Food Consumption as a Modifier of the Association between LEPR Gene Variants and Excess Body Weight in Children and Adolescents: A Study of the SCAALA Cohort

No studies showing that food consumption is a modifier of the association of variants of the leptin receptor gene (LEPR) with body weight have involved a Brazilian population. The aim of this study was to evaluate the modifying effect of dietary intake on the association between the LEPR gene and excess weight. In this study, 1211 children and adolescents aged 4–11 years were assessed. Participants were genotyped for 112 single-nucleotide variants of the LEPR gene. Anthropometric measurements were performed, and dietary data were obtained. Logistic regressions were used to study the associations of interest. Of the participants, 13.4% were overweight/obese. The risk allele (G) of the rs1137100 variant was associated with excess weight in individuals with fat consumption below the median (odds ratio OR = 1.92; 95% confidence interval CI = 1.18–3.14), with daily frequency of consumption of drink/artificial juice (OR = 2.15; 95% CI = 1.26–3.68) and refined cereals (OR = 2.17; 95% CI = 1.31–3.62) above the median. The risk allele (G) of variant rs1177681 was also associated with excess weight (OR = 2.74; 95% CI = 1.65–4.57) in subjects with a daily frequency of refined cereal consumption above the median. The association between LEPR and excess weight can be modulated by the type and distribution of dietary fatty acids, sugary drinks, and refined cereals.

and covering a variety of socioeconomic levels and environmental conditions. For this study, we also excluded 234 participants because of lack of genetic data (n = 136), lack of data on the status of excess weight (n = 10), lack of kin data (n = 61), and lack of data on food consumption (n = 27). According to these exclusions, our final sample consisted of 1211 children and adolescents.

Ethical Issues
The parents or legal guardians of each participating child signed an informed consent form in which the study procedures were described in detail. The study protocol was approved by the internal ethics committee of the Collective Health Institute of the Federal University of Bahia and by the National Research Ethics Council (CONEP) under references 003-05/CEP-ISC and 15.895/2011, respectively.

Anthropometric Data
The weight was measured in grams and height in cm at the baseline. Weight was assessed using a portable electronic scale (Filizola ® , model E-150/3P, São Paulo, Brazil), and height was measured with a portable stadiometer (Measure ® Seca, Hamburg, Germany). Anthropometric measures were obtained according to standardized techniques performed by trained interviewers. To evaluate anthropometric status, the 2006 and 2007 reference tables of the World Health Organization (WHO), which are based on a Z-scores for body mass index (BMI) according to sex and age, were employed [17,18]. For statistical analyses, the excess weight was defined as follows: for children under 5 (z > +2 score), for children aged 5 to 10 years (z > +1), and for adolescents (z ≥ +1).

DNA Extraction and Genotyping
DNA was extracted according to the protocol described by Gentra ®® Puregene ®® Blood Kit Qiagen (Germantown, ML, USA), while quantification was standardized at a concentration of 50 ng/µL and identified in barcoded tubes. This step was performed using a Qubit fluorometer ® (Invitrogen, Paisley, UK) [19]. The LEPR genetic information was extracted between positions 65,886,335 and 66,103,176 on chromosome 1 and genotyping for all 149 SNVs of the LEPR gene was conducted on the HumanOmni2.5-8 platform using the BeadChip kit (Illumina, San Diego, California, USA).

Quality Control
A genetic quality control was performed before conducting the association tests. All procedures were performed with the help of PLINK v.1.9 (https://www.cog-genomics.org/plink/1.9/) [20]. To evaluate family structure, the kinship coefficients for each possible pair of this analysis were estimated. Sixty-one individuals were removed from the sample because of their degree of kinship. Quality control of the SNVs was performed in stages: by excluding SNVs with genotyping call rates under 0.98, a Hardy-Weinberg equilibrium (using only controls) with a p-value below 0.05, and a minor frequency allele (MAF) under 1% [21]. In candidate gene studies, the withdrawal of SNVs with very low frequency (1-2%) is common.
To ensure the quality of the study, for small samples, it is recommended that MAF cut-off values be increased [22]. After the quality control, 112 SNVs were available for analysis.

