Hostname: page-component-7c8c6479df-7qhmt Total loading time: 0 Render date: 2024-03-29T00:24:20.755Z Has data issue: false hasContentIssue false

Heritability of Children's Dietary Intakes: A Population-Based Twin Study in China

Published online by Cambridge University Press:  15 August 2016

Ji Li
Affiliation:
Department of International Health, Johns Hopkins University Bloomberg School of Public Health, Baltimore, MD, USA
Huijuan Liu
Affiliation:
Jiaxing Maternity and Child Health Care Hospital, Jiaxing, China
Terri H. Beaty
Affiliation:
Department of Epidemiology, Johns Hopkins University Bloomberg School of Public Health, Baltimore, MD, USA
Hua Chen
Affiliation:
Jiaxing Maternity and Child Health Care Hospital, Jiaxing, China
Benjamin Caballero
Affiliation:
Department of International Health, Johns Hopkins University Bloomberg School of Public Health, Baltimore, MD, USA
Youfa Wang*
Affiliation:
Department of International Health, Johns Hopkins University Bloomberg School of Public Health, Baltimore, MD, USA Department of Epidemiology, Johns Hopkins University Bloomberg School of Public Health, Baltimore, MD, USA Department of Health, Behavior and Society, Johns Hopkins University Bloomberg School of Public Health, Baltimore, MD, USA Systems-Oriented Global Childhood Obesity Intervention Program, Department of Epidemiology and Environmental Health, University at Buffalo, State University of New York, Buffalo, NY, USA
*
address for correspondence: Youfa Wang MD, PhD, MS, Professor and Director, Systems-Oriented Global Childhood Obesity Intervention Program, Department of Epidemiology and Environmental Health, University at Buffalo, State University of New York, Kimball Tower, Buffalo, NY14214-8001, USA. E-mail: youfawan@buffalo.edu

Abstract

Background: Despite evidence for some genetic control of dietary intake in adults, there is little evidence of how genetic factors influence children's dietary patterns. Objective: To estimate heritability of dietary intake in twin children from China and test if genetic effects on dietary intakes vary by the children's socio-economic status (SES). Methods: A sample of 622 twins (162 monozygotic and 149 dizygotic pairs; 298 boys and 324 girls aged 7–15 years) was recruited in South China. Dietary intakes were assessed using a validated 145-item semi-quantitative food frequency questionnaire. Pooled and sex-specific dietary patterns were identified using factor analysis. Heritability was estimated using structural equation models. Results: Heritable components differed by gender and for nutrients and food groups; and estimated heritability of dietary patterns was generally greater in girls than boys. In boys, estimated heritabilities ranged from 18.8% (zinc) to 58.4% (fat) for nutrients; and for food group, 1.1% (Western fast foods) to 65.8% (soft drinks). In girls, these estimates ranged from 5.1% (total energy) to 38.7% (percentage of energy from fat) for nutrients, and 12.6% (eggs) to 94.6% (Western fast foods) for food groups. Factor analysis identified five food patterns: vegetables and fruits, fried and fast foods, beverages, snacks and meats. Maternal education and family income were positively associated with higher heritabilities for intake of meat, fried, and fast food. Conclusions: Genetic influence on dietary intakes differed by gender, nutrients, food groups, and dietary patterns among Chinese twins. Parental SES characteristics modified the estimated genetic influence.

Type
Articles
Copyright
Copyright © The Author(s) 2016 

Dietary intakes not only affect growth (Formica & Regelson, Reference Formica and Regelson1995; Silventoinen et al., Reference Silventoinen, Hasselbalch, Lallukka, Bogl, Pietilainen, Heitmann and . . . Kaprio2009), body functions (Cameron et al., Reference Cameron, Paton, Nowson, Margerison, Frame and Wark2004), cognitive and behavioral development (Kar et al., Reference Kar, Rao and Chandramouli2008) in children, but also have long-term effects on health conditions in adulthood (Kemm, Reference Kemm1987; Moore et al., Reference Moore, Singer, Bradlee, Djousse, Proctor, Cupples and Ellison2005). Childhood is a key time window for forming life-long eating behaviors. Without effective intervention, poor eating habits in early life could ‘track’ into adulthood and have long-term health consequences (Li & Wang, Reference Li and Wang2008; Wang et al., Reference Wang, Bentley, Zhai and Popkin2002; Zive et al., Reference Zive, Berry, Sallis, Frank and Nader2002).

