Physiology & Behavior 152 (2015) 494–501

Contents lists available at ScienceDirect

Physiology & Behavior

journal homepage: www.elsevier.com/locate/phb

Review

Behavioral susceptibility to obesity: Gene–environment interplay in the

development of weight
Clare Llewellyn ⁎, Jane Wardle
Health Behavior Research Center, Department of Epidemiology and Public Health, University College London, Gower Street, London WC1E 6BT, United Kingdom

H I G H L I G H T S

• There is considerable evidence for both environmental and genetic causes of obesity.
• Appetite is hypothesized to mediate genetic susceptibility to the food environment.
• Prospective data suggest that appetitive traits play a causal role in weight gain.
• Genetic data have established that appetitive traits have a strong genetic basis.
• An appetitive model of obesity allows for both genetic and environmental influences.

a r t i c l e i n f o a b s t r a c t

Article history: There is considerable evidence for both environmental and genetic causes of obesity. Increased availability of cheap,
Received 31 March 2015 palatable food plays a role, but despite the ubiquity of the ‘obesogenic’ environment there is still substantial varia-
Received in revised form 2 July 2015 tion in weight — in fact, weight variability has gone up over recent decades. Twin and adoption studies show that
Accepted 6 July 2015
adiposity is highly heritable (50–90%), and genome-wide association studies have started to identify single nucleo-
Available online 10 July 2015
tide polymorphisms (SNPs) associated with weight. We have proposed that genetic susceptibility to obesity is partly
Keywords:
attributable to appetitive phenotypes, called the behavioral susceptibility theory (BST). BST proposes that individ-
Obesity uals who inherit a more avid appetite or lower sensitivity to satiety are more likely to overeat in response to the
Appetite food environment. Our laboratory has provided considerable evidence for BST using a variety of research ap-
Eating behavior proaches. We have used prospective epidemiological studies to demonstrate that appetite plays a causal role in
Genetic the development of weight, twin designs to show that appetitive phenotypes are highly heritable and have genetic
Twins overlap with adiposity, and genomic analyses to show that obesity-related SNPs are associated with appetite and
Behavior that appetite mediates some of the SNP–adiposity association. BST has helped to resolve the seeming paradox of
Children
both genetic determination and environmental determination of weight, and points to appetite as a useful target
for pharmacological and behavioral interventions in the prevention and management of obesity.
© 2015 Elsevier Inc. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495
2. The role of the environment in the development of obesity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495
3. The genetic basis of obesity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495
4. Gene–environment interplay in the development of obesity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 496
5. Appetite as the mediator of genetic susceptibility to the environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 496
5.1. Individual differences in appetite play a causal role in the development of obesity . . . . . . . . . . . . . . . . . . . . . . . . . . . 497
5.2. Individual differences in appetite have a genetic basis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 497
6. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 498
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 500
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 500

⁎ Corresponding author.
E-mail address: c.llewellyn@ucl.ac.uk (C. Llewellyn).

http://dx.doi.org/10.1016/j.physbeh.2015.07.006
0031-9384/© 2015 Elsevier Inc. All rights reserved.
 C. Llewellyn, J. Wardle / Physiology & Behavior 152 (2015) 494–501 495

