Genetic Determinants of Childhood Obesity

Abstract

Obesity represents a major health burden to both developed and developing countries. Furthermore, the incidence of obesity is increasing in children. Obesity contributes substantially to mortality in the United States by increasing the risk for type 2 diabetes, cardiovascular-related diseases and other comorbidities. Despite environmental changes over the past decades, including increases in high-calorie foods and sedentary lifestyles, there is very clear evidence of a genetic predisposition to obesity risk.

Childhood obesity cases can be categorized in one of two ways: syndromic or non-syndromic. Syndromic obesity includes disorders such as Prader-Willi syndrome, Bardet-Biedl syndrome and Alström syndrome. Non-syndromic cases of obesity can be further separated into rarer instances of monogenic obesity and much more common forms of polygenic obesity.

The advent of genome-wide association studies (GWAS) and next generation sequencing has driven significant advances in our understanding of the genetic contribution to childhood obesity. Many rare and common genetic variants have been shown to contribute to the heritability in obesity, although the molecular mechanisms underlying most of these variants remain unclear.

An important caveat of GWAS efforts is that they do not strictly represent gene target discoveries, rather simply the uncovering of robust genetic signals. One clear example of this is with progress in understanding the key obesity signal harbored within an intronic region of the FTO gene. It has been shown that the non-coding region in which the variant actually resides in fact influences the expression of genes distal to FTO instead, specifically IRX3 and IRX5. Such discoveries suggest that associated non-coding variants can be embedded within or next to one gene, but commonly influence the expression of other, more distal effector genes.

Advances in genetics and genomics are therefore contributing to a deeper understanding of childhood obesity, allowing for development of clinical tools and therapeutic agents.

1.0. Introduction

Childhood obesity impacts over 17% of children in the United States and is associated with serious health outcomes [1]. Obesity is a key contributor to mortality, operating as a major risk factor for common diseases that include type 2 diabetes, cardiovascular diseases and hypertension [2]. Cases of childhood obesity are classified as syndromic or non-syndromic, with the latter categorized by the given genetic etiology of either being driven by monogenic or polygenic factors.

Body mass index (BMI) definitions of obesity status in children have proven to be more challenging to define than for adults where there is a single cutoff value for all ages. Adjustments for age and sex are crucial as BMI varies widely with age throughout childhood and adolescence, especially during puberty [3]. However, there is a general consensus that ‘‘overweight’ starts at the 85^th percentile of sex-specific BMI-for-age growth charts, while obesity is defined as at and above the 95^th percentile [4–7]. Severe pediatric obesity is defined as at and above 120% of the 95^th percentile [8] or an absolute BMI at and above 35 kg/m² and is the only obesity classification that is increasing in prevalence in children and adolescents [9]. Indeed, as children progress through adolescence, such percentiles converge with the definitions in adults, with the 95^th percentile being close to equivalent to 30 kg/m² [10, 11].

Of note, BMI is highly associated with direct measurements of fat mass and waist circumference. BMI is more highly correlated with total fat mass by dual-energy X-ray absorptiometry (0.91, P<0.001) than waist circumference (0.56, P<0.001) [12]. Further, BMI is more highly correlated with truncal fat mass and appendicular fat mass (arms and legs) than is waist circumference [12]. Both higher BMI and waist circumference are about equally predictive of cardiovascular disease [13]. Thus, BMI is both associated with other measures of fat mass as well as with future cardiovascular disease.

Obesity is a common phenotype which results from complex interactions between environmental factors and genetics. Together these factors contribute to an imbalance in equilibrium between energy uptake and utilization, leading to an excess in adipose tissue [3]. While current primary treatment methods for childhood obesity only address environmental and lifestyle factors [14], heritable factors are also vital to characterize. Differences in obesity prevalence between different ethnic populations [15], familial transmission and twin concordance support the existence of a substantial genetic component. The heritability for variation in BMI ranges from approximately 40 to 70% [16–20], depending on the study design. Similarly, the heritability for variation in other measures of obesity such as fat mass and waist circumference are estimated to range from 63 to 71% [21] and 72 to 82% [22], respectively. An individual-based pooled analysis of twin cohorts demonstrated that variation in BMI explained by additive genetic factors was lowest at 4 years of age and gradually increased with age until the onset of adulthood [19]. Interestingly, this may reflect children gaining more independence from their parents in eating and being able to act more according to their genetic predisposition [19].