Multiple factors influence children's dietary intake, which remain not well understood, in particular, regarding genetic influence. Family factors play an important role in affecting children's eating behaviors, which was partly reflected in the resemblance in diets between parents and their children (Beydoun & Wang, Reference Beydoun and Wang2009; Kitzman-Ulrich et al., Reference Kitzman-Ulrich, Wilson, St George, Lawman, Segal and Fairchild2010; Wang et al., Reference Wang, Li and Caballero2009). Observed parent–child correlations in dietary intakes may be due to shared genetics and environments. Population-based studies cannot distinguish genetic and environmental effects, whereas the twin study design can estimate the contribution of unobserved genetic factors to phenotypic variance of quantitative traits such as dietary intakes (Neale et al., Reference Neale and Cardon1992).

Twin studies have studied the heritabilities of dietary intakes among adults (Heitmann et al., Reference Heitmann, Harris, Lissner and Pedersen1999; Keskitalo et al., Reference Keskitalo, Silventoinen, Tuorila, Perola, Pietilainen, Rissanen and Kaprio2008; Teucher et al., Reference Teucher, Skinner, Skidmore, Cassidy, Fairweather-Tait, Hooper and MacGregor2007); however, such studies among children are scarce and have reported inconsistent findings. One study among twins aged 4–5 years in the United Kingdom showed that estimated heritability of food preference was 20% for dessert, 37% for vegetables, 51% for fruits, and 78% for meat and fish (Breen et al., Reference Breen, Plomin and Wardle2006). Based on single 24-hour dietary recall data, a recent study of U.S. twins aged 7 years found apparent genetic effects varied by food groups and sex (Faith et al., Reference Faith, Rhea, Corley and Hewitt2008). The estimated heritabilities of consumption of a particular food group ranged from 12% (fish and lemon) to 79% (peanut butter and jelly) among boys, and from 20% (bread and butter) to 56% (fish and lemon) among girls. To our knowledge, no study has been conducted in developing countries, where the food choices are different from those in developed countries, and variation in dietary intakes and in people's dietary preferences may be larger (Darnton-Hill & Coyne, Reference Darnton-Hill and Coyne1998; Khan & Bhutta, Reference Khan and Bhutta2010). Additionally, the genetic composition of the population is different, which may lead to different findings.

This study aimed to quantify potential genetic influence on usual dietary intakes, including food patterns among Chinese children in a twin study. We also tested whether the genetic and environmental effects on children's dietary intakes might vary by family socio-economic status (SES). We hypothesized that SES such as maternal education and family income could affect food environments and parenting styles related to diet, and further influence the genetic effects on dietary intakes in children. We suspected better-educated mothers might assist their children to eat healthier, and thus these children might show a lower heritability of food consumption than their counterparts, whereas children in high-income families might have access to more abundant and various food groups, and thus might show a stronger heritability of food consumption than those from lower-income families.

Materials and Methods

Study Sample

We surveyed 622 twins aged 7–15 years old (n = 311 pairs, monozygotic (MZ) pairs: dizygotic (DZ) pairs≈1:1, male:female ≈1:1) and their mothers (n = 311) in December 2009–January 2010 in Jiaxing, Zhejiang Province, China. The twins were originally enrolled in the China Birth Defects and Child Health Care Surveillance System (BDSS-China), which has collected anthropometric data from each child since birth to age 7 (Li, Moore et al., Reference Li, Moore, Li, Berry, Gindler, Hong and Erickson2003). Twins were recruited if (1) both twins participated in the BDSS-China, and (2) both twins and their mother consented or agreed to participate in the follow-up. Children with complete dietary data were included in the present study. Subjects (n = 56) were excluded mainly because their daily energy intakes were <450 kcal (n = 2) or >8,500 kcal (n = 33), or their co-twin did not have complete dietary data (n = 21). The final sample yielded 283 twin pairs (270 boys and 296 girls).

Data Collection

All twins visited one of the three clinics closest to their home on a designated weekend morning and completed the survey, which included blood draw. The mothers completed a questionnaire about demographic and family information. Children's anthropometry was measured by research physicians from the Jiaxing Maternity and Child Health Care Hospital (JMCHCH) following specific training for this study.

The study protocol was approved by the Institutional Review Boards of the Johns Hopkins University Bloomberg School of Public Health (JHBSPH) and JMCHCH, Jiaxing, China.