1. Introduction have not been uniform. The thinnest people today are as thin as the
thinnest people 30 years ago, but the fatter ones are a great deal fatter
Obesity is one of the world's great health challenges, contributing to than they used to be [16,17], and proportionally, the largest increases
chronic disease and premature mortality, and burdening health ser- have been in rates of morbid obesity (e.g. Fig. 1). To put this into context,
vices. It is increasingly developing in childhood, and once developed, the prevalence of obesity today is just over double what it was in 1980
obesity is hard to reverse. The dramatic rise in obesity rates in recent (increasing from 15% to 36%), but rates of morbid obesity (BMI N 40)
years, in both adults and children, has caused considerable concern. A have increased more than four times (from 1.4% to 6.3%) [1,2].
century ago, no more than 1 in 20 of the US population were obese,
the figure had risen to more than 1 in 5 by 1988 [1], and now it's
more than 1 in 3 [2]. Combining overweight and obesity, 2 in 3 adults 3. The genetic basis of obesity
in the US carry too much body fat [2]. The problem is not confined to
adults either – rates of obesity have more than doubled for children, Research has established that genes play a major role in determining
and quadrupled for adolescents over the last 30 years – suggesting why people vary in weight, while all are exposed to the same
that children have not been buffered from the effects of the modern en- ‘obesogenic’ environment. Much of our understanding of the genetic
vironment. In children aged 6–11 years the percentage of those obese basis of human body weight has come from twin studies. Twins provide
has risen from 7% in 1980 to nearly 18% in 2012 [3,4]. Over the same pe- a natural experiment that makes it possible to estimate the relative in-
riod, the percentage of obese adolescents increased from 5% to nearly fluence of genes and environment to variation in any trait, by comparing
21%. In total in 2012 more than 1 in 3 children and adolescents were resemblance between identical and non-identical twins [18]. Greater
overweight or obese [3,4]. similarity between identical twins indicates a genetic contribution to
Until recently the prevailing view was that childhood weight was BMI, denoted by the statistic heritability which indicates the extent to
uninformative about adult weight, partly due to the rarity of childhood which individual differences in a trait are explained by genetic variation.
obesity (few obese adults had been obese children), and in light of the Twin-estimated heritability of BMI is consistently high — genes ac-
fact that many ‘chubby’ toddlers matured into slim children. However, count for well over 50%, and sometimes as much as 90%, of individual
studies that have followed adiposity trajectories using continuous rath- differences in BMI [19]. There have also been studies of identical twins
er than categorical measures have found strong tracking of weight from who were adopted apart from one another at birth, and their weights
early childhood [5,6], and obese children are far more likely to be obese show almost exactly the same pattern of twin resemblance despite
adults than their normal weight peers [7,8]. Furthermore, in recent being brought up in different households [20]. However, it is notewor-
years there has been a strong interest in the importance of very early thy that the heritability estimates vary considerably from sample to
life experiences, and it is now clear that overweight and obesity in child- sample, suggesting that BMI can be influenced by other factors, some
hood are strongly predicted by rapid weight gain in early postnatal life of which are discussed in more detail below. A finding that has surprised
[9–11], suggesting that causal processes begin soon after birth. This many researchers is the high heritability of body weight in both infancy
has highlighted the importance of understanding the etiology of obesity and childhood. Although birth weight tends only to show moderate her-
from very early in life. itability (in the order of 30%) [21], genetic influences on weight increase
Understand the etiology of obesity is complex because there is substantially soon after birth. By 6 months of age the heritability of in-
considerable evidence for both genetic and environmental influences. fant weight is as high as 62% [22], by 1 year of age it has been estimated
The behavioral susceptibility theory (BST) of obesity hypothesizes that to be over 80% [23], and by middle childhood (8 years of age) it is as high
‘obesity genes’ influence weight partly through their effects on appetite as 90% [24]. Another striking observation from twin data has been the
and that variation in appetite emerges early in postnatal life. In our lab- low impact of the shared environment on childhood weight (any aspect
oratory, we have been developing an evidence base to support BST, of the environment that the two twins share in common that contrib-
focusing in particular on infants and children; this review largely sum- utes to their similarity, such as food availability in the home and paren-
marizes this research. tal feeding policies). There is a moderate influence of the shared
environment during early childhood (in the order of 10–30%), but by
2. The role of the environment in the development of obesity late childhood it has diminished substantially and by adolescence the
effect has disappeared [24]. These observations highlight the impor-
Environmental changes are widely agreed to account for the rapid tance of genetic variation in determining adiposity, even during child-
increases in weight over the last 40 years. The main culprit is believed hood when the home family environment has the opportunity to play
to be the changes in lifestyle and the food supply, creating what is an important role.
often called an ‘obesogenic’ environment. Developments in food pro- Genome-wide association studies (GWAS) have now identified 97
duction, processing, storage and preparation have resulted in highly- genetic variants (single nucleotide polymorphisms, SNPs) that are ro-
palatable and energy-dense foods becoming more accessible and bustly associated with BMI [25]. The first to be discovered was the fat
cheaper [12]. Portion sizes have gone up — even in recipe books [13]. mass and obesity-associated gene (FTO). Of all the obesity-associated
Foods are also heavily marketed nowadays, often with incentives for SNPs, FTO has the largest effect size. The 16% of adults who carry two
buying larger quantities [14]. On the energy expenditure side, there copies of the high risk version of the FTO SNP on both alleles are approx-
have been changes in transportation and mechanization which have re- imately 3 kg (6.6 lb) heavier than the 46% carrying two copies of the low
duced the amount of physical activity that we are required to do in ev- risk allele. FTO is also associated with childhood weight by 5 years of age
eryday life, as well as increases in the range of sedentary screen-based [26]. The full 97 SNPs can be combined into a polygenic obesity risk
entertainments, which also reduce activity levels [15], although physical score that shows a quantitative association with body weight. However,
activity as a leisure time pursuit has increased. The increased opportuni- to date, the known genetic variants collectively explain only a small
ties to eat, and the incentive structures of the current food environment fraction (2.7%) of variation in BMI, a problem that has been called ‘miss-
encourage overconsumption, so many individuals end up in positive en- ing heritability’. However, the situation is not unique to weight; height
ergy balance (energy-intake is greater than energy-expenditure), which is also highly heritable, but it has proved hard to find the specific SNPs
leads to weight gain. involved [27]. The current view is that there are thousands of SNPs,
However, despite the ubiquity of the ‘obesogenic’ environment, not each contributing a tiny amount to overall genetic risk, but with such
everyone is obese. Even within families, it is not uncommon for one small effect sizes that they cannot be detected even in huge datasets
child to be lean while his or her sibling is overweight. Notably, the aver- used in contemporary GWAS. This is partly because GWAS need to ad-
age increases in weight seen over the course of the obesity epidemic just for upwards of 1 million statistical tests (as many SNPs as there
496 C. Llewellyn, J. Wardle / Physiology & Behavior 152 (2015) 494–501

Fig. 1. Data from 10,506 women from the Health Survey for England showing unequal increases in BMI from 1993–4 to 2002–3 for UK women at differential risk of obesity. Slim women in
2002–3 were almost as slim as their counterparts in 1993–4; however, women at the upper end of the distribution in 2002–3 were considerably heavier than their counterparts in 1993–4.