There are clinical features of loci which have more than one phenotype. For example, clinically, melanocortin-4 receptor (MC4R) deficiency is associated with increased eating in children which may be a result of hyperinsulinemia [23]. Further, in MC4R deficient patients, a reduced sympathetic tone and reduction in blood pressure are also reported [24]. As described below, syndromic obesity is often accompanied by other clinical features. These sub-phenotypes are of important interest and further description is beyond the scope of this manuscript, but has significant clinical relevance.

Although it appears that the field is not yet ready to perform genetic diagnostics in children with polygenic obesity, the most common form of obesity, there are reported associations with variants and eating behavior and satiety which lead to obesity. For example, children aged 4–5 years homozygous for a low-risk FTO variant ate less in a laboratory setting than children with one or two high-risk alleles [25]. Further, parents of these children reported improved satiety in those children with the low-risk allele compared with those who had the high-risk allele [26]. Future research should include efforts to define sub-populations of children with variants associated with obesity and eating behaviors as well as response to treatment. There are defined variants which relate to both syndromic and monogenic obesity; children should be genetically tested when clinically indicated, as described below.

2.0. Syndromic Childhood Obesity

Syndromic childhood obesity is a class of rare forms of the disease, where obesity presents as part of a distinct set of clinical phenotypes, commonly including developmental delay and dysmorphic characteristics [27]. Approximately 25 different types of syndromic obesity have been described, which are based on a series of genetic abnormalities occurring on the autosomes or the X chromosome [28]. Here we outline some of the main forms of syndromic childhood obesity (Table 1), but one must note there are many others, including Borjeson-Forssman-Lehmann, Carpenter, Ayazi, Coffin-Lowry, Cohen, Fragile X, Rubinstein-Taybi and Wilson-Turner syndromes [27].

Table 1.

Most Common Forms of Syndromic Obesity [27, 29, 31]

Syndrome	Prevalence	Implicated Gene(s)	Phenotypes	Genetic Testing for Diagnosis
Prader-Willi	1 in 10,000–30,000	SNURF-SNRPN, MKRN3, NECDIN, MAGEL2	Severe neonatal hypotonia, Feeding difficulties followed by hyperphagia and excessive weight gain, Developmental delay, Hypogonadism, Intellectual disability, Characteristic facial features (almond-shaped eyes, thin upper lip, downturned corners of the mouth, narrow face), Behavioral problems, Small hands and feet, Short stature	DNA methylation testing, Fluorescence in situ hybridization, Microarray, DNA polymorphism analysis
Bardet-Biedl	1 in 13,500 (Israel and Arab countries), 1 in 160,000 (Switzerland)	BBS1-BBS20, NPHP1, FBN3, CEP19	Obesity, Cone-rod dystrophy, Postaxial polydactyly, Cognitive impairment, Hypogenitalism, Renal abnormalities, Shortened life expectancy	Targeted high-throughput sequencing
Alström	1–9 in 1,000,000	ALMS1	Obesity, Cone-rod dystrophy, Renal anomalies, Progressive sensorineural hearing impairment, Hypogonadism, Reduced adult height, Type 2 diabetes, Dilated or restrictive cardiomyopathy, Shortened life expectancy	Targeted high-throughput sequencing

Prader-Willi syndrome is the mostly commonly seen type of syndromic obesity, driven by a lack of expression of the paternally inherited genes on chromosome 15q11.2-q13 due to genomic or epigenetic variation [29]. Epigenetic variation refers to differences in heritable modifications to both DNA and histones, such as methylation or acetylation events, which do not involve altering the DNA sequence itself but can influence gene expression [30]. Some genes in the chromosome 15q11.2-q13 region, including SNURF-SNRPN, MKRN3, NECDIN and MAGEL2, are typically active exclusively from the paternally transmitted chromosome 15 due to normal genomic imprinting [29]. However, loss of paternal expression of these genes results in a total absence of expression, given that the maternal genes are already silenced by epigenetic mechanisms [29]. The paucity of gene expression from the paternally inherited chromosome 15 is often due to a 5 to 6 Mb deletion, but can also be driven by maternal uniparental disomy 15 or an imprinting error [29]. A recent study on Prader-Willi syndrome observed that 61% of patients harbored a 15q11.2-q13 deletion, 36% had maternal uniparental disomy and 3% were a consequence of imprinting defects [31].

