Grant SF. Genetics of childhood obesity. J Obes 2011;2011. 78. Song QY, Song JY, Wang Y, Wang S, Yang YD, Meng XR, et al.
845148. Association study of three gene polymorphisms recently
162 Yılmaz and Gezmen Karadağ: Role of genetic factors in adolescent obesity

identified by a genome-wide association study with obesity- 80. Schrempft S, van Jaarsveld CHM, Fisher A, Herle M, Smith AD,
related phenotypes in Chinese children. Obes facts 2017;10: Fildes A, et al. Variation in the heritability of child body mass
179–90. index by obesogenic home environment. JAMA Pediatr 2018;172:
79. Shinozaki K, Okuda M. The effects of fat mass and obesity- 1153–60.
associated gene variants on the body mass index among ethnic 81. Llewellyn CH, Trzaskowski M, Plomin R, Wardle J. Finding the
groups and in children and adults. Indian J Endocrinol Metab missing heritability in pediatric obesity: the contribution of
2012;16:S588–95. genome-wide complex trait analysis. Int J Obes 2013;37:1506–9.