are available on the genotyping array that is used), and so a stringent intensified. This has led researchers to explore the mechanisms through
significance threshold is set (typically p b .5 × 10−8) [28]. which the ‘obesogenic’ environment might elicit genetic effects on
weight.
4. Gene–environment interplay in the development of obesity
5. Appetite as the mediator of genetic susceptibility to the
The strong genetic basis of human body weight seems to be para- environment
doxical in light of the evidence for the importance of the environment,
as well as the fact that the gene pool cannot be changed over the short We have been developing an appetitive model of obesity – so-called
time period during which the obesity epidemic has developed. Much behavioral susceptibility theory (BST) [33] – that attempts to resolve the
of the confusion can be resolved if we consider that environmentalists apparent paradox of high genetic and environmental influences on
study mean change in population weights over time, whereas genetic human body weight. The two important aspects to appetite are: eating
researchers have been more interested in understanding individual dif- onset (responsiveness to signals to start eating) and eating offset (re-
ferences in weight gain over time. And as highlighted in Section 2, while sponsiveness to signals to stop eating). On the onset side, people who
it is true that mean population weights have increased over the course are more responsive to food cues (smell, sight and taste) are more likely
of the obesity epidemic, increases in weight have not been uniform to eat in response to tempting opportunities. On the offset side, people
across the weight spectrum but have been much greater at the upper who have weaker internal signaling from their gut once they start eat-
end of the distribution than in the middle [16,17]. One theory that ge- ing, or are less aware of the signals (sensitivity to satiety) are likely to
netic researchers have put forward to explain this phenomenon is that carry on eating for longer once they have started. BST proposes that ge-
the effect of obesity risk genes depends on environmental exposure, netic risk operates through these two appetitive traits, which confer dif-
which implies gene–environment interaction — i.e. individuals who ferential susceptibility to the food environment. That is, obesity genes
are most genetically susceptible to the effects of the ‘obesogenic’ envi- ‘work’ by determining how responsive each of us is to the opportunities
ronment have gained the most weight. To put it another way, the to eat. As the food environment becomes more ‘permissive’ (increased
more ‘obesogenic’ the environment, the stronger the genetic effect on opportunity to eat) genes for high food responsiveness or low satiety re-
body weight. sponsiveness are better able to be expressed in terms of behavior, pro-
This idea has been supported by both twin and molecular genetic viding greater explanatory power for both genetic and environmental
data. A large Swedish study of around 2000 identical and non- factors. This helps explain how genetic influences have got stronger in
identical twin pairs and 115,000 siblings born between 1951 and 1983 line with increasing power of the ‘obesogenic’ environment. George
showed that genetic variance in BMI at age 18 was higher for those Bray put it well when he said: “Genes load the gun, the environment
born later, and by implication living in a more ‘obesogenic’ environ- pulls the trigger” [34].
ment. A recent meta-analysis of the heritability of BMI derived from We have used data from two population-based pediatric birth co-
twin studies concluded that estimates are higher in countries with horts of twins to test some of the assumptions of BST — namely, that ap-
higher gross domestic product (again, by inference have a more petite plays a causal role in the development of obesity, and that
‘obesogenic’ environment), and in samples with a higher average BMI appetite itself has a genetic basis. We have focused on infants and chil-
[29]. We, along with others have shown that the heritability of BMI in- dren for three reasons: (1) infancy has been identified as a key period
creases steadily from early childhood to late adolescence, which could during which obesity risk begins, and genetic influence on weight is
reflect increasing exposure to the ‘obesogenic’ environment [29,30]. highly established within the few months of life suggesting that the
The findings from twin studies have been supported by molecular ge- mechanisms through which ‘obesity genes’ influence weight begin
netic studies, with evidence that the effects of FTO [31] and a polygenic soon after birth; (2) infants and young children are not actively modify-
obesity risk score [32] on BMI are significantly stronger for those born ing their eating behavior in order to manage their weight, unlike some
later, suggesting that genetic effects are moderated by environmental older children and adults; and (3) infants and young children have
exposure. Collectively, these findings provide evidence that as the envi- not had long-standing obesity which itself can cause aberrations in
ronment has become more ‘obesogenic’, genetic effects on weight have appetite.
 C. Llewellyn, J. Wardle / Physiology & Behavior 152 (2015) 494–501 497