Population Structure
A principal component analysis (PCA) was performed to identify a possible population structure (population groups differentiated due to the ancestry/origin of each individual). Details on the PCA for population stratification are available in the work of Costa et al., 2015 [19].

Food Consumption
Dietary intake was obtained through food recall for the last 24 h (24 h food recall-24 h) and a food frequency questionnaire (FFQ). Information was obtained from the parent or legal child's guardian and collected by previously trained nutritionists and nutrition academics. Children over eight years of age also provided information on their food consumption outside of the home environment.
The 24-h diet-recall method (R24h) was used to determine dietary intake. Parents reported their children's dietary intake. However, the information given by the children at the time of the interview complemented the information given by their parents. Food consumed in school or at day-care centers was also recorded. The food intake was converted into energy and macronutrient percentages using the Diet Pro program [23]. Foods that were not part of the software database were added using information contained in the Brazilian Food Composition Table [ , and packaged-food labels. The macronutrient intake was expressed as the percentage of the total energy intake. The ratio between polyunsaturated and saturated fat (POLY:SAT) was also considered.

Other Variables
The variables for adjustment were age, sex, population structure, and energy intake. The birth dates of the participants were obtained from their birth certificate, and the ages in years were calculated by subtracting the birth date from the interview date.

Statistical Analyses
The population was characterized by a descriptive analysis. The χ 2 test was employed for categorical variables, and the Mann-Whitney U test was used to compare food intake variables of the "overweight/obese" and "not overweight/obese" groups.
Logistic regression analysis was used to evaluate the association between variants of the LEPR gene and overweightness. Each SNV of the LEPR gene was analyzed separately. Each LEPR gene SNV were determined using logistic regression models. Due to the low number of homozygotes for the risk allele, the genotype was analyzed using a dominant heredity genetic model adjusted for sex, age, population structure (determined by the first three major components), and energy intake. These confounding variables were selected on the basis of data published in previous studies [29,30].
Genomic and proteomic analyses regularly involve the simultaneous testing of hundreds of hypotheses in both numerical and categorical data. To correct for the occurrence of false positives, validation tests based on correction of multiple tests, such as Bonferroni and Benjamini, and Hochberg's false discovery rate, and re-sampling techniques (permutation-based tests) are frequently employed. In this study, we used a permutation test because it has become a widely accepted and recommended approach for studies involving multiple statistical tests for genetic markers [31]. As such, empirical p-values were obtained after 50,000 phenotype permutations to limit the occurrence of type I errors (false-positive results). After the permutation tests, values of p < 0.05 were considered statistically significant.
Additional SNVs associated with overweightness/obesity in the SCAALA population were added to the interaction analyses, in addition to three LEPR SNVs (rs1137100, rs1137101, and rs8179183); these variants are commonly associated with excess weight, and have been frequently mentioned in the literature [8,[32][33][34]. The interactions between LEPR gene SNVs and dietary intake were tested with respect to their effects on overweight/obesity, including the product terms in the models. The effect modification was analyzed by a likelihood-ratio test after estimation (lrtest).
To illustrate the interaction, the association between LEPR gene variants and excess weight was stratified by food consumption. To achieve this stratification, we considered calorie contribution percentages above or below the medians provided by carbohydrates, proteins, and total fat. The median daily intake frequencies of the 11 food groups were also the cut point for consideration in the analyses. The ratio between polyunsaturated and saturated fat (POLY:SAT) was also an independent variable.
All statistical tests were two-tailed, and the significance level considered was 5%. Statistical analyses were performed using PLINK version 1.9 [20] and STATA version 12.0. (Colllege Station, TX, USA).

