Cerebral Cortex, April 2020;30: 2519–2528

doi: 10.1093/cercor/bhz257
Advance Access Publication Date: 24 October 2019
Original Article

ORIGINAL ARTICLE

Childhood Obesity, Cortical Structure, and Executive

Function in Healthy Children

Downloaded from https://academic.oup.com/cercor/article/30/4/2519/5603326 by guest on 29 May 2023

Lisa Ronan1,*, Aaron Alexander-Bloch2 and Paul C. Fletcher1,3,4
1 Department of Psychiatry, University of Cambridge, Cambridge CB2 8HA UK, 2 Department of Child and
Adolescent Psychiatry and Behavioral Science, Children’s Hospital of Philadelphia, PA 19104, USA,
3 Cambridgeshire and Peterborough NHS Foundation Trust, Cambridge CB21 5EF, UK, and 4 The Wellcome-MRC

Institute of Metabolic Science-Metabolic Research Laboratories (IMS-MRL), University of Cambridge,

Cambridge CB2 0QQ, UK
Address correspondence to Lisa Ronan, Department of Psychiatry, University of Cambridge, Douglas House, 18b Trumpington Road, Cambridge CB2 8HA
UK. Email: [email protected]

Abstract
The development of executive function is linked to maturation of prefrontal cortex (PFC) in childhood. Childhood obesity
has been associated with changes in brain structure, particularly in PFC, as well as deficits in executive functions. We aimed
to determine whether differences in cortical structure mediate the relationship between executive function and childhood
obesity. We analyzed MR-derived measures of cortical thickness for 2700 children between the ages of 9 and 11 years,
recruited as part of the NIH Adolescent Brain and Cognitive Development (ABCD) study. We related our findings to
measures of executive function and body mass index (BMI). In our analysis, increased BMI was associated with significantly
reduced mean cortical thickness, as well as specific bilateral reduced cortical thickness in prefrontal cortical regions. This
relationship remained after accounting for age, sex, race, parental education, household income, birth-weight, and
in-scanner motion. Increased BMI was also associated with lower executive function. Reduced thickness in the rostral
medial and superior frontal cortex, the inferior frontal gyrus, and the lateral orbitofrontal cortex partially accounted for
reductions in executive function. These results suggest that childhood obesity is associated with compromised executive
function. This relationship may be partly explained by BMI-associated reduced cortical thickness in the PFC.

Key words: ABCD, childhood obesity, cortical thickness, executive function, prefrontal cortex

Like adult obesity (Graham et al. 2014; Yang et al. 2018),

Introduction childhood obesity has been linked to impairments in executive
Although the rise in incidence of childhood obesity appears to functioning (Maayan et al. 2011; Reinert et al. 2013; Yau et al.
have plateaued in some developed nations, the condition is still 2014; Ross et al. 2015; Alarcón et al. 2016; Li et al. 2018), although
estimated to effect 124 million children worldwide (NCD-RisC studies produce conflicting results (Gunstad et al. 2008). Exec-
2017) or an estimated 1 in 3 children in the US (Ogden et al. utive function is an umbrella term for several different cogni-
2014). Children who are overweight or obese are more likely tive dimensions, including inhibitory control, decision-making,
to become obese adults, and have an increased risk of poorer working memory, and reward sensitivity, broadly referring to a
health outcomes in later life including diabetes, heart disease, set of processes that enable planning, problem-solving, flexible
cancer, and overall mortality (Biro and Wien 2010). reasoning, and regulation of behaviors and emotions. Children

© The Author(s) 2019. Published by Oxford University Press.

This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/),
which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
 2520 Cerebral Cortex, 2020, Vol. 30, No. 4

who are overweight or obese generally score lower on various (Sowell et al. 2004; Shaw et al. 2006), the consolidation of
measures of executive function (Liang et al. 2014), including regional activity (Durston et al. 2006), and the emergence of more
working memory (Riggs et al. 2012), reward-sensitivity (Verbeken comprehensive and extensive network connections (Ezekiel et
et al. 2012) and inhibitory control (Guerrieri et al. 2008; Verbeken al. 2013). An important question therefore is whether childhood
et al. 2009). obesity is characterized by structural changes in cortical regions
At a biological level, various hypotheses exist relating execu- important for executive function at this critical developmental
tive functions to body mass index (BMI). One prominent theory period and further, whether these structural changes mediate
is that the role of executive function in planning and decision- the link between BMI and differences in executive function
making, response inhibition, and reward evaluation influences observed in childhood.
food intake (Gluck et al. 2017), contributing to increased BMI. We sought to address this question using data from the NIH
Prospective studies of bariatric patients following surgery sup- Adolescent Brain and Cognitive Development (ABCD) dataset
port this association (Spitznagel et al. 2013), as do neuroimag- (Jernigan and Brown 2018) of n = 2700 children between the
ing investigations. For example, functional magnetic resonance ages of 9 and 11 years. Specifically we related measures of
imaging studies have demonstrated that the dorsolateral pre- cortical thickness to measures of executive function and BMI,

Downloaded from https://academic.oup.com/cercor/article/30/4/2519/5603326 by guest on 29 May 2023