Prader-Willi syndrome is defined by several presentations including endocrine abnormalities, such as growth hormone deficiency, central adrenal insufficiency, hypothyroidism and hypogonadism [32], and challenges with feeding in early life, followed by over-eating in later childhood, eventually resulting in morbid obesity [29]. Genetic testing is used to diagnose this syndrome. Specifically, assessment of DNA methylation can reveal changes in parent-specific imprinting on chromosome 15 [31], while conducting fluorescence in situ hybridization [29] or leveraging microarray-based approaches are options to detect chromosome 15 deletions [31]. DNA polymorphism analysis in affected individuals and their parents is used to detect uniparental disomy [29]. Genetic testing has therefore proven vital to confirm the diagnosis of Prader-Willi syndrome, especially in cases with atypical presentation or for individuals too young for such clinical diagnostic approaches [29].

Bardet-Biedl syndrome is a rare form of syndromic obesity, which is transmitted in an autosomal recessive manner [33]. It presents along with learning difficulties and vision, genital, renal and structural abnormalities, such as polydactyly, brachydactyly and syndactyly [33]. Genes identified to contribute to this syndrome include BBS1-BBS20, NPHP1, FBN3 and CEP19 [27]. These genes all participate in cilia function [33] but appear to lack natural redundancy, as disruption of any of one these genes can lead to cilia impairment [34]. Most tested Bardet-Biedl syndrome cases present with mutations in either BBS1 or BBS10, although the clinical presentation does not substantially differ based on the impacted gene in each case [35]. The understanding of the precise genetic contributors to Bardet-Biedl syndrome, however, remains limited [35] as many of the identified mutations represent simple amino acid substitutions and are therefore not overtly deleterious [36–40]. Targeted high-throughput sequencing offers the best approach to provide the most efficient diagnostic yields, as other methods have time and cost constraints given the known locus heterogeneity for this disorder [27]. Even with these capabilities, the relationship between genotype and phenotype in Bardet-Biedl syndrome is not yet fully established [27].

Alström syndrome represents another rare autosomal recessive disease, with similar clinical features to Bardet-Biedl syndrome [27]. This disorder presents with obesity, along with visual, stature, renal and gonadal abnormalities [27]. Discerning the differences between the two syndromes is principally based on when visual issues present and if post-axial polydactyly is observed [27]. Alström syndrome is driven by mutations in ALMS1, situated on chromosome 2p13 [27], with additional triallelic mutations having been reported in less severe cases [41]. The known function of ALMS1 suggests it plays a role in cilia function [27]. An excess of 200 mutations in ALMS1 have been reported to date, with most occurring within exons 8, 10 and 16 [42]. However, the most impactful mutations are either nonsense or dramatic frameshifts, resulting in a non-functional protein [42, 43]. As such, sequencing approaches are the most commonly used methods to determine the causal genetic events for Alström syndrome [43]. The precise molecular mechanism for this disorder has not yet been fully elucidated [42].

In addition, there is a growing body of evidence implicating genomic events on chromosome 16p11.2 driving a penetrant and severe form of obesity, along with features of intellectual disability [44, 45].

3.0. Non-Syndromic Monogenic Childhood Obesity

Several forms of non-syndromic monogenic childhood obesity exist, although they are observed in less than 1% of children evaluated in tertiary level pediatric clinics [46]. Monogenic obesity is defined by a mutation occurring in just a single gene involved in regulation of body weight [14]. As obesity is such a highly heterogeneous disease, relatively severe cases presenting with an early age of onset are typically driven by highly penetrant rare genetic variants [47]. Most genes implicated in this form of childhood obesity are involved in the leptin-melanocortin signalling pathway [48]. Here we describe five genes causing some of the most prevalent forms of non-syndromic monogenic childhood obesity (Table 2), while other genes include BDNF, MC3R, MRAP2, NTRK2, SH2B1 and SIM1 [49].