Gemini — Health and Development in Twins is a prospective birth when they were 10 years old, and for a community sample of 572 3–
cohort of 2404 families with identical and non-identical twins born in 5 year olds, to demonstrate that greater FR and lower SR are traits that
England and Wales in 2007 [35]; it is the largest pediatric twin cohort characterize greater adiposity [33]. In subsequent analyses we com-
set up specifically to assess early childhood growth. The Twins Early De- pared the appetite scores of a group of severely obese 8–13 year old Brit-
velopment Study (TEDS) is a British cohort of more than 16,000 families ish children currently in treatment for ‘simple’ obesity (no identifiable
with twins born in 1994–1996 [36]. Both Gemini and TEDS are broadly medical cause for their obesity), with those of a sample of 7–12 year
comparable to the general population in terms of distribution across the old children, which included a subsample from TEDS. These analyses
country, and with national averages for twins in sex, zygosity (propor- allowed us to show that these traits do not simply distinguish the clin-
tion of identical and non-identical pairs), gestational age and birth- ical from the non-clinical (as Schachter had showed in early studies
weight; all children participating in TEDS and Gemini are twins. The with adults), but that they are linearly related to weight across the
prospective element of these cohorts has been used to examine the whole spectrum — with progressively higher FR and lower SR from
likelihood that appetite plays a causal role in the development of thin, to normal-weight, to overweight, to obese children [42] (Fig. 2).
weight, and the twin design has been exploited to quantify the extent A wealth of other studies have demonstrated the same associations in
to which individual differences in these traits are explained by genetic pediatric samples from infancy to adolescence providing strong evi-
variation. DNA has also been collected which has enabled us to establish dence that appetite is associated with weight in childhood [43–48].
associations between these appetitive characteristics with known The majority of these studies have been cross-sectional, making it
obesity-related genetic variants. The evidence for BST provided by impossible to be sure about the causal direction — increases in adiposity
these two cohorts is described in more detail below. may drive increases in appetite. Recently, we used longitudinal data
from 2213 infants in Gemini (one of the two twin cohorts described in
5.1. Individual differences in appetite play a causal role in the development Section 5) to test the causal direction. We used structural equation
of obesity modeling to demonstrate that the prospective associations between
SR or FR at 3 months and weight at 9 months were significantly stronger
An appetitive model of obesity was first proposed by Stanley than those between weight at 9 months and SR or FR at 15 months,
Schachter in a seminal paper in Science in 1968 [37]. He demonstrated when tested simultaneously [49]. This suggested that appetitive traits
that obese adults did not show the same compensatory down- are more likely to drive early weight gain than the other way around.
regulation of food intake following a high-calorie snack as normal Furthermore, it indicated that variations in these appetitive characteris-
weight controls; in other words, they were less sensitive to satiety. tics are present from the first few months of life, even before any solid
Obese adults also ate more ice-cream than normal weight adults food has been introduced, and contribute to the development of adipos-
when the ice-cream that they were offered was highly palatable, but ity in early infancy.
the groups did not differ in their intake of ‘not-so-nice’ ice-cream; in The twin design also made it possible to test the robustness of the
other words, they were more responsive to food cues. Obese adults appetite-weight association by using a discordant-sibling analysis
also ate more nuts if they were shelled than if they had to do the shelling which controls for potential shared environmental confounders. We
themselves, indicating greater responsiveness to opportunity. demonstrated that dizygotic twin pairs who differed by at least 1 stan-
Since the 1960s a wealth of evidence has emerged in support of an dard deviation in FR (n = 121 pairs) or SR (n = 172 pairs) at 3 months,
appetitive model of obesity. Early behavioral studies gave rise to a rich had divergent growth trajectories from 3 to 15 months. The twin who
research base comparing overweight/obese and normal weight adults, was more food responsive or less satiety sensitive gained significantly
which eventually extended to pediatric samples. The drawback of be- more weight over the 12 month period; by 15 months of age there
havioral research is that it is time-consuming and costly, so it cannot was nearly a 1 kg difference in weight, representing approximately
be used in large population-based samples that are needed to detect 10% of average weight at that age [50] (Fig. 3). Together, these findings
small effects and demonstrate generalizability. ‘Self-conscious’ over- strongly support the idea that variation in appetite in the first few
weight adults may also modify their eating behavior when being ob- months of life plays a causal role in weight gain, and predisposes to
served by researchers. In addition, a single eating episode may not obesity.
provide a reliable representation of an underlying appetitive trait due
to other factors at play during only one observation period. 5.2. Individual differences in appetite have a genetic basis
In order to study appetite more easily, and in the large numbers
needed for genetic analyses, we developed a questionnaire measure of Another key hypothesis of BST is that individual differences in appe-
appetite called the Child Eating Behaviour Questionnaire (CEBQ) [38]. titive characteristics have a genetic basis and that genes indirectly influ-
The CEBQ is a psychometric measure of a range of appetitive character- ence body weight partly through their direct effects on appetite. Data
istics including food responsiveness and satiety sensitivity designed for from Gemini and TEDS have established that the heritability of appeti-
pediatric samples on which parents respond to items describing a range tive characteristics in both infancy and childhood is high, in keeping
of eating behaviors, using a 5-point Likert scale (‘never, ‘rarely’, ‘some- with the size of the estimates observed for body weight itself. Using
times’, ‘often’, ‘always’). Parent report is of course subjective and has the CEBQ in 5435 pairs of 10 year old twins from TEDS, we demonstrat-
the potential for bias, but parents also know their child's behavior, and ed that the majority of individual differences in responsiveness to food
therefore can respond in general terms about the child's tendency to cues and satiety responsiveness are determined by genes (75% and
demonstrate these appetitive traits across many meals and situations 63% respectively) [51]. Similarly high heritability of appetite was ob-
rather than relying on the single indicator that a test meal provides. served at 3 months measured using the BEBQ in 2402 pairs of twins
The CEBQ has been validated using behavioral measures [39], and is sta- from Gemini: FR (59%) and SR (72%) [52].
ble over time [40]. An infant version, the Baby Eating Behaviour Ques- Twins can also be used to quantify the extent to which the genes that
tionnaire (BEBQ) [41] has also been developed to assess the same influence one trait (e.g. BMI) are the same as the genes that influence
appetitive characteristics during the earliest period of life when infants another trait (e.g. appetite), denoted by a statistic called the genetic cor-
are exclusively milk-fed. The availability of these measures has led to relation. Using this method with Gemini data we were able to demon-
the emergence of a large pediatric literature exploring ‘food responsive- strate genetic correlations between 0.22 and 0.37 for appetite and
ness’ (FR) and ‘satiety responsiveness’ (SR), and allowed relationships weight at 3 months of age. These data suggest that between a quarter
with weight to be elucidated in large population-based samples. and a third of the genetic effects on appetite are the same as those on
We collected information on appetite using the CEBQ for 10,364 chil- weight, further supporting the idea that obesity genes work through ef-
dren from TEDS (one of the two twin cohorts described in Section 5) fects on appetite [53].
498 C. Llewellyn, J. Wardle / Physiology & Behavior 152 (2015) 494–501

Fig. 2. Mean scores for the ‘food responsiveness’ (on the left) and ‘satiety responsiveness’ (on the right) scales of the Children's Eating Behaviour Questionnaire (CEBQ) by weight group
show a linear association between these appetitive characteristics and weight.

The discovery of FTO provided a unique opportunity to test BST with SNPs bestow obesity risk. Many of these explorations pointed towards
a common genetic variant that is robustly associated with BMI in adults an appetite pathway as well. Gene expression analyses revealed that
and children. 3337 children from TEDS were genotyped for the common 27 out of 31 significantly enriched tissues were in the central nervous
SNP in the FTO gene, to explore associations with SR; children with two system, and in particular in the hypothalamus and pituitary gland —
versions of the high risk allele of FTO (AA) had significantly reduced SR, key sites in the central regulation of appetite. 405 genes that were in
and this appetitive trait significantly mediated part of the association close proximity to, and correlated with the 97 SNPs were categorized
between FTO status and adiposity [54]. into 25 biological groups, of which the largest comprised genes involved
The discovery of many more SNPs associated with obesity has made in neuronal processes including monogenic obesity genes involved in
it possible to quantify genetic risk of obesity in the form of a polygenic the hypothalamic melanocortin appetite pathway. Lastly, many of the
obesity risk score that reflects the number of obesity-risk increasing al- identified SNPs themselves sit in or near genes known to regulate appe-
leles carried. We created polygenic obesity risk scores for TEDS children tite centrally (e.g. FTO, MC4R, TMEM18, KCTD15, SH2B1, GNPDA2, NEGR1,
based on 28 known obesity-related variants. Using these scores we BDNF and POMC) or peripherally (e.g. GIPR, a receptor of gastric inhibi-
were able to show that higher genetic risk of obesity was associated tory polypeptide), and in genes involved in monogenic forms of obesity
with lower SR, and as before, SR mediated some of the association be- known to be caused by disruption to the hypothalamic melanocortin
tween the genetic risk score and weight [55] (Fig. 4). The association appetite pathway (e.g. MC4R, BDNF, BBS4, SH2B, NEGR1 and POMC).
remained after exclusion of FTO, suggesting that the other obesity- These analyses suggest that the biological pathways of a number of
related genes also influence adiposity partly through their direct effects these loci are likely to be appetitive [25].
on appetite.
The most recent meta-analysis of GWAS that identified 97 BMI- 6. Conclusion
associated SNPs carried out a series of explorations to identify biologi-
cally relevant tissues (those with evidence of increased gene expres- There is now considerable evidence for the behavioral susceptibility
sion) and gene sets (those highly correlated with the 97 SNPs, theory (BST) of obesity. Prospective analyses suggest that from the first
r2 N 0.2) to shed light on the likely mechanisms through which the few months of life, appetite influences the development of weight [49,