Dietary Assessment

Food consumption was assessed using a 145-item self-administered food frequency questionnaire (FFQ) that measured children's food intakes during the previous 12 months. In addition to 117 food items, the FFQ contained 28 questions on cooking methods, edible oil consumption, and so forth. The FFQ was initially designed for children in Beijing and validated against four 24-hour recalls, and showed good reliability and moderate validity (Wang, Reference Wang2009). Subjects provided information on the amount and type of food consumed. Children were also provided with two-dimensional colorful pictures of measuring plates and bowls to facilitate estimation of portion size. Standardized interviews were conducted when necessary to help children better understand the questionnaire. All the interviewers were required to follow a standardized protocol for data collection.

Nutrients and Food Groups

Energy and nutrient intakes were calculated using the Nutrition Data Systems established by the Chinese Centers for Disease Control and Prevention (Yang, Reference Yang2004; Yang et al., Reference Yang, Wang and Pan2002). For micronutrients such as vitamin (Vit) A, Vit C, Vit E, calcium, zinc, and iron, nutrient density (μg/1000 kcal or mg/1000 kcal) was calculated by using 1,000X (absolute value of nutrient daily intake, μg or mg)/(daily calorie intake, kcal). High fat intake was defined as the age-specific top quartile of % energy (E) from fat. Based on similarity in nutrient content, the 117 food items were further categorized into 23 food groups (see Supplementary material). These food groups were used in our factor analysis to identify the dietary patterns.

Zygosity

MZ and DZ twins were ascertained by 19 questions in the mothers’ questionnaire. The questions were adapted from a validated Taiwan twin similarity questionnaire (Chen et al., Reference Chen, Chang, Wu, Lin, Chang, Chiu and Soong1999) and based on other related research (Ooki & Asaka, Reference Ooki and Asaka2004; Rietveld et al., Reference Rietveld, van Der Valk, Bongers, Stroet, Slagboom and Boomsma2000). The accuracy rate of the Taiwan questionnaire was 97.4% for parental reports and 95.6% for adolescent twins’ self-reports compared to results based on DNA diagnosis (Chen et al., Reference Chen, Chang, Wu, Lin, Chang, Chiu and Soong1999). Only one twin pair was asked to answer the questions because their mother did not answer them. Based on the participants’ answers, we identified 146 MZ pairs, 73 same-sex (SS) DZ pairs, and 64 opposite-sex (OS) DZ pairs.

Other Measurements

Family information was collected from the parents. Mothers’ education was grouped into two categories: (1) higher education (22 MZ and 11 same-sex DZ pairs) referred to senior high school or above, and (2) lower education (124 MZ and 62 same-sex DZ pairs) referred to junior high school, elementary school, or below. Annual family income was categorized into two groups: (1) higher income (42 MZ and 21 same-sex DZ pairs) meant income ≥50,000 Yuan (≈7,457 USD) per year, and (2) lower income (102 MZ and 44 same-sex DZ pairs) an income <50,000 Yuan per year.

Statistical Analysis

First, to test if selection bias existed, we compared 566 eligible twins with 56 twins (n = 28 pairs) who were excluded due to incomplete FFQ data for one or both twins. The two groups did not differ significantly in their age, sex, zygosity, weight, height, BMI, maternal education, or family income status (p > .05). Second, we conducted descriptive analysis and analyzed the twin cohort as individuals; t tests, ANalysis Of VAriance (ANOVA), and χ 2 tests were conducted to compare differences between groups in continuous and categorical variables. Shrout–Fleiss intraclass correlation coefficients (ICCs) were calculated in boys and girls respectively using SAS 9.2 (SAS Institute, Cary, NC). ICC measures the degree of similarity in a quantitative trait between pairs of individuals. By comparing MZ and DZ ICCs, we estimated the genetic and environmental influences on each trait.

Third, factor analysis was conducted using SAS 9.2 (SAS Institute, Cary, NC) to identify dietary patterns for consumption of the 23 food groups separately among boys, girls, and in the pooled sample. Uncorrelated factors were derived using Varimax rotation. The number of factors was determined by considering the scree plots of eigenvalues (Supplementary material), as well as whether a factor had meaningful contents. Patterns were named based on the food items with the highest factor scores and the general nutritional content of foods that loaded highly on the pattern (Newby et al., Reference Newby, Weismayer, Akesson, Tucker and Wolk2006). A final five-factor solution was selected for each group.