Table 2.

Most Common Forms of Non-Syndromic Monogenic Obesity [49, 60]

Gene	Mutation(s)	Prevalence	Phenotypes	Genetic Testing for Diagnosis
MC4R	Homozygous, Compound heterozygous, Heterozygous	3–5% of children with early-onset severe obesity	Hyperphagia, Early-onset obesity, Increased linear growth and height, Increased bone mass, Increased fat and lean mass	Direct sequencing of MC4R
LEP	Homozygous	27 patients reported	Severe hyperphagia, Incapacity of feeling satiety, Early-onset obesity, Hypogonadotropic hypogonadism, Hypothalamic hypothyroidism with puberty delay, Reduced adult height, Increased risk of infections	Direct sequencing of LEP
LEPR	Homozygous, Compound heterozygous	3% of children with severe obesity from a cohort enriched in consanguineous families	Severe hyperphagia, Incapacity of feeling satiety, Early-onset obesity, Hypogonadotropic hypogonadism, Hypothalamic hypothyroidism with puberty delay, Reduced adult height, Increased risk of infections, Increased bone mineral density	Direct sequencing of LEPR
POMC	Homozygous, Compound heterozygous	11 patients reported	Neonatal adrenal insufficiency, Early-onset obesity, Hyperphagia, Central hypothyroidism, Gonadotropin deficiency	Direct sequencing of POMC
PCSK1	Homozygous, Compound heterozygous	19 patients reported	Hyperphagia, Obesity, Intestinal dysfunction, Postprandial hypoglycemia, Central hypothyroidism, Hypogonadotropic hypogonadism, Diabetes insipidus	Direct sequencing of PCSK1

Mutations in MC4R, resulting in melanocortin 4 receptor deficiency, are the most common cause of monogenic childhood obesity [50]. These mutations are usually inherited co-dominantly and loss-of-function homozygous mutations result in more severe obesity [51–54]. The bulk of these genomic events are missense mutations, being almost always non-recurrent and unique to single families [51–54], with no evidence of common founder mutations [50]. Individuals with MC4R mutations have an altered ratio between fat mass and lean mass, along with changes in bone strength, growth and feeding behavior [55]. Given that pathogenic MC4R variants are of relatively low prevalence and that many variants are non-pathogenic, routine clinical MC4R testing is not typically recommended [48]. The clinical picture is complicated further by the penetrance of MC4R-linked childhood obesity being sometimes incomplete and clinical expression being variable [28]. Furthermore, a recent study revealed gain-of-function MC4R variants conferring protection from obesity [56].

Leptin, encoded by LEP, and its receptor, encoded by LEPR, are key regulators of several endocrine functions [48]. LEP and LEPR mutations are very rare and only normally observed in consanguineous families [14]. Transmitted in an autosomal recessive fashion, the c.398delG homozygous frameshift mutation in LEP results in leptin deficiency, leading to a non-secreted truncated protein, while other homozygous mutations lead to diminished levels of leptin in circulation [49, 50]. Clinical characteristics of leptin deficiency include severe obesity, short stature, hyperphagia, emotional lability and social disability [48]. Leptin receptor deficiency is also an autosomal recessive disorder and, at least in some cases, results from a splicing mutation of LEPR which truncates the gene product prior to the transmembrane domain [50]. To date, approximately seven homozygous mutations in LEPR have been reported in severe childhood obesity cases [48]. In addition to obesity, clinical characteristics of LEPR mutations include hyperphagia, impact on immunity and gonadal changes influencing onset of puberty [48].

The secretion of pro-opiomelanocortin (POMC) from hypothalamic neurons is vital for energy balance regulation and neuroendocrine function [48]. POMC is a precursor for pituitary adrenocorticotropic hormone and alpha-melanocyte-stimulating hormone [48]. Mutations in and hypermethylation of POMC leads to a lack of alpha-melanocyte-stimulating hormone and therefore inhibits food intake through interactions with MC4R in the hypothalamus [48]. POMC deficiency is an autosomal recessive disease driven by loss-of-function mutations in POMC [50, 57]. The resulting clinical presentation is severe early-onset obesity [48]. A dosage effect of POMC has been reported [50], where heterozygous individuals have intermediate weight gain, less than those with homozygous mutations.