Fig. 3. Growth trajectories for dizygotic sibling pairs who are discordant for ‘food responsiveness’ (figure on the left) and ‘satiety responsiveness’ (figure on the right). The blue curve shows
the growth of the twin with the higher ‘food responsiveness’ score (left) and the higher ‘satiety responsiveness score’ (right). The yellow curve shows the growth of the twin with the lower
‘food responsiveness’ score (left) and the lower ‘satiety responsiveness’ score (right). The 95% confidence intervals are indicated by the dashed lines.
 C. Llewellyn, J. Wardle / Physiology & Behavior 152 (2015) 494–501 499

Fig. 4. Association between a polygenic obesity risk score comprising 28 known obesity-associated genetic variants, with adiposity (figure on the right) and satiety responsiveness (figure
on the left). The number of obesity-risk increasing alleles is normally distributed. As genetic risk of obesity increases, BMI and waist circumference increase (figure on the left), but satiety
responsiveness reduces.

50]. Twin studies have supported the hypothesis that individual differ- and parenting styles) on child weight. Contrary to common beliefs
ences in appetitive traits have a strong genetic basis [51,52], and molec- about the causes of childhood obesity, the influence of the home family
ular genetic studies indicate that obesity-associated SNPs influence environment appears to be very low. It is clear that children who differ
weight partly through effects on appetite [54,55]. BST helps to address genetically (e.g. non-identical twins) often respond quite differently to
the seeming paradox of high genetic and environmental determination the opportunities offered by the same family environment — the twin
of weight based on the dependence of genetically driven risk behaviors with the more avid appetite is more likely to overeat at home and
on the environmental opportunities to express such tendencies. It also gain more weight than his or her co-twin, by virtue of his or her
helps explain why the population variability in weight has increased genes. Within-family variation in these appetitive dispositions begins
as the environment has become more ‘obesogenic’. Under conditions to be expressed in early postnatal life during the period of exclusive
of famine no one is obese, regardless of genetic susceptibility to obesity, milk-feeding, even before any solid food has been introduced.
but under conditions of abundance (like the modern food environment Evidence that genetic risk for obesity operates at least partly through
in many Western countries) highly genetically susceptible individuals appetitive mechanisms suggests that pharmacological or behavioral in-
become obese unless they are able to exert high levels of self-control, terventions that target over-responsiveness to food cues or under-
while those with low genetic susceptibility remain lean (Fig. 5). sensitivity to satiety are likely to be important. Strategies such as careful
Twin studies have also highlighted the overriding importance of ge- portion control and slow eating are already used to circumvent poor sa-
netic factors versus those of the shared environment (aspects of the en- tiety sensitivity, and there may be other opportunities to blunt respon-
vironment shared by two twins, including the home food environment siveness to food cues, such as attention control or self-regulation

Fig. 5. A hypothetical demonstration of the percentage of children who are obese under three different environmental conditions, according to whether they have a low, average or high
genetic susceptibility to obesity. Under conditions of famine there would be no obese children, regardless of genetic susceptibility to obesity; when the food supply is limited slightly more
of the children at high genetic susceptibility would be obese than those at average susceptibility, but none of the children at low genetic susceptibility would be obese; under conditions of
abundance (like the modern food environment in the US) the majority of children at high genetic susceptibility would be obese, a considerable number at average genetic susceptibility,
but very few children at low genetic susceptibility would be obese.
500 C. Llewellyn, J. Wardle / Physiology & Behavior 152 (2015) 494–501