Fourth, estimated heritabilities of intake of nutrients, food groups, and dietary patterns were computed for each gender using structural equation models (SEMs) with maximum likelihood estimation of the genetic and environmental component. Only MZ and same-sex DZ twins were included in the SEM analysis. If 1/2 rMZ was less than rDZ, ACE models were indicated as discussed below; if 1/2 rMZ was greater than rDZ, ADE models were indicated (Neale et al., Reference Neale and Cardon1992). We used energy adjustment in these models to control for recall bias, which adjusts for children's age when studying macronutrient and micronutrient density, and for age and daily calorie intake when studying food groups and dietary patterns. Genetic models were fitted using the OpenMx statistical package 0.3.3, which allows for variance component decomposition (http://openmx.psyc.virginia.edu). Submodels were compared. Model fit for SEM was assessed based on χ 2 tests. The best-fitting models were those with small χ 2 and high p value. A parsimonious model was identified by using Akaike's information criterion (AIC), and the model with the lowest AIC was preferred.

Total calorie intake was transformed to 100 times the natural log (ln) scale to correct its skewed distribution. The variance is partitioned into an additive genetic component (a), a non-additive genetic component (d), a shared non-genetic component (c), and a random component (e). MZ twins are genetically identical and DZ twins share on average 50% of their genetic background. Therefore, in MZ twins, ai 1 = ai 2 and di 1 = di 2 (where i means the ith twin pair; 1 and 2 means the birth order of this ith twin pair), cov(a i1, a i2) = σ 2 a and cov(di 1, di 2) = σ 2 d (cov means the covariance); in DZ twins, cov(ai 1, ai 2) = σ 2 a/2 and cov(di 1, di 2) = σ 2 d /4 (Falconer & Mackay, Reference Falconer and Mackay1996). The heritability is then estimated as (estimated genetic variance)/(total variance) × 100%.

Finally, to assess whether genetic effects on dietary intakes varied by family SES, we fit univariate Cholesky decomposition models for food groups and dietary patterns stratified by maternal education and family income levels, respectively. The Cholesky decomposition is more frequently used in describing multivariate models. However, univariate Cholesky decomposition models have been used to examine genetic and environmental influences on various traits in previous twin studies as well (Hansell et al., Reference Hansell, Halford, Andrews, Shum, Harris, Davies and Wright2015).

All genetic models were fitted using OpenMx for SEM, adjusting for child's age, sex, and daily calorie intake. Statistical significance of the regression coefficients was set at p < .05; p < .10 was set for marginal significance.

Results

Subject Characteristics

The mean (±SD) age was 11.7±2.6 years in boys (n = 270) and 11.6±2.4 years in girls (n = 296). OS DZ twins (11.3±2.5 years) were younger than MZ (11.8±2.5 years) and SS DZ (11.4±2.6 years) twins (p < .05). Table 1 shows the distribution of nutrient and food group intakes. Compared to girls, boys had higher intakes of zinc, refined grains, meats, eggs, soft drinks, snacks, and western fast foods, but lower intakes of Vit C and vegetables (all p < .05).

TABLE 1 Average Daily Dietary Intakes by Sex in Twin Pairs of Chinese Children Aged 7–15 Years (n = 566)a, b

aBased on FFQ. bDifferences between groups were tested by t-tests. cPercentage of energy from the macronutrient. *Significantly different by sex, p < .05. **Significantly different by sex, p < .01. §Marginally significantly different by sex, p < .1.

Intraclass correlation coefficients (ICCs) and heritabilities of energy and nutrient intakes

Twin-pair ICCs and heritability estimates suggested that genetic factors had important effects on nutrient intake among boys and girls. The results were shown by zygosity status and gender in Table 2. In boys, ICCs were greater for all nutrient measures except for percentage E from protein and zinc among MZ twins. For macronutrients, ICCs ranged from 0.26 (percentage E from protein) to 0.61 (total energy) for MZ twins and from 0 (percentage E from fat) to 0.68 (percentage E from protein) for DZ twins. For micronutrients, ICCs ranged from 0.38 (Vit A) to 0.55 (Vit E) for MZ twins, and from 0 (Vit E) to 0.52 (Zinc) for DZ twins.

TABLE 2 Intraclass Correlation Coefficients (ICCs) and Heritability Estimates of Daily Nutrient and Food Group Intakes in Chinese MZ and DZ Twinsa,b

ICC = intraclass correlation coefficients. aShrout–Fleiss ICCs were calculated using SAS. bHeritability was estimated using ACE model: A, additive genetic effect; C, common environmental effect shared by a twin pair; E, residual environmental effect. The model was fit on each dietary phenotype using OpenMx. Macronutrients and micronutrients were adjusted for age; food groups were adjusted for age and daily calorie intake. ~ = heritability estimate was zero. c100 × log (the variable). dPercentage of energy from the macronutrient.

Because rDZ was >1/2 rMZ for most dietary variables, ACE models were fitted. In the SEMs for heritability estimation, additive genetic components were detected for all dietary elements except percentage E from protein and Vit C after adjustment for age. Final estimates of heritability calculated under ACE models were moderate or strong.