Proprotein convertase 1, encoded by PCSK1, is a neuroendocrine enzyme responsible for processing POMC [48]. Rare mutations in PCSK1 disrupt normal functions of POMC and other prohormones and neuropeptides such as insulin and glucagon [48]. Biallelic mutations leading to PCSK1 deficiency are inherited in an autosomal recessive manner [58]. Characteristics of PCSK1 deficiency include very early-onset obesity, moderate hyperphagia and postprandial hypoglycaemia [48]. Common variation at the PCSK1 locus has also been reported to be associated with obesity through case-control studies [59].

4.0. Polygenic Childhood Obesity

Polygenic obesity is the most common form of the childhood disease [61]. The term polygenic refers to cases where obesity is caused through the influence of susceptibility variants in multiple genes, with each having a relatively small effect. Genome-wide association studies (GWAS) represent a global method to identify associations between genetic loci and a trait influenced by many common susceptibility variants, such as polygenic childhood obesity. A limitation of GWAS is that they use single nucleotide polymorphisms (SNPs) to uncover alleles associated with traits of interest or within the same haplotype as a variant associated with a phenotype. This means that GWAS do not generally identify causal variants, but identify a region of linkage disequilibrium containing a potentially causal event. Follow-up studies, such as expression quantitative trait loci mapping, chromatin conformation experiments and targeted clustered regularly interspaced short palindromic repeats (CRISPR) validation, are therefore required to fully gain mechanistic insights from GWAS signals.

Across the last fifteen years, hundreds of BMI loci have been uncovered through GWAS efforts, with an emphasis on the adult setting [62]. Insulin-induced gene 2 (INSIG2) was the debut obesity locus, observed in both adults and children [63], although attempts at validating this association in other cohorts resulted in inconsistent observations [64–68]. However, the association at the next reported locus, namely the ‘fat mass- and obesity-associated’ (FTO) [69] gene, has consistently been replicated by other groups [70–73], including our own [74]. Furthermore, common variation at the MC4R locus represented one of the early, stronger loci revealed by GWAS [75], complementing the monogenic observations described above.

While GWAS has revealed a multitude of adult BMI and obesity loci, the genetic architecture of childhood obesity is far less characterized, despite there being what appears to be a large overlap [3]. Following our investigations of individual adult loci in the context of childhood obesity and BMI [76–80], assessment of copy number variation [81] and initial GWAS efforts in European ancestry of both childhood obesity [82] and pediatric BMI treated as a quantitative trait [83], our latest trans-ancestral meta-analysis of childhood obesity GWAS represents the largest such effort to date [3] and was conducted in the context of the Early Growth Genetics (EGG) consortium. This study was performed to gain greater insights into obesity in the pediatric setting. Using 30 studies of up to 13,005 cases and 15,599 controls derived from multiple ancestry groups, we found 18 previously implicated obesity loci and one novel locus (Table 3) [3]. This multiethnic approach not only facilitated the discovery of a new childhood obesity locus; it also allowed us to fine-map a number of known loci given the linkage disequilibrium pattern differences between the different ancestral populations studied [3]. Indeed, we could refine the search space for the causative variant down to fewer than 10 SNPs at four known loci, MC4R, FAIM2, SEC16B and GNPDA2, leading to an informative list referred to as a 99% credible set [3]. As study sizes continue to increase and as more non-European ancestry populations are included, GWAS will gain strength to identify more causal loci for polygenic childhood obesity.

Table 3.