Characteristics of Participants and Dietary Intake
The eligible study population was 1445 children aged 4-11 years, 1211 of whom were included in the descriptive analysis. Of the children included in the study, 13.4% had excess weight (8.8% overweight, 4.6% obese). A slightly higher percentage was observed in boys compared with girls (53.8 vs. 46.2%). The median energy intake was 1651.78 (58.13-19307.12) Kcal. The median energy intake levels from carbohydrates, proteins, and fats were 61.58% (range: 27.08-87.38%), 12.60% (range: 4.28-82.69%), and 26.06% (range: 5.28-51.28%), respectively. The median intake of the polyunsaturated and saturated fat ratio (POLY/SAT) was 0.95 (range: 0.00-3.89). Significant differences were observed in total energy (p = 0.003), energy from protein (p = 0.034), and fat (p = 0.022) percentages between the groups studied ( Table 1). The data also showed statistically significant differences in the median daily intake frequencies of milk and dairy products (p = 0.000), and processed meats (p = 0.042) groups amongst those who were overweight when compared with the non-overweight group. Additional information is provided in Table 2.

The Role of Diet in the Association between LEPR Variants and Overweight/Obesity
An effect modification analysis of food intake regarding the association between LEPR gene variants and overweightness was performed (Table 4). An interaction was observed between fat intake (the ratio between PUFA:SFA) and the LEPR gene variant rs1137100 (interaction p = 0.049). The risk allele (G) for the rs1137100 variant was positively associated with overweightness/obesity in individuals whose POLY:SAT ratio was below the median (OR = 1.92; 95% CI = 1.18-3.14). Regarding the analyses run for the food groups, an interaction was found between daily soft drinks/artificial juices intake frequency and LEPR gene variant rs1137100 (interaction p = 0.019) relative to overweight/obesity. The risk allele (G) of the rs1137100 variant was positively associated with overweight/obesity in individuals whose daily soft drink/artificial juice intake frequency was above the median (OR = 2.15; 95% CI = 1.26-3.68). In addition, interactions between daily refined cereal intake frequency and LEPR gene variants rs1177681, rs1137100, and rs8179183 (interaction p < 0.001, 0.011 and 0.005, respectively) were observed. Positive associations were found between the risk alleles of the variants studied, and overweight/obesity in individuals whose daily refined cereal intake frequency was above the median: rs1177681 (G) (OR = 2.74; 95% CI = 1.65-4.57), rs1137100 (G) (OR = 2.17; 95% CI = 1.31-3.62). The risk allele (C) of variant rs8179183 was positively associated with overweightness/obesity in individuals whose refined cereal intake frequency was below the median (OR: 1.75; 95% CI: 1.06-2.90). Additional information is provided in Supplementary Tables S2 and S3.