In girls, ICCs were greater for 6 of the 13 variables among MZ twins compared to those in DZ twins. For macronutrients, the average and median correlations were 0.52 and 0.51 in MZ twins and 0.48 and 0.52 in DZ twins, respectively, suggesting that strong correlations not solely due to genes; for micronutrients, the average and median were 0.29 and 0.32 in MZ twins and 0.53 and 0.57 in DZ twins, respectively, again suggesting that genetics cannot explain much of the variation in these traits. Estimated heritability of nutrient intake ranged from 5.1% (total energy; A: 0.05, 95% CI: 0.00–0.58; C: 0.40, 95% CI: 0.00–0.58; E: 0.55, 95% CI: 0.40–0.71) to 38.7% (% E from fat; A: 0.39, 95% CI: 0.00–0.75; C: 0.27, 95% CI: 0.00–0.62; E: 0.34, 95% CI: 0.24–0.50). High fat intake was defined as the age-specific top quartile of %E from fat. Heritability of high fat intake in girls was 16.3% (A: 0.16, 95% CI: 0.00–0.91; C: 0.66, 95% CI: 0.00–0.91; E: 0.17, 95% CI: 0.05–0.41) greater than that estimated in boys (0.0%; A: 0.00, 95% CI: 0.00–0.72; C: 0.56, 95% CI: 0.00–0.79; E: 0.44, 95% CI: 0.21–0.76). Estimated additive genetic components were all non-significant for carbohydrate, %E from protein, Vit A, Vit C, calcium, zinc, and iron.

ICCs and heritabilities of food group intakes

Twin-pair ICCs and heritability estimates for each food group intake are shown by zygosity status and sex in Table 2. In MZ boys, ICC was lowest for eggs (0) and highest for soft drinks (0.69). In DZ boys, ICC was lowest for eggs and Chinese fried foods (0.03) and highest for vegetables and tofu (0.65). After adjustment for age and daily calorie intake, additive genetic components were statistically significant for 2 of 13 food groups (meats and soft drinks). The average heritability was 10.5%, which was low and non-significant.

Although the average ICC over all traits was 0.35 for MZ twins and 0.31 for DZ twins in girls, ICCs were greater for nine food groups among MZ twins than among DZ twins. Even after adjustment for covariates, the additive genetic components were still statistically significant for seven food groups. The average heritability was 36.7%, and the greatest heritability was for Western fast foods (94.6%; A: 0.95, 95% CI: 0.88–0.97; C: 0.00, 95% CI: 0.00–0.06; E: 0.05, 95% CI: 0.04–0.09), followed by Chinese fried foods (87.0%; A: 0.87, 95% CI: 0.75–0.92; C: 0.00, 95% CI: 0.00–0.05; E: 0.13, 95% CI: 0.08–0.25) and juices (82.1%; A: 0.82, 95% CI: 0.40–0.90; C: 0.03, 95% CI: 0.00–0.44; E: 0.15, 95% CI: 0.10–0.24).

Dietary patterns and their ICCs and heritabilities

Our factor analysis and five-factor solutions are presented in Table 3. The general and gender-specific patterns were similar and named according to their major contents. In girls, we excluded dairy foods from the factor analysis because of its low loading score (0.23). Factor analysis was then conducted among 22 food groups and also derived five factors.

TABLE 3 Dietary Patterns and Loading Scores of Food Items From Factor Analysis in Pooled Sample and by Sexa

aLoading scores were obtained from principle component analysis (PCA). The percentage of variance explained by each factor was presented in parenthesis. bPooled patterns were obtained from the pooled sample (both boys and girls). A total of 23 food groups were included in factor analysis. cBoy-specific patterns were obtained from boys. A total of 23 food groups were included in factor analysis. dGirl-specific patterns were obtained from girls. Milk and yogurt was dropped due to their low loading (0.23). Then, a total of 22 food groups were included in factor analysis.

Twin-pair correlation coefficients and heritability estimates for dietary patterns are shown by zygosity status and sex in Table 4. In terms of general patterns, ICCs ranged from 0.22 (snacks) to 0.65 (beverages) for MZ boys and from 0 (beverages and fried and fast foods) to 0.53 (vegetables and fruits) for DZ boys. Correlations were greater among MZ girls than among DZ girls for meats, and fried and fast foods. After adjustment for age and daily calorie intake, only heritability of fried and fast foods significantly differed between two sexes (based on 95% CI).