training. As personalized medicine gains traction, genetic information [22] L. Johnson, C.H. Llewellyn, C.H. van Jaarsveld, T.J. Cole, J. Wardle, Genetic and
environmental influences on infant growth: prospective analysis of the Gemini
about individual risk may generate even more precise targeting, and twin birth cohort, PLoS One 6 (2011) e19918.
personalized prevention advice could result in better adherence. [23] K. Silventoinen, K.H. Pietilainen, P. Tynelius, T.I.A. Sorensen, J. Kaprio, F. Rasmussen,
Genetic and environmental factors in relative weight from birth to age 18: the
Swedish young male twins study, Int. J. Obes. 31 (2007) 615–621.
Acknowledgments [24] K. Silventoinen, B. Rokholm, J. Kaprio, T.I. Sorensen, The genetic and environmental
influences on childhood obesity: a systematic review of twin and adoption studies,
Int. J. Obes. 34 (2010) 29–40.
This manuscript is based on work presented during the 2014 [25] A.E. Locke, B. Kahali, S.I. Berndt, A.E. Justice, T.H. Pers, F.R. Day, C. Powell, S.
Annual Meeting of the Society for the Study of Ingestive Behavior, July Vedantam, M.L. Buchkovich, J. Yang, D.C. Croteau-Chonka, T. Esko, T. Fall, T. Ferreira,
30th, 2014. I would like to thank the program committee for the invita- S. Gustafsson, Z. Kutalik, J. Luan, R. Magi, J.C. Randall, T.W. Winkler, A.R. Wood, T.
Workalemahu, J.D. Faul, J.A. Smith, Z.J. Hua, W. Zhao, J. Chen, R. Fehrmann, A.K.
tion to share our data on this research topic. The Twins Early Develop- Hedman, J. Karjalainen, E.M. Schmidt, D. Absher, N. Amin, D. Anderson, M. Beekman,
ment Study (TEDS) is supported by a grant from the UK Medical J.L. Bolton, J.L. Bragg-Gresham, S. Buyske, A. Demirkan, G. Deng, G.B. Ehret, B.
Research Council (G0901245; and previously G0500079), with addi- Feenstra, M.F. Feitosa, K. Fischer, A. Goel, J. Gong, A.U. Jackson, S. Kanoni, M.E. Kleber,
K. Kristiansson, U. Lim, V. Lotay, M. Mangino, L.I. Mateo, C. Medina-Gomez, S.E.
tional support from the US National Institutes of Health (HD044454; Medland, M.A. Nalls, C.D. Palmer, D. Pasko, S. Pechlivanis, M.J. Peters, I. Prokopenko,
HD059215); the anthropometric data were collected as part of a grant D. Shungin, A. Stancakova, R.J. Strawbridge, S.Y. Ju, T. Tanaka, A. Teumer, S. Trompet,
from the UK Biological and Biotechnology Research Council (D19086) S.W. van der Laan, J. van Setten, J.V. van Vliet-Ostaptchouk, Z. Wang, L. Yengo, W.
Zhang, A. Isaacs, E. Albrecht, J. Arnlov, G.M. Arscott, A.P. Attwood, S. Bandinelli, A.
with additional support from program grant funding from Cancer Barrett, I.N. Bas, C. Bellis, A.J. Bennett, C. Berne, R. Blagieva, M. Bluher, S. Bohringer,
Research UK; the genome-wide genotyping was funded by the L.L. Bonnycastle, Y. Bottcher, H.A. Boyd, M. Bruinenberg, I.H. Caspersen, Y.D. Ida
Wellcome Trust Case Control Consortium 2 project (085475/Z/08/Z). Chen, R. Clarke, E.W. Daw, A.J. de Craen, G. Delgado, M. Dimitriou, A.S. Doney, N.
Eklund, K. Estrada, E. Eury, L. Folkersen, R.M. Fraser, M.E. Garcia, F. Geller, V.
The Gemini study was funded by a grant from Cancer Research UK Giedraitis, B. Gigante, A.S. Go, A. Golay, A.H. Goodall, S.D. Gordon, M. Gorski, H.J.
(C1418/A7974). Grabe, H. Grallert, T.B. Grammer, J. Grassler, H. Gronberg, C.J. Groves, G. Gusto, J.
Haessler, P. Hall, T. Haller, G. Hallmans, C.A. Hartman, M. Hassinen, C. Hayward,
N.L. Heard-Costa, Q. Helmer, C. Hengstenberg, O. Holmen, J.J. Hottenga, A.L. James,
References J.M. Jeff, A. Johansson, J. Jolley, T. Juliusdottir, L. Kinnunen, W. Koenig, M. Koskenvuo,
W. Kratzer, J. Laitinen, C. Lamina, K. Leander, N.R. Lee, P. Lichtner, L. Lind, J.
[1] C.L. Ogden, M.D. Carroll, Prevalence of overweight, obesity, and extreme obesity Lindstrom, L.K. Sin, S. Lobbens, R. Lorbeer, Y. Lu, F. Mach, P.K. Magnusson, A.
among adults: United States, trends 1960–1962 through 2007–2008, National Cen- Mahajan, W.L. McArdle, S. McLachlan, C. Menni, S. Merger, E. Mihailov, L. Milani,
ter for Health Statistics 2010, pp. p1–p6. A. Moayyeri, K.L. Monda, M.A. Morken, A. Mulas, G. Muller, M. Muller-Nurasyid,
[2] K.M. Flegal, M.D. Carroll, B.K. Kit, C.L. Ogden, Prevalence of obesity and trends in the A.W. Musk, R. Nagaraja, M.M. Nothen, I.M. Nolte, S. Pilz, N.W. Rayner, F. Renstrom,
distribution of body mass index among US adults, 1999–2010, JAMA 307 (2012) R. Rettig, J.S. Ried, S. Ripke, N.R. Robertson, L.M. Rose, S. Sanna, H. Scharnagl, S.
491–497. Scholtens, F.R. Schumacher, W.R. Scott, T. Seufferlein, J. Shi, S.A. Vernon, J.
[3] C.L. Ogden, M.D. Carroll, B.K. Kit, K.M. Flegal, Prevalence of childhood and adult Smolonska, A.V. Stanton, V. Steinthorsdottir, K. Stirrups, H.M. Stringham, J.
obesity in the United States, 2011–2012, JAMA 311 (2014) 806–814. Sundstrom, M.A. Swertz, A.J. Swift, A.C. Syvanen, S.T. Tan, B.O. Tayo, B. Thorand, G.
[4] National Center for Health Statistics, Health, United States, 2011: With Special Thorleifsson, J.P. Tyrer, H.W. Uh, L. Vandenput, F.C. Verhulst, S.H. Vermeulen, N.
Features on Socioeconomic Status and Health, US Department of Health and Verweij, J.M. Vonk, L.L. Waite, H.R. Warren, D. Waterworth, M.N. Weedon, L.R.
Human Services, Hyattsville, MD, 2012. Wilkens, C. Willenborg, T. Wilsgaard, M.K. Wojczynski, A. Wong, A.F. Wright, Q.
[5] F. Trudeau, R.J. Shephard, S. Bouchard, L. Laurencelle, BMI in the Trois-Rivieres Zhang, E.P. Brennan, M. Choi, Z. Dastani, A.W. Drong, P. Eriksson, A. Franco-
study: child–adult and child–parent relationships, Am. J. Hum. Biol. 15 (2003) Cereceda, J.R. Gadin, A.G. Gharavi, M.E. Goddard, R.E. Handsaker, J. Huang, F.
187–191. Karpe, S. Kathiresan, S. Keildson, K. Kiryluk, M. Kubo, J.Y. Lee, L. Liang, R.P. Lifton,
[6] J.K. Lake, C. Power, T.J. Cole, Child to adult body mass index in the 1958 British birth B. Ma, S.A. McCarroll, A.J. McKnight, J.L. Min, M.F. Moffatt, G.W. Montgomery, J.M.