Discussion
The prevalence of excess weight 13.4% (8.8% overweight, 4.6% obesity) in this study is lower than those recorded by the Family Budget Survey (POF) (Pesquisa de Orçamento Familiar) 2008-2009, in which the prevalence of overweightness fluctuated between the different regions of Brazil. Among children of 5 to 9 years, 33.5% were overweight and 14.3% were obese, and among adolescents aged 10-19 years, 20.5% were overweight and 4.9% were obese [1]. The criteria adopted for the diagnosis of overweight/obesity, the differences in the studied age groups and the cultural differences in each region may justify the variability observed in the studies. However, regardless of this variability, the prevalence of overweightness and obesity among children and adolescents in all macro-regions of the country, especially in the southeast, south, and center-west regions, stands out in general.
The value of this study derives in part from the characteristics of the population in which it was conducted. The population of Salvador is highly miscegenated. Lima- Costa et al. (2015) showed that 50% of the genetic composition of the Salvador-SCAALA cohort is of African origin [35] with the remainder being of European and Amerindian origin. For the LEPR gene from which the SNVs were evaluated, four variants were found to be positively associated with excess weight: rs115650230, rs116239759, rs202069668, and rs79353784. However, the rs78005150 variant showed a negative association with excess weight. Variants of the LEPR gene influencing obesity and obesity-related traits have been studied in children and adolescents in India [8], Spain [9], Mexico [36], [32] with the variants rs1137101, rs1137100, and rs8179183 being the most commonly studied. However, other studies, conducted with children of the same age group in Poland [12], Japan [37], Turkey [11], Malaysia [33], Denmark [34], and Brazil [38,39], failed to identify any significant association. Already in a study with young adults and Danish children [34], a negative association was observed. The heterogeneity between the results of association between overweightness/obesity, and different variants of the LEPR gene in different populations may depend on the ethnicity or the differences in the gene-environment interaction.
The role of diet in modulating the association between LEPR gene variants and overweight/obesity was examined through logistic regression analysis on food consumption in a well-characterized cohort of children and adolescents in Salvador, Bahia, Brazil. We found that a POLY:SAT ratio below the median potentiated the association between the risk allele of LEPR variant rs1137100 (G) (OR = 1.92; 95% CI = 1.18-3.14) and overweight/obesity. In contrast, a soft drinks/artificial juices intake frequency above the median maximized the obesogenic effect of the risk allele of variant rs1137100 (G) (OR = 2.15; 95% CI = 1.26-3.68). Similar results were observed when the intake frequency of the refined cereal group was above the median for SNVs rs1177681 (G) (OR = 2.74; 95% CI = 1.65-4.57), rs1137100 (G) (OR = 2.17; 95% CI = 1.31-3.62).
The literature offers few studies that evaluate the modifying effect of diet on the association between LEPR and overweightness/obesity, especially in children and adolescents. Dominguez-Reyes et al. [15] found an association in young Mexican individuals encoding alleles (AG + GG) for variant rs1137101 with saturated fat intake (SAT) ≥12 g/d (OR = 2.9; 95% CI = 1. 5-5.8), and a greater risk of overweight/obesity when the total fat intake was ≥ 83 g/d (OR = 3.0; 95% CI = 1.5-6.2, p < 0.001). The beneficial effects of fatty acids, especially of polyunsaturated fatty acids (PUFA), on the modulation of the association between LEPR and overweight/obesity, has been noted by Jourdan et al. [40]. Likewise, the benefit of PUFA has been observed for polymorphisms of other genes, such as FTO [30,41], and studies have indicated an "anti-obesity" effect from PUFA. This finding is supported by the notion that unsaturated fats lead to greater diet-induced thermogenesis, energy expenditure, or fat oxidation, compared with saturated fatty acids (SFA) [42]. Multiple researchers have shown that fat oxidation is directly proportional to the PUFA:SFA ratio [43,44], which justifies the testing of this relationship. The mechanisms that allow dietary SFA to interact with LEPR are unknown; however, both the quality and quantity of dietary fat have been shown to influence the methylation of CpG regions [45]. Methylation consists of the addition of methyl group to cytosine (C), usually at CpG dinucleotides [46].
Our study also shows a significant interaction between higher soft drink/artificial juice intake frequency, and genetic predisposition related to overweight and obesity. The association between the frequent consumption of sugary beverages and overweightness in children and adults has been demonstrated in different prospective cohort studies, and randomized clinical trials that were reported in a systematic review and meta-analysis [47]. In addition to BMI and obesity, direct associations between sugary drink ingestion and waist circumference have been shown for these populations [48,49]. The ingestion of sugary drinks contributes to obesity through several potential mechanisms, including a high caloric content and reduced satiety. As liquids do not activate the inhibitory ingestion mechanisms in subsequent meals, increased total energy intake results [50]. Furthermore, the increasing portion sizes offered, combined with the presence of high concentrations of rapidly absorbed carbohydrates, such as sucrose or fructose-rich corn syrup, may increase the risk of visceral adiposity and other metabolic alterations [51]. Sugary drink ingestion may lead to increased genetic susceptibility to obesity, as seen from results obtained with three large cohorts