frontal cortex (DPFC) is differentially activated in obesity in both and further examined whether cortical thickness confounds the
adults (Le et al. 2006) and children (Davids et al. 2010; Reinert et observed relationship between these traits. We chose to focus
al. 2013). This region is critical to cognitive control over eating on measures of cortical thickness as cortical thinning in child-
(Gluck et al. 2017), as well as reward-motivated behavior via hood has been linked to the emergence of executive function
its links to the mesolimbic and mesocortical regions of the (Kharitonova et al. 2013; Bathelt et al. 2018), cortical thickness
brain. Other regions of the prefrontal cortex (PFC) such as the changes have been linked to childhood obesity (Maayan et al.
orbitofrontal cortex (OFC) have also been linked to obesity. The 2011; Reinert et al. 2013; Alosco et al. 2014b; Yau et al. 2014; Ross
OFC is involved in inhibition and reward processing (Gehring et al. 2015; Medic et al. 2016), and previous studies in adulthood
and Willoughby 2002) and has been shown to be differentially have demonstrated that cortical thickness in the PFC mediates
activated in lean and obese people depending on the level of the relationship between executive function and BMI (Lavagnino
satiety (Del Parigi 2010), including in children and adolescents et al. 2016).
(Holsen et al. 2005; Batterink et al. 2010; Bruce et al. 2010; Stice
et al. 2010). This region is also implicated in the response to
food stimuli (Killgore and Yurgelun-Todd 2005) in both children
Methods and Materials
(Holsen et al. 2005) and adults (Kringelbach 2005). The ventral Subjects
lateral PFC, linked to impulsivity, has also been implicated in
A total of 3923 children aged 9–11 years from the ABCD dataset
obesity (Batterink et al. 2010). More generally, children with lower
were initially included (10.15154/1504466). The ABCD dataset is
levels of executive abilities are more likely to be sedentary and
a longitudinal study of over 10 000 children recruited from 21
have higher rates of snack consumption (Liang et al. 2014), and
centers throughout the US, with participants largely recruited
are less likely to benefit from weight-loss interventions (Eichen
through the school system. Sampling plans and recruitment
et al. 2018). While these studies support a link between execu-
procedures based on considerations of age, gender, race, socio-
tive function and BMI, the direction of causality is unclear. For
economic status, and urbanicity were designed to reflect the
example, there is compelling evidence suggesting that the low-
sociodemographics of the US. Details of recruitment and study
grade inflammatory response that characterizes obesity may
design are described elsewhere (Garavan et al. 2018). Details
have a causal impact on the brain and impair executive function
of demographic, physical, and mental health assessments are
(Shields et al. 2017; Yang et al. 2018). Supporting this causal
described elsewhere (Barch et al. 2018).
pathway, studies of postsurgical executive function in bariatric
BMI was based on measures of height and weight, which were
patients have demonstrated that weight loss is correlated with
taken as the average of up to 3 separate measures. BMI was
an improvement in cognitive abilities (Alosco et al. 2014a).
calculated as weight in lbs divided by height in inches squared,
In summary, while the direction of causal association
multiplied by 703 (Eq. (1)).
between BMI and executive function is not well understood,
neuroimaging studies support the hypothesis that cortical  
structure and function are important to characterizing the weight lbs
BMI = 703 ×  2 (1)
relationship between executive function and BMI. Various height in
investigations have linked adolescent obesity with changes
in grey matter volume, connectivity, and reduced cortical
BMI z-scores (BMIz ) were defined using lookup tables from
thickness, commonly in prefrontal regions known to be
the Center of Disease Control 2001 (CDC Growth Charts 2018),
associated with executive function (Maayan et al. 2011; Alosco
where BMI was adjusted for sex and age. Subjects with a
et al. 2014b; Yau et al. 2014; Gupta et al. 2015; Ross et al. 2015). In
diagnosis of ADHD (n = 536), autism spectrum disorder (n = 49),
addition, high fat diets have been linked to changes in microRNA
schizophrenia (n = 2), intellectual disability (n = 2), and diabetes
expression related to axonal guidance in the PFC of adolescents
(n = 9) were excluded from analysis.
(Labouesse et al. 2018). Thus, there is a demonstrable association
Additional analyses were carried out using measures of waist
in adolescents between BMI and cortical structure in regions
circumference and waist-to-height ratio in place of BMI.
associated with executive function. Whether such structural
changes mediate the relationship between BMI and executive
abilities is not established. Late childhood and early adolescence Imaging Protocols
is a critical period for the emergence and consolidation of Imaging protocols for the ABCD dataset are described elsewhere
executive function, which is strongly linked to maturation of (Casey et al. 2018), and were harmonized for three 3T scanner
the PFC (Gray et al. 2003; Frangou et al. 2004; Tamnes et al. 2013). platforms (Siemens Prisma, General Electric 750 and Philips)
This maturation is characterized by reduced cortical thickness used across the 21 data acquisition sites.
 Childhood BMI, Brain Structure, and Executive Function Ronan et al. 2521

Cortical Reconstruction and Brain Structural Measures linked to changes in cortical thickness between the ages of 4
and 13 years (Nguyen et al. 2013). To account for the possible
Cortical reconstructions were carried out using FreeSurfer v5.3.0
confounding effects of the age of onset of puberty, we included
(Dale et al. 1999; Fischl et al. 1999a, 1999b), as part of the initial
salivary DHEA levels (Uban et al. 2018) as a covariate in our
baseline processing of the ABCD dataset. Reconstructions were
analysis. Birth weight was also included as a nuisance variable,
visually inspected for quality control purposes. Only those
as studies have indicated that this may play a role in intelligence
reconstructions deemed of sufficient quality were included
scores at 11 years (Korpela et al. 2018) and has been demon-
in this study. Based on these surface reconstructions, cortical
strated to be significantly predictive of childhood obesity (Biro
thickness (Fischl and Dale 2000) values were processed for
and Wien 2010; Glavin et al. 2014). We also included estimates of
the Deskian-Killiany atlas (Desikan et al. 2006), with data
head movement during scanning. Such micromotions have been
unavailable for 3/36 regions per hemisphere (omitted regions
demonstrated to be genetically correlated with BMI (Hodgson et
included “unknown”, “corpus callosum”, and “insula”). Derived
al. 2017) and associated with biases in MR-derived parameters
results per individual per region were provided as part of the
of cortical structure (Alexander-Bloch et al. 2016). For these
ABCD curated annual release 1.0 (DOI 10.15154/1412097).
reasons, frame-wise displacement (FWD) derived from resting-

Downloaded from https://academic.oup.com/cercor/article/30/4/2519/5603326 by guest on 29 May 2023

state data was adopted as an estimate of average head motion
Executive Function and included as a covariate. We also included brain volume and
Participants involved in the ABCD study participated in a battery a self-reported measure of physical activity, which was recorded
of tests designed to test their executive function. An overview as the number of days in the week prior to interview where the
of the baseline neurocognition battery is described elsewhere subject had been moderately physically active for more than
(Luciana et al. 2018). Tests were based on the National Institute of 60 min. Finally, we included covariates of household income,
Health (NIH) Toolbox (https://nihtoolbox.desk.com). A compos- race, and parental education in our analysis as these have been
ite score of executive function was generated based on results of demonstrated to be associated with BMI (Strauss and Knight
several tests, namely the Flanker inhibitory control and atten- 1999).
tion test, the dimensional change card sort test, the picture
sequence memory test, the list sorting working memory test,
and the pattern comparison processing speed test (Akshoomoff
Results
et al. 2013, 2014, 2018). The age-corrected standard scores for Our results support statistically significant associations between
each test were based on a normative sample of 2917 children BMI and executive function, between BMI and cortical thick-
and adolescents (Casaletto et al. 2015). The composite score ness, and between cortical thickness and executive function.
was derived by averaging the standard scores of each of the Individuals with higher BMI tend to have lower scores on
measures and then deriving standard scores based on this new executive function tests and thinner cerebral cortices, while
distribution. These age-corrected composite scores were used individuals with thinner cerebral cortex tend to have lower
in subsequent analysis. Full data were available for n = 2352 scores on executive function tests. The association between
subjects (n = 1802, 352 and 702 for lean, overweight and obese, BMI and executive function may be mediated by their shared
respectively). relationship with the thickness of a subset of regions in PFC.