Top 10 Loci Associated with Childhood Obesity [3]

Locus Name	Marker SNP	Risk Allele	Non-Risk Allele	Bayes’ Factor
FTO	rs56094641	A	G	31.88
TMEM18	rs7579427	A	C	20.25
SEC16B	rs539515	A	C	18.07
FAIM2	rs7132908	A	G	16.39
ADCY3	rs4077678	C	G	13.38
MC4R	rs6567160	T	C	13.20
TNNI3K	rs10493544	T	C	11.81
TFAP2B	rs2206277	T	C	11.63
GNPDA2	rs925494	T	C	8.57
BDNF	rs17309874	A	G	8.52

Clearly childhood obesity is not just one trait. Apart from the obvious mutations seen in extreme forms of syndromic obesity, different ways of assessing BMI and obesity across childhood have been employed. Although many of these approaches consistently reveal the ‘usual suspects,’ in particular the FTO locus, each of the different treatments of the trait definitions have yielded additional genetic insights. In parallel to our childhood obesity EGG efforts, carrying out GWAS on the dichotomously defined trait and corresponding controls described above, the EGG has also conducted GWAS of pediatric BMI treated as a continuous trait. In that setting, we revealed three novel loci, ELP3, RAB27B and ADAM23, which were not apparent in the dichotomous setting [83]. Limiting GWAS of BMI to just those subjects in early life and childhood revealed the FAM120AOS locus [84]. Going a step further, the EGG leveraged BMI data from pediatric subjects with measures at multiple time-points, allowing us to conduct longitudinal analyses, revealing an additional locus, LEPR/LEPROT [85]. As such, leveraging various strategies to further explore the definition of the trait during childhood, including accounting for measures for such factors as leptin and adiponectin [86], hold promise to reveal even further loci relevant to childhood obesity.

5.0. Finding Causal Common Variants from GWAS Signals

Although initial bioinformatic-based efforts with GWAS loci related to obesity have implicated central nervous system processes [87, 88], beyond identifying associated loci and fine-mapping to narrow lists of credible sets of SNPs, more efforts are required to elucidate the actual causal variants at GWAS signals. Virtually all GWAS signals are located in non-coding genomic regions [89] and therefore likely reside in cis-regulatory elements that operate as enhancers or silencers. As such motifs can function across large linear distances, chromosome conformation experiments can be used to determine physical interactions between non-coding variants and gene promotors to implicate the correct putative causal gene. A variety of computational analyses can be carried out to predict the functional role of non-coding variants. The Ensembl Variant Effect Predictor can assess if certain functional groups are over-represented [3]. Colocalization with publicly available Genotype-Tissue Expression (GTEx) project data determines the likelihood of whether a suggestively associated variant shares the same underlying culprit variant as an expression quantitative trait locus for a particular gene in the region [3]. Public resources can also provide epigenomic annotations and knowledge of regulatory motif conservation [89]. Epigenomic annotations, including DNA methylation patterns, DNase I hypersensitivity sites and histone modifications, can be used to imply functional or regulatory potential in a cell type- or tissue-type specific manner [30]. After determining high confidence causal variants based on a confluence of evidence, targeted functional studies can confirm the identity of effector genes at GWAS loci and investigate their molecular mechanism.

Given that the FTO locus is consistently associated with obesity in both children and adults, it is the most extensively studied and therefore serves as a good example of how to translate GWAS signals to biological discovery [82, 90]. Despite early high profile efforts to characterize how FTO confers its effect on obesity risk [91, 92], the identity of the actual effector gene(s) operating at this locus has been a point of great discussion. This key childhood obesity GWAS signal resides in introns 1 and 2 of FTO [89]. Epigenomic data has been leveraged to infer where the causal variant resides, which was then validated in reporter assays [89]. Chromatin interactions in the FTO genomic region implicated candidate effector genes which were supported by expression quantitative trait locus analyses. Cellular processes affected by the causal variant were predicted on the basis of correlated expression. In the same study, the variant implicated in the dysregulation of expression of the potential effector genes was predicted with a quantitative analysis of regulator-motif conservation [89]. All predictions were then validated with the CRISPR-Cas9 system in patient cells and mouse models. Taken together, this team determined that rs1421085 impacts a conserved ARIB5B repressor motif, leading to a pronounced impact on expression of both IRX3 and IRX5 during early-stage adipocyte differentiation [89]. Furthermore, when IRX3 and IRX5 expression increased, it influenced adipocyte fate among other processes [89]. As such, this signal harbored within FTO influences expression of IRX3 and IRX5, as well as growing evidence for the other neighboring RFGRIP1L gene [93], as opposed to the FTO gene itself [89, 94].