in the United States of America [52]. These results were replicated in two other cohorts that showed an association between genetic variants with BMI and the incidence of obesity (based on 32 loci of GWAS associated with BMI) [53,54]. The effect was more pronounced amongst adult individuals with high sugary drink intake frequency, compared with those with infrequent consumption. In addition, an effect modification of these beverages on the association between genetic variants and abdominal obesity was found by Olsen et al. (2016) [55]. These and other findings suggest that the consumption of sweetened beverages potentiates the effect on weight gain in individuals who are genetically predisposed to storing body fat [53]; however, the mechanisms by which these relations take place are not clear. Studies involving the effect of these beverages on the association between the LEPR gene and weight gain in children and adolescents were not identified, which highlights the novelty of our findings.
This study showed an obesogenic effect from LEPR gene variants amongst those with increased refined cereal consumption frequency, which is a proxy for carbohydrates (CHO), with the exception of variant rs8179183. Although studies evaluating the modulating effects of specific carbohydrate sources in the relationship between LEPR and obesity were not identified, studies reporting the interactions between different genetic variants and carbohydrate intake are available [56,57]. Marti et al. (2006) [56] observed a higher risk between SNV pro12ala of the peroxisome proliferator-activated receptor gamma-gene (PPARy) and obesity (OR = 5.12, p < 0.04, 95% CI: 1.01-25.80) amongst individuals with higher CHO consumption (>49% energy). Likewise, Martinez et al. [57] observed that women with the Gln27Glu polymorphism of the β2-adrenergic receptor gene (ADRB2) with higher CHO intake (>49% energy) showed a higher risk of obesity (OR = 2.56, p = 0.051) than those with lower CHO intake, although with borderline statistical significance. Certainly, the cereal refinement process-which increases caloric density by 10%, reduces dietary fiber by 80%, and reduces the amount of dietary protein by almost 30%-may disqualify CHOs, and may be associated with insulin and leptin resistance, and body weight gain [58]. However, the genetic susceptibility to cereal-dependent weight gain still requires scientific investigation.
The literature shows that the leptin (LEP) and LEPR genes have been defined as a biological pathway for the regulation of food intake and energy expenditure [10]. For leptin to promote the appropriate neural response, it needs to bind to specific receptors (of the ObRb type) on the cell surface, promoting the activation of the JAK2/STAT3 system (Janus Kinase 2/signal transducer and activator of transcription 3), which regulates the synthesis of different neuropeptides that play important roles in the orexigenic and anorexigenic system involved in the control of food intake and energy balance [59]. Mutations in the leptin receptor gene compromise the transmission of the signal from the leptin binding to the receptor into the cell, and lead to leptin resistance, which is the reduced capacity of leptin to regulate appetite and weight gain [7].
Although the mechanisms involved in hypothalamic leptin resistance have been clarified, little is known about the effects of dietary components on the development of leptin resistance. Recently, new data have emerged from experimental animal studies indicating that specific types of dietary sugars or fats are capable of inducing leptin resistance in the absence of high levels of circulating leptin and/or body fat [60]. Shapiro et al. (2008) demonstrate that the dietary intake of fructose alone or in combination with high-fat diets contribute to obesity through two mechanisms: (1) induction of hyperphagia driven by increased palatability, and (2) blockade of leptin transport through the blood-brain barrier [61]. Despite advances in this field, studies dealing with this topic in human populations are still rare, and further studies are required to accurately define the role of the LEPR gene in the development of overweight and obesity.
Our study was based on an analysis of cross-sectional data, which limits the capacity to investigate causality. Furthermore, it was not possible to examine other adiposity measures, limiting this study to consider only BMI, which cannot distinguish body composition, and does not provide any indication of body fat distribution. The R24 h method employed in the present study to investigate food intake may also represent another limitation. Although the R24 h method is fast, relatively inexpensive, and easily applied, its success depends on the memory of the respondent and requires a well-trained investigator to obtain accurate estimates of the portions consumed. Nevertheless, this method provides reliable estimates of the average dietary intake of a population even when applied a single time, provided that the designated methodology is followed and the analytical resources are appropriate [62]. Moreover, most of the children included in this analysis are predominantly of African descent; therefore, our results cannot be generalized to other ethnic groups. In contrast, the strengths of this study include the interaction analysis with dietary intake and the analysis of a varied number of SNVs of the LEPR gene.

Conclusions
The association between LEPR and overweightness/obesity can be modified by the dietary characteristics, especially by the types and distributions of dietary fatty acids, sugary drinks, and refined cereals. The results of this study offer new insights into the interrelationships between the genetic variants of LEPR, and dietary intake and obesity. Nevertheless, further studies are needed to clarify the mechanisms involved in these relationships.