cohort: associations with parental obesity, Arch. Dis. Child. 77 (1997) 376–381. Murabito, G. Nicholson, D.R. Nyholt, Y. Okada, J.R. Perry, R. Dorajoo, E. Reinmaa,
[7] K.M. McTigue, J.M. Garrett, B.M. Popkin, The natural history of the development of R.M. Salem, N. Sandholm, R.A. Scott, Genetic studies of body mass index yield new
obesity in a cohort of young U.S. adults between 1981 and 1998, Ann. Intern. Med. insights for obesity biology, Nature 518 (2015) 197–206.
136 (2002) 857–864. [26] U. Sovio, D.O. Mook-Kanamori, N.M. Warrington, R. Lawrence, L. Briollais, C.N.
[8] R.M. Fuentes, I.L. Notkola, S. Shemeikka, J. Tuomilehto, A. Nissinen, Tracking of body Palmer, J. Cecil, J.K. Sandling, A.C. Syvanen, M. Kaakinen, L.J. Beilin, I.Y. Millwood,
mass index during childhood: a 15-year prospective population-based family study A.J. Bennett, J. Laitinen, A. Pouta, J. Molitor, S.G. Davey, Y. Ben-Shlomo, V.W.
in eastern Finland, Int. J. Obes. Relat. Metab. Disord. 27 (2003) 716–721. Jaddoe, L.J. Palmer, C.E. Pennell, T.J. Cole, M.I. McCarthy, M.R. Jarvelin, N.J.
[9] K.K. Ong, R.J. Loos, Rapid infancy weight gain and subsequent obesity: systematic Timpson, Association between common variation at the FTO locus and changes in
reviews and hopeful suggestions, Acta Paediatr. 95 (2006) 904–908. body mass index from infancy to late childhood: the complex nature of genetic
[10] U. Ekelund, K.K. Ong, Y. Linne, M. Neovius, S. Brage, D.B. Dunger, N.J. Wareham, S. association through growth and development, PLoS Genet. 7 (2011) e1001307.
Rossner, Association of weight gain in infancy and early childhood with metabolic [27] J. Hebebrand, A.L. Volckmar, N. Knoll, A. Hinney, Chipping away the ‘missing herita-
risk in young adults, J. Clin. Endocrinol. Metab. 92 (2007) 98–103. bility’: GIANT steps forward in the molecular elucidation of obesity — but still lots to
[11] J. Baird, D. Fisher, P. Lucas, J. Kleijnen, H. Roberts, C. Law, Being big or growing fast: go, Obes. Facts 3 (2010) 294–303.
systematic review of size and growth in infancy and later obesity, BMJ 331 (2005) [28] G.S. Barsh, G.P. Copenhaver, G. Gibson, S.M. Williams, Guidelines for genome-wide
929. association studies, PLoS Genet. 8 (2012) e1002812.
[12] B.A. Swinburn, G. Sacks, K.D. Hall, K. McPherson, D.T. Finegood, M.L. Moodie, S.L. [29] J. Min, D.T. Chiu, Y. Wang, Variation in the heritability of body mass index based on
Gortmaker, The global obesity pandemic: shaped by global drivers and local envi- diverse twin studies: a systematic review, Obes. Rev. 14 (2013) 871–882.
ronments, Lancet 378 (2011) 804–814. [30] C.H. Llewellyn, M. Trzaskowski, R. Plomin, J. Wardle, From modeling to measure-
[13] C. Piernas, B.M. Popkin, Food portion patterns and trends among U.S. children and ment: developmental trends in genetic influence on adiposity in childhood, Obesity
the relationship to total eating occasion size, 1977–2006, J. Nutr. 141 (2011) (Silver Spring) 22 (2014) 1756–1761.
1159–1164. [31] J.N. Rosenquist, S.F. Lehrer, A.J. O'Malley, A.M. Zaslavsky, J.W. Smoller, N.A.
[14] N. Zlatevska, C. Dubelaar, S.S. Holden, Sizing up the effect of portion size on con- Christakis, Cohort of birth modifies the association between FTO genotype and
sumption: a meta-analytic review, J. Mark. 78 (2014) 140–154. BMI, Proc. Natl. Acad. Sci. U. S. A. (2014).
[15] R.C. Brownson, T.K. Boehmer, D.A. Luke, Declining rates of physical activity in the [32] E.W. Demerath, A.C. Choh, W. Johnson, J.E. Curran, M. Lee, C. Bellis, T.D. Dyer, S.A.
United States: what are the contributors? Annu. Rev. Public Health 26 (2005) Czerwinski, J. Blangero, B. Towne, The positive association of obesity variants with
421–443. adulthood adiposity strengthens over an 80-year period: a gene-by-birth year
[16] C.L. Ogden, S.Z. Yanovski, M.D. Carroll, K.M. Flegal, The epidemiology of obesity, interaction, Hum. Hered. 75 (2013) 175–185.
Gastroenterology 132 (2007) 2087–2102. [33] S. Carnell, J. Wardle, Appetite and adiposity in children: evidence for a behavioral
[17] J. Wardle, D. Boniface, Changes in the distributions of body mass index and waist cir- susceptibility theory of obesity, Am. J. Clin. Nutr. 88 (2008) 22–29.
cumference in English adults, 1993/1994 to 2002/2003, Int. J. Obes. 32 (2008) [34] G.A. Bray, Leptin and leptinomania, Lancet 348 (1996) 140–141.
527–532. [35] C.H. van Jaarsveld, L. Johnson, C. Llewellyn, J. Wardle, Gemini: a UK twin birth cohort
[18] R. Plomin, J.C. DeFries, V.S. Knopik, J.M. Neiderhiser, Behavioral Genetics, 6th ed. with a focus on early childhood weight trajectories, appetite and the family environ-
Worth Publishers, New York, 2012. ment, Twin Res. Hum. Genet. 13 (2010) 72–78.
[19] C.E. Elks, M. den Hoed, J.H. Zhao, S.J. Sharp, N.J. Wareham, R.J. Loos, K.K. Ong, Vari- [36] C.M. Haworth, O.S. Davis, R. Plomin, Twins Early Development Study (TEDS): a
ability in the heritability of body mass index: a systematic review and meta- genetically sensitive investigation of cognitive and behavioral development from
regression, Front Endocrinol. (Lausanne) 3 (2012) 29. childhood to young adulthood, Twin Res. Hum. Genet. 1–9 (2012).
[20] C.M. Grilo, M.F. Pogue-Geile, The nature of environmental influences on weight and [37] S. Schachter, Obesity and eating. Internal and external cues differentially affect the
obesity: a behavior genetic analysis, Psychol. Bull. 110 (1991) 520–537. eating behavior of obese and normal subjects, Science 161 (1968) 751–756.
[21] C. Bouchard, Childhood obesity: are genetic differences involved? Am J Clin Nutr 89 [38] J. Wardle, C.A. Guthrie, S. Sanderson, L. Rapoport, Development of the Children's
(2009) 1494S–1501S. Eating Behaviour Questionnaire, J. Child Psychol. Psychiatry 42 (2001) 963–970.
 C. Llewellyn, J. Wardle / Physiology & Behavior 152 (2015) 494–501 501