TABLE 4 Intraclass Correlation Coefficients and Heritability of Dietary Patterns by Sexa,b

ICCs = intraclass correlations. aShrout–Fleiss ICCs were calculated using SAS. bHeritability was estimated using ACE model: A, additive genetic effect; C, common environmental effect shared by a twin pair; E, residual environmental effect. The model was fit on each dietary phenotype using OpenMx and adjusted for age and daily calorie intake. ~ = heritability estimate was zero. cFood patterns were named the same for pooled and sex-specific analysis, but each food pattern may include different food groups (see Table 3). dFood patterns were obtained from the pooled sample (both boys and girls). eSex-specific patterns were obtained from boys and girls, respectively. *Significantly different between boys and girls (based on 95% CI).

Regarding sex-specific patterns, in boys, ICCs ranged from 0.10 (snacks) to 0.68 (beverages) for MZ twins and from 0 (fried and fast foods and beverages) to 0.46 (vegetables and fruits) for DZ twins. In girls, ICCs were greater among MZ twins than among DZ twins for meats (0.57 vs. 0.06), beverages (0.42 vs. 0.20), and fried and fast foods (0.56 vs. 0.11).

After adjusting for age and daily calorie intake, estimated heritability was 35.2% (A: 0.35, 95% CI: 0.00–0.51; C: 0.00, 95% CI: 0.00–0.50; E: 0.65, 95% CI: 0.49–0.83) for fried and fast foods, 67.9% (A: 0.68, 95% CI: 0.46–0.81; C: 0.00, 95% CI: 0.00–0.10; E: 0.32, 95% CI: 0.19–0.54) for beverages, and 32.5% (A: 0.32, 95% CI: 0.00–0.51; C: 0.00, 95% CI: 0.00–0.37; E: 0.68, 95% CI: 0.49–0.89) for meats among boys, whereas in girls, it was 53.1% (A: 0.53, 95% CI: 0.22–0.71; C: 0.00, 95% CI: 0.00–0.16; E: 0.47, 95% CI: 0.29–0.73) for meats, 58.0% (A: 0.58, 95% CI: 0.02–0.71; C: 0.00, 95% CI: 0.00–0.49; E: 0.42, 95% CI: 0.29–0.59) for beverages, and 89.8% (A: 0.90, 95% CI: 0.82–0.94; C: 0.00, 95% CI: 0.00–0.07; E: 0.10, 95% CI: 0.06–0.18) for fried and fast foods. The additive genetic components were all non-significant for vegetables and snacks in both sexes. In addition, the variance components models showed common environmental components were statistically significant for vegetable dietary intakes and snacks in girls (p < .01) but not for any other food pattern in boys (p > .05), suggesting that common environmental effects were less important among boys.

Estimated heritability of children's dietary intakes by parental characteristics

Figure 1a and b shows heritability estimates of consumption of food groups and dietary patterns by maternal education and family income level, respectively. We did not present food groups or dietary patterns where the estimated heritability was zero in these two categories; that is, higher education versus lower education, richer versus poorer. Maternal education was positively associated with heritability for refined grain, meats, eggs, nuts, snacks, and Western fast foods, but negatively associated with that for fruits, soft drinks, and Chinese fast foods.

FIGURE 1 Estimated heritabilities of dietary intakes (food groups and dietary patterns) in Chinese child twins, by maternal education and family income. Note: Heritability was estimated using variance components models. The model was fit on each food group or dietary pattern and adjusted for age, sex and daily energy intake using OpenMx. Higher education: senior high school or above; lower education: junior high school, elementary school or below; higher income: income ≥50,000 Yuan (≈7,457 USD) per year; lower income: income <50,000 Yuan per year.

The twins from higher-income families had greater estimated heritabilities for most food groups (except for refined grains and Western fast foods). Consumption of meats, fried, and fast foods appeared to be more heritable, but less so for vegetables among children from richer families and among children with more educated mothers.

Discussion

To our knowledge, this study is the first to investigate the estimated genetic influence on usual dietary intakes among children in a developing country, China, and how the influence may vary by family characteristics. We found that estimated genetic effects were generally moderate, but varied by nutrients and food groups, and by maternal education and annual family income levels. ICCs were greater for 10 of 13 nutrient measures and 8 of 13 food groups in MZ boys than in DZ boys. In girls, MZ ICCs were greater for 6 of 13 nutrients and 9 food groups than DZ ICCs. Heritability of sex-specific dietary patterns ranged from 0% to 67.9% in boys and from 0% to 89.8% in girls. The apparent genetic component of intake of fried and fast foods was significantly greater among girls compared to boys. Overall, the results support our hypothesis that children in high-income families might have a stronger heritability of food consumption than those from lower-income families, but do not support our hypothesis for children whose mothers had more education.