Statistical Analysis Demographic Variables, BMI, and Executive Function

We conducted a mediation analysis to determine whether corti- As expected from prior literature, many demographic and bio-
cal thickness mediated the relationship between BMI and exec- logical variables were related to BMI and executive function,
utive function. As part of this analysis, we determined the fol- supporting their inclusion as covariates in our statistical model.
lowing relationships using multivariate methods: regression of There was no association between BMIZ and age; however,
executive function on BMI, regression of regional cortical thick- males were significantly heavier than females (β = 0.1, t = 2.4,
ness on BMI, and regression of executive function on cortical P = 0.02) (see Table 1). Birth weight was significantly associated
thickness. Subsequently, we determined whether regional corti- with BMIZ (β =0.1, t = 5.9, P < 0.001), as was household income
cal thickness (mediator) was a significant predictor of executive (F(2, 2387) = 52, P < 0.001), race (F(3, 3286) = 49, P < 0.001), and level
function (the dependent variable) in a model that also included of parental education (F(2, 2387) = 21, P < 0.001) (see Table 1). In
BMI (the independent variable) (Baron and Kenny 1986). We line with previous analysis, in-scanner motion was positively
used Mahalanobis distance to identify and remove outliers in all associated with BMIZ (β = 0.16, t = 7.6, P < 0.001), while self-
regression models, and false discovery rate (FDR) methods (Ben- reported levels of physical activity were negatively associated
jamini and Hochberg 1995) were used to correct cortical results with increasing BMIz (β = −0.07, t = 3, P = 0.001). There was no
for multiple comparisons. Standardized regression coefficients association between BMIZ and total brain volume; however, BMIZ
were reported. was positively associated with salivary DHEA levels (β =0.17, t = 6,
Mediation was conducted using the “mediation” package in P < 0.001), suggesting that increased BMI was associated with
R (Tingley et al. 2014). We assessed the significance of our medi- more advanced pubertal stages.
ation models using bootstrapping methods to increase power Executive abilities were significantly associated with age
(Hayes 2009), with 1000 bootstrap samples used to generate 95% (β = 0.09, t = 4.5, P < 0.001) and slightly higher in females (β = 0.1,
confidence intervals for the indirect effect. All analyses were t = 2.4, P = 0.02). Birth weight was not associated with executive
conducted in R (v.3.3.3). abilities; however, total brain volume (β = 0.07, t = 3.3, P < 0.001),
in-scanner motion (β = 0.2, t = 7.5, P < 0.001), physical activity
Covariates (β = 0.05, t = 2.2, P = 0.03), parental education (F(2, 2387) = 37,
Puberty is known to influence brain development, and pubertal P < 0.001), household income (F(2, 2387) = 46, P < 0.001), and race
hormones such as dehydroepiandrosterone (DHEA) have been (F(2, 2386) = 27, P < 0.001) were associated. Overall, there was also
 2522 Cerebral Cortex, 2020, Vol. 30, No. 4

Table 1 Demographics and variables by BMI class

Underweight (<5th) Lean (5th—85th) Overweight (85th—95th) Obese (>95th)

N 127 2197 472 501

Age (months) 122 120 120 120
Sex (F/M) 77/50 1113/1084 225/247 231/270
Birth weight (lbs) 6 (na = 3) 6.5 (na = 68) 6.7 (na = 18) 6.7 (na = 23)
Income 6/49/64 (na = 8) 250/748/1045 (na = 154) 98/167/162 (na = 45) 123/209/133 (na = 36)
(lower/middle/higher)
Race (white/black/Hispan- 92/7/14/14 1501/199/334/163 236/74/135/27 218/113/141/29
ic/other)
FWD 0.19 (na = 4) 0.24 (na = 164) 0.28 (na = 54) 0.31 (na = 51)
Parental education (high 10/75/42 215/1289/693 87/283/102 102/311/88
school/college/postgrad.)
DHEA (pg/mL) 64 (na = 57) 61 (na = 1044) 65 (na = 216) 84 (na = 231)

Downloaded from https://academic.oup.com/cercor/article/30/4/2519/5603326 by guest on 29 May 2023

Physical activity (no. days) 3.7 3.8 (na = 1) 3.5 3.5 (na = 2)
Executive function 99.6 (na = 12) 99.7 (236) 95.8 (na = 52) 93.9 (na = 57)
(age-corrected)
Brain volume (cm3 ) 1180 (na = 30) 1224 (na = 550) 1216 (na = 131) 1207 (na = 112)

BMI was classified using percentile growth charts stratified according to age based on CDC 2001 look up tables (CDC Growth Charts 2018, https://www.cdc.gov/
growthcharts/html_charts/bmiagerev.htm). For statistical assessment, household income levels were categorized as less than $35 000, less than $100 000, and greater
than $100 000. Race was categorized as white, black, Hispanic and other. Parental education was categorized as up to and included General education diploma (GED),
up to and including college or associated degrees, and postgraduate. Mean FWD was used as a measure of head motion during scanning. Physical activity was a
self-reported record of number of days in past week where the subject was physically active for more than 60 min/day.

Relationship Between BMI and Cortical Thickness

Across individuals with complete data (n = 2668), there was a
significant, negative association between BMIZ and mean global
cortical thickness (β = −0.5, t = 2.6, P = 0.01) accounting for demo-
graphic and other covariates excepting DHEA.
After FDR correction for multiple comparisons, regions of sig-
nificant cortical thickness reductions bilaterally included lateral
OFC, inferior frontal gyrus (parsorbitalis and pars triangularis),
and rostral middle frontal and superior frontal cortex. In the
left hemisphere, additional significant differences were found
in entorhinal cortex, while in the right hemisphere additional
changes were found in medial OFC and temporal pole (see
Fig. 2, Table 1, Supplementary Material). In all cases, changes
took the form of a decrease in cortical thickness associated with
Figure 1. Relationship between BMI and executive function after adjustment BMIZ .
for age, sex, race, birth weight, in-scanner motion, parental education, and When we repeated our analysis using waist circumference
household income. and waist-to-height ratio in place of BMI, we found that the
relationship to cortical thickness was broadly similar to the
a positive association between levels of DHEA and executive pattern seen with BMI (see Supplementary Table 1 and Supple-
abilities (β = 0.09, t = 2.9, P = 0.005). mentary Materials).
In a separate analysis, there was a negative association
between DHEA level and average global cortical thickness
Relationship Between BMI and Executive Function (β = −0.06, t = 2.2, P = 0.03) in line with the hypothesis that
There was a significant negative relationship between BMIZ increases in pubertal hormonal levels are associated with
and age-corrected executive function (β = −0.05, t = 2.4, P = 0.02, maturation-related cortical thinning in this age range. In a
n = 2389) accounting for other variables except cortical thickness regional analysis of the association between cortical thickness
and levels of pubertal hormones (see Fig. 1). When levels of and BMI taking DHEA levels into account, only cortical
DHEA were taken into account, the relationship between BMI thickness in rostral middle frontal cortex was associated
and executive function was reduced to trend-level (β = −0.05, with BMI (β = −0.13, t = 3.8, P = 0.01). Given that the effect
t = 1.6, P = 0.1). Because there were far fewer subjects with levels size was not decreased compared with the analysis in the
of DHEA (n = 1227), and the effect size is identical, this likely larger dataset, this suggests that DHEA was not a signifi-
reflects reduced power in the smaller dataset rather than DHEA cant confound of the relationship between BMI and cortical
mediating the relationship between BMI and executive function. thickness, and that differences in results were instead more
Both waist circumference and waist-to-height ratio were also likely due to a comparative lack of power in the smaller
negatively associated with executive function (see Supplemen- dataset (see Supplementary Table 2 and Supplementary
tary Materials). Material).
 Childhood BMI, Brain Structure, and Executive Function Ronan et al. 2523