[39] S. Carnell, J. Wardle, Measuring behavioural susceptibility to obesity: validation of [48] L. Webber, C. Hill, J. Saxton, C.H. van Jaarsveld, J. Wardle, Eating behaviour and
the Child Eating Behaviour Questionnaire, Appetite 48 (2007) 104–113. weight in children, Int. J. Obes. 33 (2009) 21–28.
[40] J. Ashcroft, C. Semmler, S. Carnell, C.H. van Jaarsveld, J. Wardle, Continuity and sta- [49] C.H. van Jaarsveld, C.H. Llewellyn, L. Johnson, J. Wardle, Prospective associations be-
bility of eating behaviour traits in children, Eur. J. Clin. Nutr. 62 (2008) 985–990. tween appetitive traits and weight gain in infancy, Am. J. Clin. Nutr. 94 (2011)
[41] C.H. Llewellyn, C.H. van Jaarsveld, L. Johnson, S. Carnell, J. Wardle, Development and 1562–1567.
factor structure of the Baby Eating Behaviour Questionnaire in the Gemini birth co- [50] C.H. van Jaarsveld, D. Boniface, C.H. Llewellyn, J. Wardle, Appetite and growth: a lon-
hort, Appetite 57 (2011) 388–396. gitudinal sibling analysis, JAMA Pediatr. 168 (2014) 345–350.
[42] H. Croker, L. Cooke, J. Wardle, Appetitive behaviours of children attending obesity [51] S. Carnell, C.M. Haworth, R. Plomin, J. Wardle, Genetic influence on appetite in
treatment, Appetite 57 (2011) 525–529. children, Int. J. Obes. 32 (2008) 1468–1473.
[43] D.L. Jahnke, P.A. Warschburger, Familial transmission of eating behaviors in [52] C.H. Llewellyn, C.H. van Jaarsveld, L. Johnson, S. Carnell, J. Wardle, Nature and
preschool-aged children, Obesity (Silver Spring) 16 (2008) 1821–1825. nurture in infant appetite: analysis of the Gemini twin birth cohort, Am J Clin
[44] K.N. Parkinson, R.F. Drewett, A.S. Le Couteur, A.J. Adamson, Do maternal ratings of Nutr 91 (2010) 1172–1179.
appetite in infants predict later Child Eating Behaviour Questionnaire scores and [53] C.H. Llewellyn, C.H. van Jaarsveld, R. Plomin, A. Fisher, J. Wardle, Inherited behavior-
body mass index? Appetite 54 (2010) 186–190. al susceptibility to adiposity in infancy: a multivariate genetic analysis of appetite
[45] E.F. Sleddens, S.P. Kremers, C. Thijs, The Children's Eating Behaviour Questionnaire: and weight in the Gemini birth cohort, Am. J. Clin. Nutr. 95 (2012) 633–639.
factorial validity and association with body mass index in Dutch children aged 6–7, [54] J. Wardle, S. Carnell, C.M. Haworth, I.S. Farooqi, S. O'Rahilly, R. Plomin, Obesity
Int. J. Behav. Nutr. Phys. Act. 5 (2008) 49. associated genetic variation in FTO is associated with diminished satiety, J. Clin.
[46] J.C. Spence, V. Carson, L. Casey, N. Boule, Examining behavioural susceptibility to Endocrinol. Metab. 93 (2008) 3640–3643.
obesity among Canadian pre-school children: the role of eating behaviours, Int. J. [55] C.H. Llewellyn, M. Trzaskowski, C.H. van Jaarsveld, R. Plomin, J. Wardle, Satiety
Pediatr. Obes. 6 (2010) e501–e507. mechanisms in genetic risk of obesity, JAMA Pediatr. 168 (2014) 338–344.
[47] V. Viana, S. Sinde, J.C. Saxton, Children's Eating Behaviour Questionnaire: associa-
tions with BMI in Portuguese children, Br. J Nutr 100 (2008) 445–450.