Our study supports a genetic influence on usual dietary intakes in both boys and girls. We found some interesting sex differences in the estimated heritabilities. Sex-specific dietary patterns had greater point estimates of heritability than did the pooled patterns. Among boys, estimated heritability was higher for beverage consumption (67.9% vs. 58.0%), lower for meat consumption (32.5% vs. 53.1%), and lower for fried and fast foods (35.2% vs. 89.8%) than those of girls, whereas only heritability of ‘fried and fast foods pattern’ significantly differed between boys and girls. Girls had higher heritability of high fat intake than boys. This gender difference in estimated heritability of high fat food consumption may be related to human evolution, since energy and fat accumulation favors maturation and pregnancy (Dunger et al., Reference Dunger, Ahmed and Ong2006; Paul et al., Reference Paul, Muller and Whitehead1979; Stoll, Reference Stoll1998). It may be also possible that social factors are prompting families with boys to feed them higher fat foods (Chunming, Reference Chunming2000) so that social factors override heritability for boys. Similar gender differences have been observed in previous research in children and adults (Faith et al., Reference Faith, Rhea, Corley and Hewitt2008).

In our study, although stronger genetic effects on nutrient intakes were found among boys than girls, who may be more susceptible to environmental factors, including influences from peers (Field et al., Reference Field, Camargo, Taylor, Berkey and Colditz1999; Field et al., Reference Field, Camargo, Taylor, Berkey, Roberts and Colditz2001) and their parents or other care providers (Levine et al., Reference Levine, Smolak, Moodey, Shuman and Hessen1994; Smolak et al., Reference Smolak, Levine and Schermer1999), girls had higher heritability in more food groups. Such a discrepancy between nutrients and food groups could be due to the strong sex differences in usual dietary intakes among children. Note that although we named the sex-specific patterns using the same terms, each pattern included different food groups in boys and girls, and the boys’ patterns were more similar to the pooled ones. For example, for boys, ‘vegetables’ included vegetables, tofu, and fruits, and ‘meats’ included meats, grains, and seafood; for girls, ‘vegetables’ solely included different types of vegetables, and ‘meats’ included meats, seafood, and tofu. Another reason may be gender difference in food sources of nutrients. In considering Vit A as an example, meats, eggs, and vegetables are Vit-A-rich foods. In our study, boys consumed significantly more meats and eggs, but less other vegetables than girls did. In addition, our results showed girls gave a higher estimated heritability for consumption of fried and fast food, but relatively weak estimates for total energy, fat and protein intake. This observation is not totally surprising since fried and fast foods are not the main food sources for total energy, fat and protein intakes among girls, which can be reflected by their daily consumption (4.4±14.5 g/day for Chinese fried foods, and 8.1±26.4 g/day for Western fast foods, Table 1). On the other hand, we found heritability of high fat intake (the top quartile of %E from fat) was greater among girls than boys.

Our SEMs suggested that shared environmental factors significantly influenced two dietary patterns among girls, but not among boys in our study. This result confirmed previous findings (Castro et al., Reference Castro, Saxton, Haghighat, Reisler, Plant and Soudant1993; Faith et al., Reference Faith, Rhea, Corley and Hewitt2008). No evidence of genetic influence was detected for consumption of several distinct food groups (including some vegetables, fish and other seafood) in our study. Previous studies have shown strong environmental influences on vegetable and fish consumptions. For example, at the family level, vegetable intakes were positively associated with education, household income, and availability of healthful foods in the home (Dehghan et al., Reference Dehghan, Akhtar-Danesh and Merchant2011; Ding et al., Reference Ding, Sallis, Norman, Saelens, Harris, Kerr and Glanz2012). At the community level, residential food retailers may also improve vegetable consumptions (Giskes et al., Reference Giskes, van Lenthe, Kamphuis, Huisman, Brug and Mackenbach2009; Rasmussen et al., Reference Rasmussen, Krolner, Klepp, Lytle, Brug, Bere and Due2006). Lower family food budget, income, and education levels may limit fish consumption (Jahns et al., Reference Jahns, Raatz, Johnson, Kranz, Silverstein and Picklo2014). An adult twin study reported similar findings as ours and showed strong shared environmental influence but not additive genetic influence on vegetable and fish intakes (Hasselbalch et al., Reference Hasselbalch, Heitmann, Kyvik and Sorensen2008). It is noted that our findings could reflect true differences between these food groups and others (Faith et al., Reference Faith, Rhea, Corley and Hewitt2008), or it may be also due to the limitations of our FFQ assessment or the modest sample size.