Downloaded from https://academic.oup.com/cercor/article/30/4/2519/5603326 by guest on 29 May 2023

Figure 2. Map of reduced cortical thickness (beta regression coefficients) associated with BMI, adjusted for demographic and other confounder variables.

Relationship Between Cortical Thickness thickness was a significant mediator of the relationship between
and Executive Function BMIZ and executive function.
Results of analysis revealed that while global mean cortical
Across individuals with complete data (n = 2389), there was a
thickness was not a significant mediator between BMI and exec-
significant negative relationship between mean global cortical
utive function, cortical thickness in 11 regions partially mediates
thickness and executive abilities (β = −0.07, t = 3.1, P = 0.002),
the relationship (see Fig. 4, Supplementary Table 3 and Supple-
without adjusting for BMIZ . At a local level and after FDR-
mentary Material). These regions included the parsorbitalis, pars
correction, mean cortical thickness in several regions was
triangularis, rostral middle frontal and superior frontal cortex
predictive of executive function (see Fig. 3), including cuneus,
bilaterally, and additionally lateral OFC in the left hemisphere,
fusiform, lateral occipital, rostral anterior cingulate, rostral
and fusiform and medial OFC in the right hemisphere (see
middle frontal gyrus, superior and inferior parietal cortex,
Supplementary Table 3 and Supplementary Material).
middle and superior temporal gyrus, pars opercularis, pars
triangularis, postcentral, and supramarginal cortices bilaterally,
and additionally the caudal anterior cingulate, superior tem-
poral sulcus, caudal middle frontal gyrus, lateral orbitofrontal Discussion
cortex, pars orbitalis, precuneus, precentral sulcus, posterior In this study, we investigated the association between BMI,
cingulate and the superior frontal cortex in the left hemisphere, cortical thickness, and executive function in 2700 9–11 year
and lingual region, the precuneus, and the transverse temporal olds recruited as part of the ABCD NIH study. We observed a
sulcus in the right hemisphere (see Supplementary Table 1 and negative association between executive function and cortical
Supplementary Material). In all regions, this association took thickness across the cortical surface. Increased BMI was asso-
the form of a negative relationship between cortical thickness ciated with lower scores on a composite measure of executive
and executive function. This is in line with previous results for function. We also found significant BMI-related differences in
this age-range (Shaw et al. 2006). cortical thickness in line with similar studies (Maayan et al. 2011;
Taking DHEA levels into account in the subset of the sample, Yau et al. 2014; Ross et al. 2015). In particular, reduced cortical
the negative relationship between global mean cortical thick- thickness was pronounced in orbitofrontal cortex, ventromedial
ness and executive abilities remained (β = −0.08, t = 2.6, P = 0.02). PFC, and DPFC, regions involved in executive functions includ-
At a regional level, executive function was again associated ing decision-making, response inhibition, working memory, and
with reduced cortical thickness in several regions including cognitive flexibility.
cuneus and superior parietal cortex in the left hemisphere and The changes that such reduced cortical thickness reflects
pars triangularis and transverse temporal cortex in the right are unknown. For example, previous studies have suggested
hemisphere. Again, the effect size was not reduced compared to that MR-based measures of changes in cortical thickness during
the larger dataset suggesting that DHEA was not a significant childhood may reflect, in part, increases in cortical myelination,
confound of the relationship between cortical thickness and particularly in frontal association areas (Croteau-Chonka et al.
executive function (see Supplementary Table 2 and Supplemen- 2016). Interpreting the results of the current study along these
tary Material). lines, reduced cortical thickness associated with childhood
obesity may be a function of an increase in cortical myelination.
Future studies may consider more direct measures of myeli-
Mediation
nation with a view to increasing power to detect the extent
Having established a relationship between (1) BMI and cortical of structural mediation between BMI and executive function.
thickness, (2) cortical thickness and executive function, and (3) Moreover, other brain parameters may also be important.
BMI and executive function, we next examined whether cortical For example, relative change in degree of connectivity or
 2524 Cerebral Cortex, 2020, Vol. 30, No. 4

Downloaded from https://academic.oup.com/cercor/article/30/4/2519/5603326 by guest on 29 May 2023

Figure 3. Map of reduced cortical thickness (beta regression coefficients) associated with executive function adjusted for demographic and other confounder variables.

Figure 4. Estimate of mediation effect of regional cortical thickness on the relationship between BMI and executive function.

consolidation of activity may more closely index the develop- and changes in this area are convincingly related to risk-taking
ment of executive function and thus may be more sensitive to behavior and substance abuse (Goldstein and Volkow 2011). In
BMI-related differences (Durston et al. 2006; Ezekiel et al. 2013). addition, the relatively extended maturational trajectory of PFC
Previous studies have reported a degree of regional- is thought to subserve experience-dependent learning (Romine
specificity to changes in cortical structure in relation to and Reynolds 2005). As such, differences in PFC structure during
childhood obesity. The current study, capitalizing on a uniquely early adolescence may possibly increase the vulnerability of this
large dataset, demonstrates that increased BMI is associated region to external stressors. Thus, BMI-associated brain changes
with pervasive reductions in cortical thickness across much of in PFC may be regarded as a risk factor to the developing brain.
the PFC. While our study does not allow a clear mechanistic In a complementary analysis, we found that reduced cortical
interpretation of this predominance of effect in PFC, one thickness in the PFC partially mediated the relationship between
possibility is that, since this region is associated with top- BMI and executive function. This observation is compatible with
down control and inhibitory processes, then BMI-related the idea that elevated BMI causes cortical thinning in turn
changes could in turn lead to further difficulties in resisting leading to a reduction in executive function score. The direction
external drives to consumption and attenuated learning from of this causality model is supported by some observational
experience. This could entail a positive feedback in which early studies. For example, in adult populations, a significant number
detrimental changes to PFC structure and function lead to of studies have suggested that obesity may play a causal role
ensuing behavioral changes that exacerbate weight gain. in the onset of brain structural changes and cognitive decline
More generally, PFC is involved in top-down regulation and (Bruce-Keller et al. 2009; Arnoldussen et al. 2014). It is hypoth-
inhibitory control as well as emotion and motivation regulation, esized that factors related to increased body mass such as an
 Childhood BMI, Brain Structure, and Executive Function Ronan et al. 2525