6.0. Polygenic Risk Scores

A polygenic risk score (PRS) incorporates the weights of respective risk alleles determined by GWAS and addresses the overall risk burden within a given individual [95]. A key study leveraged genotyping and sequencing data available in a large cohort of individuals, along with a strong replication strategy, to reveal subjects at higher risk of presenting with obesity using a PRS [96, 97]. Although the PRS was not particularly informative for birth weight and in very early life, by the age of 8 years old this approach was much more predictive [97]. This highlights the limitation of using a PRS for predicting childhood obesity risk across the entire age range. This is especially notable, given that the optimal timeframe of opportunity for prevention of childhood obesity is between infancy and early childhood [97]. This PRS may also not be useful in non-European populations [97]. Although other clinical factors, such as birth weight and gestational age, have predictive power for childhood obesity risk [97, 98], there is an appeal to leveraging genetic risk to drive prevention strategies; however, its utility in intervention has not been fully assessed [97].

7.0. Overlap Between Forms of Childhood Obesity

Increasing evidence suggests that forms of syndromic and monogenic childhood obesity exist on a continuum [99]. SNPs in or near syndromic obesity genes have been reported by GWAS, including BDNF, NTRK2, SIM1, BBS2, BBS4, SH2B1 and SDCCAG8 [62]. Many loci at non-syndromic monogenic childhood obesity loci, including POMC, PCSK1 and MC4R, have also been associated with polygenic childhood obesity [100]. A recent study involving single-marker, tag-SNP and gene-based analyses demonstrated that GWAS-implicated SNPs mapped to 17 of 54 candidate syndromic childhood obesity genes [99]. The utility of their candidate gene approach therefore showed an overlap between genes and pathways functioning in both syndromic childhood obesity and non-syndromic polygenic obesity [99].

8.0. Therapeutic Developments

Identifying genetic factors influencing childhood obesity risk offers the promise of tailored advice for patients and their families [47]. As novel therapeutics for genetically defined obesities become available, albeit there is a paucity of therapeutics currently, the identification of a specific defect at the molecular level is very important [47]. While routine clinical genetic testing is not always recommended for childhood obesity cases, this may need to change as the prevalence of individualized therapies increase. Monogenic causes offer insights for understanding hormonal and neural regulation of adiposity, identifying pathways driving the pathogenesis of common obesity and providing targets for therapeutic intervention [50]. A number of therapeutic agents are being developed, including setmelanotide [101]. Setmelanotide is a melanocortin receptor agonist which influences feeding behaviour in the brain, with an independent impact on insulin sensitivity [102]. It is already being evaluated for treatment of many of the monogenic obesity disorders described above, including Prader-Willi and Alström syndromes [103–105].

9.0. Conclusions and Perspectives

Childhood obesity is an incredibly complex trait, impacted by environmental factors and genetics. Elucidating how genetics fuel its pathogenesis is crucial, as the prevalence of pediatric obesity continues to increase in both in the United States and globally. It has been demonstrated that genetic factors contributing to adult and childhood obesity largely overlap, increasing the utility of genetic studies in a pediatric setting where environmental factors have had less time to have an impact. Conclusions drawn from childhood obesity studies may be informative for obesity genetics across the lifespan. Syndromic and monogenic forms of childhood obesity can provide clues for which genes and pathways should be targeted with therapeutic agents. The power to detect causal loci for polygenic forms of childhood obesity are increasing as GWAS samples sizes increase and more non-European populations are included. Trans-ancestral meta-analysis of GWAS provide opportunities to identify novel loci and fine-map known loci, making it easier to determine which common variants are causal at implicated loci. The functional studies performed at the FTO locus demonstrate the types of experiments needed to elucidate the molecular mechanisms of action at GWAS-implicated loci.

In the vast majority of childhood obesity cases, heritable factors are not considered when determining treatment, which usually only involves lifestyle intervention. The future of prediction, diagnosis and treatment relies on greater understanding of the genetic determinants of childhood obesity.

Key Points.

• Childhood obesity cases present in syndromic, monogenic and polygenic forms, with all having a pronounced genetic component

• Greater understanding of the genetic causes of childhood obesity will allow for advances in prediction, diagnosis and treatment