elevated inflammatory response or neuroendocrine dysfunction cardio-metabolic risks than other metrics such as waist circum-
might impact on brain structure and cognitive function in a ference and waist-to-height ratio (Sharma et al. 2015). When
manner akin to neurodegenerative processes observed with we repeated our analysis for waist circumference and waist-
aging. Indeed, many studies have associated increased BMI in to-height ratio, we found that both measures were associated
midlife with increased rates of neurodegeneration and a sig- with lower levels of executive function, as well as regional
nificant elevated risk of dementia and Alzheimer’s disease in reductions in cortical thickness in a manner similar to what
old age (Singh-Manoux et al. 2017). In children, a large-scale we observed using BMI. However, both measures additionally
longitudinal analysis of early childhood development reported identified regions where increased waist circumference and
that obesity in very early childhood is a risk factor for reduced waist-to-height ratio were associated with increases in cortical
cognitive function years later (Li et al. 2018). Important corollar- thickness. These results illustrate that BMI may not capture
ies to these studies are reports of significant improvement of the total variation of cortical structure with increased adiposity.
memory and executive function following weight-loss (Gunstad Finally, although we have confined this extensive analysis to the
et al. 2011; Veronese et al. 2016), as well as the general neuropro- cortical sheet, we acknowledge that subcortical structures have
tective effects of severe caloric restriction (Colman et al. 2009). also been implicated in obesity. Therefore, it will be important

Downloaded from https://academic.oup.com/cercor/article/30/4/2519/5603326 by guest on 29 May 2023

However, care must be taken when interpreting our results, for future work to extend such analyses to subcortical regions
and we note that the causal model of BMI impacting cortical and, critically, to examine covariance relationships between key
thickness which then further impacts executive function is just cortical and subcortical structures in order to more fully char-
one of a possible six models and it is not possible to distinguish acterize the relationship between childhood obesity and brain
these statistically. For example, our data may also fit a model structure.
whereby BMI impacts executive function, which in turn impacts
cortical thickness. Alternatively, our data would be equally com-
patible with a hypothesis that executive function influences
Conclusions
BMI, which in turn may influence cortical structure. Indeed, In a large, population-based cohort, reduction in PFC cortical
various studies support such a hypothesis. For example, lon- thickness was associated with childhood obesity. Higher BMI
gitudinal studies suggest that the early cognitive environment was also associated with reduced scores on a composite cogni-
may be a risk factor for developing obesity in later childhood, tive measure reflecting executive processes, and a complemen-
with children in lower cognitive stimulation environments at tary mediation analysis was consistent with cortical thickness
a 2-fold greater risk of developing obesity (Strauss and Knight change mediating the relationship between BMI and executive
1999). Meanwhile, in bariatric patients, executive function has functioning. The data are consistent with a mechanism whereby
been shown to predict postsurgical weight-loss (Spitznagel et al. PFC changes in childhood obesity may lead to altered regula-
2013). tion of inhibitory control and risk-taking behavior and further
We also acknowledge the possibility that there is no causal difficulties in weight control. However, due to the limitations
relationship between BMI and executive function. This would of our data, care must be taken in interpreting our results and
be compatible with a model in which cortical structural features follow-up studies will be critical to establishing causal pathways
drive altered BMI and executive functioning independently. This between BMI, brain structure, and executive function, as well as
is feasible given that the genes associated with obesity-risk determining if longitudinal changes in BMI have a measurable
are predominantly and significantly expressed in the central impact on these traits.
nervous system and linked to basic functions such as glutamate
signaling and synaptic function (Locke et al. 2015) and that BMI,
brain structure, and various aspects of cognitive function share
Supplementary Material
common genetic influences (Curran et al. 2013; Hagenaars et al. Supplementary Material is available at Cerebral Cortex online.
2016; Marioni et al. 2016). The finding that BMI shares common
genetic influences with various aspects of brain structure and
cognition highlights the difficulty in isolating causal associa-
Funding
tions in noninterventional studies and underscores the impor- The Bernard Wolfe Health Neuroscience Fund and the Wellcome
tance of more direct studies in nonhumans. In this regard, it may Trust (grant number RNAG/259).
be that BMI and executive function are not causally related, and
structural changes associated with each may simply be regarded
as an important confound of the relationship (MacKinnon et
Notes
al. 2000). Indeed, we note that it is not possible to statistically Data used in the preparation of this article were obtained
distinguish between a confound and a mediator. In this regard, from the Adolescent Brain Cognitive Development (ABCD)
the results of this study suggest that BMI, cortical thickness or Study (https://abcdstudy.org), held in the NIMH Data Archive
executive function should be included as a potential confound (NDA). This is a multisite, longitudinal study designed to
in any future analysis that seeks to investigate the relationship recruit more than 10,000 children aged 9-10 and follow
between the other 2 variables. them over 10 years into early adulthood. The ABCD Study
There were a number of significant limitations in this study. is supported by the National Institutes of Health and addi-
Primarily, due to the cross-sectional nature of our data, we tional federal partners “under award numbers U01DA041022,
were not able to distinguish between different possible mod- U01DA041028, U01DA041048, U01DA041089, U01DA041106,
els to determine the causal relationship between BMI, execu- U01DA041117, U01DA041120, U01DA041134, U01DA041148,
tive function, and cortical thickness. This may be addressed U01DA041156, U01DA041174, U24DA041123, and U24DA041147”.
by future studies based on longitudinal data. In addition, our A full list of supporters is available at https://abcdstudy.org/nih-
analysis was based on measures of BMI. While BMI is the most collaborators. A listing of participating sites and a complete
commonly used index of adiposity, it is less directly related to listing of the study investigators can be found at https://
 2526 Cerebral Cortex, 2020, Vol. 30, No. 4

abcdstudy.org/principal-investigators.html. ABCD consortium among adolescent girls: an fMRI study. Neuroimage. 52:
investigators designed and implemented the study and/or 1696–1703.
provided data but did not necessarily participate in analysis Benjamini Y, Hochberg Y. 1995. Controlling the false discovery
or writing of this report. This manuscript reflects the views of rate: a practical and powerful approach to multiple testing.
the authors and may not reflect the opinions or views of the JSTOR. 57:289–300.
NIH or ABCD consortium investigators. Biro FM, Wien M. 2010. Childhood obesity and adult morbidities.
The ABCD data repository grows and changes over time. The Am J Clin Nutr. 91:1499S–1505S.
ABCD data used in this report came from the curated annual Bruce AS, Holsen LM, Chambers RJ, Martin LE, Brooks WM,
release 1.0 (DOI 10.15154/1412097). DOIs can be found at https:// Zarcone JR, Butler MG, Savage CR. 2010. Obese children
ndar.nih.gov/study.html?id=500. show hyperactivation to food pictures in brain networks
Conflict of Interest: P.C.F. has received money in the past for ad linked to motivation, reward and cognitive control. Int J Obes.
hoc consultancy services to GlaxoSmithKline. All other authors 34:1494–1500.
declare no competing financial interests. Bruce-Keller AJ, Keller JN, Morrison CD. 2009. Obesity and vul-
nerability of the CNS. Biochim Biophys Acta. 1792:395–400.

Downloaded from https://academic.oup.com/cercor/article/30/4/2519/5603326 by guest on 29 May 2023

CDC Growth Charts. 2018. https://www.cdc.gov/growthcharts/
References clinical_charts.htm (last accessed 30 May 2018).
Akshoomoff N, Beaumont JL, Bauer PJ, Dikmen SS, Gershon RC, Casaletto KB, Umlauf A, Beaumont J, Gershon R, Slotkin J,
Mungas D, Slotkin J, Tulsky D, Weintraub S, Zelazo PD et al. Akshoomoff N, Heaton RK. 2015. Demographically corrected
2013. NIH toolbox cognition battery (CB): composite scores of normative standards for the English version of the NIH tool-
crystallized, fluid, and overall cognition. Monogr Soc Res Child box cognition battery. J Inter Neuropsych Soc. 21:378–391.
Dev. 78:119–132. Casey BJ, Cannonier T, Conley MI, Cohen AO, Barch DM, Heitzeg
Akshoomoff NN, Thompson E, McCabe WK, Bloss C, Chang CS, MM, Soules ME, Teslovich T, Dellarco DV, Garavan H et al.
Amarai L, Casey DG, Ernst BJ, Frazier TM, Gruen JA et al. 2018. The adolescent brain cognitive development (ABCD)
2014. The NIH toolbox cognition battery: results from a large study: imaging acquisition across 21 sites. Dev Cogn Neurosci.
normative developmental sample (PING). Neuropsychol. 28: 32:43–54.
1–10. Colman RJ, Anderson RM, Johnson SC, Kastman EK, Kosmatka
Akshoomoff N, Brown TT, Bakerman R, Hagler DJ. 2018. KJ, Beasley TM, Allison DB, Cruzen C, Simmons HA, Kemnitz
Developmental differentiation of executive functions on JW et al. 2009. Caloric restriction delays disease onset and
the NIH toolbox cognition battery. Neuropsychology. 32: mortality in rhesus monkeys. Science. 325:201–204.
777–783. Croteau-Chonka EC, Dean DC, Remer J, Dirks H,
Alarcón G, Ray S, Nagel BJ. 2016. Lower working memory per- O’Muircheartaigh J, Deoni SCL. 2016. Examining the
formance in overweight and obese adolescents is medi- relationships between cortical maturation and white
ated by white matter microstructure. J Inter Neuropsych Soc. matter myelination throughout early childhood. NeuroImage.
22:281–292. 125:413–421.
Alexander-Bloch A, Clasen L, Stockman M, Ronan L, Lalonde F, Curran JE, McKay DR, Winkler AM, Olvera RL, Carless MA, Dyer
Giedd J, Raznahan A. 2016. Subtle in-scanner motion biases TD, Kent JW, Kochunov P, Sprooten E, Knowles EE et al. 2013.
automated measurement of brain anatomy from in vivo MRI. Identification of pleiotropic genetic effects on obesity and
Hum Brain Mapp. 37:2385–2397. brain anatomy. Hum Hered. 75:136–143.
Alosco ML, Galioto R, Spitznagel MB, Strain G, Devlin M, Cohen Dale AM, Fischl B, Sereno MI. 1999. Cortical surface-based analy-
R, Crosby RD, Mitchell JE, Gunstad J. 2014a. Cognitive function sis. I. Segmentation and surface reconstruction. NeuroImage.
after bariatric surgery: evidence for improvement 3 years 9:179–194.
after surgery. Am J Surg. 207:870–876. Davids S, Lauffer H, Thoms K, Jagdhuhn M, Hirschfeld H, Domin
Alosco ML, Stanek KM, Galioto R, Korgaonkar MS, Grieve SM, M, Hamm A, Lotze M. 2010. Increased dorsolateral prefrontal
Brickman AM, Spitznagel MB, Gunstad J. 2014b. Body mass cortex activation in obese children during observation of food
index and brain structure in healthy children and adoles- stimuli. Int J Obes. 34:94–104.
cents. Int J Neurosci. 124:49–55. Del Parigi A. 2010. Neuroanatomical correlates of hunger and
Arnoldussen IAC, Kiliaan AJ, Gustafson DR. 2014. Obesity and satiety in lean and obese individuals. In: Dube L, Bechara A,
dementia: adipokines interact with the brain. Eur Neuropsy- Dagher A, Drewnowski A, LeBel J, James P & Yada R, editors.
chopharmacol. 24:1982–1999. Obesity Prevention: The Role of Brain and Society on Individual
Barch DM, Albaugh MD, Avenevoli S, Chang L, Clark DB, Glantz Behavior. Elsevier: Academic Press. 1:253–271.
MD, Hudziak JJ, Jernigan TL, Tapert SF, Yurgelun-Todd D et al. Desikan RS, Ségonne F, Fischl B, Quinn BT, Dickerson BC, Blacker
2018. Demographic, physical and mental health assessments D, Buckner RL, Dale AM, Maguire RP, Hyman BT et al. 2006.
in the adolescent brain and cognitive development study: An automated labeling system for subdividing the human
Rationale and description Dev Cogn Neurosci. 32:55–66. cerebral cortex on MRI scans into gyral based regions of
Baron RM, Kenny DA. 1986. The moderator-mediator vari- interest. NeuroImage. 31:968–980.
able distinction in social psychological research: conceptual, Durston S, Davidson MC, Tottenham N, Galvan A, Spicer J, Fos-
strategic and statistical considerations. J Pers Soc Psychol. sella JA, Cassey BJ. 2006. A shift from diffuse to focal cortical
51:1173–1182. activity with development. Dev Sci. 9:1–8.
Bathelt J, Gathercole SE, Johnson A, Astle DE. 2018. Differences Eichen DM, Brittany EM, Liang J, Strong DR, Rhee K, Boutelle KN.
in brain morphology and working memory capacity across 2018. The relationship between executive functioning and
childhood. Dev Sci. 21:e12579. weight loss and maintenance in children and parents partici-
Batterink L, Yokum S, Stice E. 2010. Body mass correlates pating in family-based treatment for childhood obesity. Behav
inversely with inhibitory control in response to food Res Ther. 105:10–16.
 Childhood BMI, Brain Structure, and Executive Function Ronan et al. 2527

Ezekiel F, Bosma R, Morton JB. 2013. Dimensional change card Hodgson K, Poldrack RA, Curran JE, Knowles EE, Mathias S, Göring
sort performance associated with age-related differences in HHH, Yao N, Olvera RL, Fox PT, Almasy L et al. 2017. Shared
functional connectivity of lateral prefrontal cortex. Dev Cogn genetic factors influence head motion during MRI and body
Neurosci. 5:40–50. mass index. Cereb Cortex. 27:5539–5546.
Fischl B, Dale AM. 2000. Measuring the thickness of the human Holsen LM, Zarcone JR, Thompson TI, Brooks WM, Anderson MF,
cerebral cortex from magnetic resonance images. Proc Natl Ahluwalia JS, Nollen NL, Savage CR. 2005. Neural mechanisms
Acad Sci USA. 97:11050–11055. underlying food motivation in children and adolescents. Neu-
Fischl B, Sereno MI, Dale AM. 1999a. Cortical surface-based anal- roImage. 27:669–676.
ysis. II: inflation, flattening, and a surface-based coordinate Jernigan TL, Brown SA, ABCD consortium coordinators: intro-
system. NeuroImage. 9:195–207. duction. 2018.Dev Cogn Neurosci. 32:1–3.
Fischl B, Sereno MI, Tootell RB, Dale AM. 1999b. High-resolution Kharitonova M, Martin RE, Gabrieli JDE, Sheridan MA. 2013.
intersubject averaging and a coordinate system for the corti- Cortical gray-matter thinning is associated with age-related
cal surface. Hum Brain Mapp. 8:272–284. improvements on executive function tasks. Dev Cogn Neurosci.
Frangou S, Chitins X, Williams SCR. 2004. Mapping IQ and 6:61–71.

Downloaded from https://academic.oup.com/cercor/article/30/4/2519/5603326 by guest on 29 May 2023

gray matter density in healthy young people. NeuroImage. Killgore W, Yurgelun-Todd DA. 2005. Body mass predicts
23:800–805. orbitofrontal activity during visual presentations of high-
Garavan H, Bartsch H, Conway K, Decastro A, Goldstein RZ, calorie foods. NeuroReport. 16:859–863.
Heeringa S, Jernigan T, Potter A, Thompson W, Zahs D. 2018. Korpela S, Nyman A, Munck P, Ahtola A, Matomäki J, Korhonen
Recruiting the ABCD sample: Design considerations and pro- T, Parkkola R, Haataja L, PIPARI Study Group. 2018. Working
cedures Dev Cogn Neurosci. 32:16–22. memory in very-low-birthweight children at the age of 11
Gehring WJ, Willoughby AR. 2002. The medial frontal cortex and years. Child Neuropsychol. 24:338–353.
the rapid processing of monetary gains and losses. Science. Kringelbach M. 2005. The human orbitofrontal cortex: linking
295:2279–2282. reward to hedonic experience. Nat Rev Neurosci. 6:691–702.
Glavin K, Roelants M, Strand BH, Júlíusson PB, Lie KK, Helseth S, Labouesse MA, Polesel M, Clementi E, Müller F, Markkanen E,
Hovengen R. 2014. Important periods of weight development Mouttet F, Cattaneo A, Richetto J. 2018. MicroRNA expression
in childhood: a population-based longitudinal study. BMC profiling in the prefrontal cortex: putative mechanisms for
Public Health. 14:306. the cognitive effects of adolescent high fat feeding. Sci Rep.
Gluck ME, Viswanath P, Stinson EJ. 2017. Obesity, appetite, and (1):14.
the prefrontal cortex. Curr Obes Rep. 6:380–388. Lavagnino L, Mwangi B, Bauer IE, Cao B, Selvaraj S, Prossin
Goldstein RZ, Volkow ND. 2011. Dysfunction of the prefrontal A, Soares JC. 2016. Reduced inhibitory control mediates the
cortex in addiction: neuroimaging findings and clinical relationship between cortical thickness in the right superior
implications. Nat Rev Neurosci. 12:652–669. frontal Gyrus and body mass index. Neuropsychopharmacology.
Graham AL, Gluck ME, Votruba SB, Krakoff J, Thearle MS. 2014. 41:2275–2282.
Perseveration augments the effects of cognitive restraint on Le DSNT, Pannacciulli N, Chen K, Parigi AD, Salbe A, Reiman
ad libitum food intake in adults seeking weight loss. Appetite. EM, Krakoff J. 2006. Less activation of the left dorsolateral
82:78–84. prefrontal cortex in response to a meal: a feature of obesity.
Gray JR, Chabris CF, Braver TS. 2003. Neural mechanisms of Am J Clin Nutr. 84:725–731.
general fluid intelligence. Nat Neurosci. 6:316–322. Li N, Yolton K, Lanphear BP, Chen A, Kalkwarf HJ, Braun JM.
Guerrieri R, Nederkoorn C, Jansen A. 2008. The interac- 2018. Impact of early-life weight status on cognitive abilities
tion between impulsivity and a varied food environment: in children. Obesity. 26:1088–1095.
its influence on food intake and overweight. Int J Obes. Liang J, Matheson BE, Kaye WH, Boutelle KN. 2014. Neurocog-
32:708–714. nitive correlates of obesity and obesity-related behaviors in
Gunstad J, Spitznagel MB, Paul RH, Cohen RA, Kohn M, Luyster children and adolescents. Int J Obes. 38:494–506.
FS, Clark R, Williams LM, Gordon E. 2008. Body mass index Locke AE, Kahali B, Berndt SI, Justice AE, Pers TH, Day FR, Powell
and neuropsychological function in healthy children and C, Vedantam S, Buchkovich ML, Yang J et al. 2015. Genetic
adolescents. Appetite. 50:246–251. studies of body mass index yield new insights for obesity
Gunstad J, Strain G, Devlin MJ, Wing R, Cohen RA, Paul RH, Crosby biology. Nature. 518:197–206.
RD, Mitchell JE. 2011. Improved memory function 12 weeks Luciana M, Bjork JM, Nagel BJ, Barch DM, Gonzalez R, Nixon
after bariatric surgery. Surg Obes Relat Dis. 7:465–472. SJ, Banich MT. 2018. Adolescent neurocognitive development
Gupta A, Mayer EA, Sanmiguel CP, Van Horn JD, Woodworth and impacts of substance use: overview of the adolescent
D, Ellingson BM, Fling C, Love A, Tillisch K, Labus JS. 2015. brain cognitive development (ABCD) baseline neurocognition
Patterns of brain structural connectivity differentiate nor- battery. Dev Cogn Neurosci. 32:67–79.
mal weight from overweight subjects. NeuroImage: Clinical. Maayan L, Hoogendoorn C, Sweat V, Convit A. 2011. Dis-
7:506–517. inhibited eating in obese adolescents is associated with
Hagenaars SP, Harris SE, Davies G, Hill WD, Liewald DCM, Ritchie orbitofrontal volume reductions and executive dysfunction.
SJ, Marioni RE, Fawns-Ritchie C, Cullen B, Malik R et al. 2016. Obesity. 19:1382–1387.
Shared genetic aetiology between cognitive functions and MacKinnon DP, Krull JL, Lockwood CM. 2000. Equivalence of
physical and mental health in UK biobank (N = 112 151) and the mediation. Confounding and suppression effect. Prev Sci.
24 GWAS consortia. Mol Psychiatry. 21:1624–1632. 1:173–181.
Hayes AF. 2009. Beyond Baron and Kenny: statistical media- Marioni RE, Yang J, Dykiert D, Mõttus R, Campbell A, CHARGE
tion analysis in the new millennium. Commun Monograph. Cognitive Working Group, Davies G, Hayward C, Porteous DJ,
76:408–420. Visscher PM et al. 2016. Assessing the genetic overlap
 2528 Cerebral Cortex, 2020, Vol. 30, No. 4

between BMI and cognitive function. Mol Psychiatry. Sowell ER, Thompson PM, Leonard CM, Welcome SE, Kan E, Toga
21:1477–1482. AW. 2004. Longitudinal mapping of cortical thickness and
Medic N, Ziauddeen H, Erche KD, Bullmore ET, Nathan PJ, Ronan brain growth in normal children. J Neurosci. 24:8223–8231.
L, Fletcher PC. 2016. Increased body mass index is associated Spitznagel MB, Alosco M, Strain G, Devlin M, Cohen R, Paul R,
with specific regional alterations in brain structure. Int J Obes. Crosby RD, Mitchell JE, Gunstad J. 2013. Cognitive function
40:1177–1182. predicts 24-month weight loss success after bariatric surgery.
NCD Risk Factor Collaboration (NCD-RisC). 2017. Worldwide Surg Obes Relat Dis. 9:765–770.
trends in body-mass index, underweight, overweight, and Stice E, Yokum S, Bohon C, Marti N, Smolen A. 2010. Reward cir-
obesity from 1975 to 2016: a pooled analysis of 2416 cuitry responsivity to food predicts future increases in body
population-based measurement studies in 128.9 million chil- mass: moderating effects of DRD2 and DRD4. NeuroImage.
dren, adolescents, and adults. Lancet. 390:2627–2642. 50:1618–1625.
Nguyen T-V, McCracken JT, Ducharme S, Cropp BF, Botteron Strauss RS, Knight J. 1999. Influence of the home environment
KN, Evans AC, Karama S. 2013. Interactive effects of dehy- on the development of obesity in children. Pediatrics. 103:e85.
droepiandrosterone and testosterone on cortical thickness Tamnes CK, Walhovd KB, Grydeland H, Holland D, Østby Y, Dale

Downloaded from https://academic.oup.com/cercor/article/30/4/2519/5603326 by guest on 29 May 2023

during early brain development. J Neurosci. 33:10840–10848. AM, Fjell AM. 2013. Longitudinal working memory devel-
Ogden CL, Carroll MD, Kit BK, Flegal KM. 2014. Prevalence of opment is related to structural maturation of frontal and
childhood and adult obesity in the United States, 2011–2012. parietal cortices. J Cog Neurosci. 25:1611–1623.
JAMA. 311:806–814. Tingley D, Yamamoto T, Hirose K, Keele L, Imai K. 2014. Medi-
Reinert KRS, Po’e EK, Barkin SL. 2013. The relationship between ation: R package for causal mediation analysis. J Stat Softw.
executive function and obesity in children and adolescents: 59:1–38.
a systematic literature review. J Obes. 820956–820910. Uban KA, Horton MK, Jacobus J, Heyser C, Thompson WK, Tapert
Riggs NR, Huh J, Chou CP, Spruijt-Metz D, Pentz AM. 2012. Exec- SF, Madden PAF, Sowell ER, Adolescent Brain Cognitive Devel-
utive function and latent classes of childhood obesity risk. J opment Study. 2018. Biospecimens and the ABCD study:
Behav Med. 35:642–650. rationale, methods of collection, measurement and early
Romine CB, Reynolds CR. 2005. A model of the development of data. Dev Cogn Neurosci. 32:97–106.
frontal lobe functioning: findings from a meta-analysis. Appl Verbeken S, Braet C, Claus L, Nederkoorn C, Oosterlaan J.
Neuropsychol. 12:190–201. 2009. Childhood obesity and impulsivity: an investiga-
Ross N, Yau PL, Convit A. 2015. Obesity, fitness and brain integrity tion with performance-based measures. Behav Change. 26:
in adolescence. Appetite. 93:44–50. 153–167.
Sharma AK, Metzger DL, Daymont C, Hadjiyannakis S, Rodd CJ. Verbeken S, Braet C, Lammertyn J, Goossens L, Moens E. 2012.
2015. LMS tables for waist-circumference and waist-height How is reward sensitivity related to bodyweight in children?
ratio Z-scores in children aged 5-19 y in NHANES III: asso- Appetite. 58:478–483.
ciation with cardio-metabolic risks. Pediatr Res. 78:723–729. Veronese N, Facchini S, Stubbs B, Luchini C, Solmi M, Manzato
Shaw P, Greenstein D, Lerch J, Clasen L, Lenroot R, Gogtay N, E, Sergi G, Maggi S, Cosco T, Fontana L. 2016. Weight loss is
Evans A, Rapoport J, Giedd J. 2006. Intellectual ability and associated with improvements in cognitive function among
cortical development in children and adolescents. Nature. overweight and obese people: a systematic review and meta-
440:676–679. analysis. Neurosci Biobehav Rev. 72:87–94.
Shields GS, Moons WG, Slavich GM. 2017. Inflammation, self- Yang Y, Shields GS, Guo C, Liu Y. 2018. Executive func-
regulation, and health: an immunologic model of self- tion performance in obesity and overweight individuals:
regulatory failure. Perspect Psychol Sci. 8:588–612. a meta-analysis and review. Neurosci Biobehav Rev. 84:
Singh-Manoux A, Dugravot A, Shipley M, Brunner EJ, Elbaz A, 225–244.
Sabia S, Kivimaki M. 2017. Obesity trajectories and risk of Yau PL, Kang EH, Javier DC, Convit A. 2014. Preliminary evidence
dementia: 28 years of follow-up in the Whitehall II study. of cognitive and brain abnormalities in uncomplicated ado-
Alzheimers Dement. 14:178–186. lescent obesity. Obesity. 22:1865–1871.