Our data demonstrated that family SES factors modified the genetic influence on consumption of fried and fast foods, and beverages. Both higher maternal education and income magnified genetic effects on fried and fast foods, but lowered those on beverage consumption. The magnitude of changes in the heritable component of children's foods and dietary patterns across categories of maternal education and family income has not been examined in previous research. Maternal education was associated with food-related parenting practices (Saxton et al., Reference Saxton, Carnell, van Jaarsveld and Wardle2009) and family income might affect household food environments (Larson et al., Reference Larson, Story and Nelson2009). These family SES factors could potentially influence environmental variation in children's dietary intakes or diet-related genotypic expression, and further inflate the differences in estimated heritability. Our results supported our hypothesis about family income, but not regarding maternal education. It is possible that better-educated mothers do not necessarily have better nutritional knowledge, feeding, or parenting practices. For example, a study in urban areas of Beijing, China found that higher maternal education level was positively associated with inappropriate feeding practices (OR = 2.44, 95% CI: 1.42–4.19, p < .05; Li, Li et al., Reference Li, Li, Ali and Ushijima2003). In addition, our sample size may not be large enough to detect an education effect.

This study has several main strengths. First, the twin study design allows separating influences of the genetic and environmental factors, and estimating their contributions to children's dietary intakes for different food groups (Neale et al., Reference Neale and Cardon1992). Second, compared with previous research (Breen et al., Reference Breen, Plomin and Wardle2006; Faith et al., Reference Faith, Rhea, Corley and Hewitt2008), here, usual food intake was measured using a validated FFQ. The FFQ collected answers on the amount (weight or volume) and type of consumed foods. Third, we conducted comprehensive and vigorous analysis of dietary intakes—for example, nutrients, food groups, and dietary patterns. Heritability estimates were examined and stratified by sex and SES groups. Fourth, our study filled a major literature gap as no previous research has been done in developing countries to investigate heritability of dietary intakes.

Our study also has some limitations. First, although our FFQ was validated before use and showed good reliability with moderate validity (Wang, Reference Wang2009), both boys and girls still tended to over-report their dietary intake. We tried to minimize the impact of over reporting by using nutrient density and energy-adjusted food groups and dietary patterns in our later analysis. Second, our sample may not have adequate statistical power to detect small genetic effects on dietary intake, although previous studies have used similar or smaller sample sizes (Breen et al., Reference Breen, Plomin and Wardle2006; Faith et al., Reference Faith, Rhea, Corley and Hewitt2008; Hasselbalch et al., Reference Hasselbalch, Heitmann, Kyvik and Sorensen2010). Third, although the food composition table used here reflects the most recent nutrient information, its food items are less than 3,000 and most of them are raw foods. The limitations of the food composition table may compromise our estimates of intakes of energy and nutrients, which may result in underestimation of heritability for nutrient intakes among our subjects. However, our main goal here was to examine the dietary patterns and rank the subjects based on their food consumptions. Thus, the weaknesses of the food composition table are not a major concern. Additionally, the food composition table and its earlier versions have been used for previous research, including some nationally representative studies (Ma et al., Reference Ma, Jin, Li, Zhai, Kok, Jacobsen and Yang2008; Popkin & Du, Reference Popkin and Du2003; Popkin et al., Reference Popkin, Lu and Zhai2002; Wang et al., Reference Wang, Bentley, Zhai and Popkin2002).

In conclusion, genetic factors affect usual dietary intakes in Chinese children, but the magnitude of the effects differed considerably by gender, nutrients, food groups, and dietary patterns. Genetic influence on dietary intakes can also be modified by family SES factors. This study sheds new light on genetic and environmental effects on children's dietary intakes in developing countries. Further longitudinal and genetic studies will be needed to fully understand the importance of genetic and environmental factors in children's eating behaviors, which is beneficial for early prevention of diet-related health risks.

Acknowledgments

We are grateful for the several agencies that funded this study, and thank all the participating mothers and children. We acknowledge the support from the physicians and nurses from the Jiaxing Maternity and Child Health Care Hospital in China. We thank Professor Zhu Li, Dr. Yexuan Tao, and Dr. Yinghui Liu, who have made key contributions to the study design and data collection. We thank Dr. Brion Maher for his help on data analysis. We also thank Dr. Tao Huang for quality control, review and comments on the paper. The study is part of Dr. Ji Li's PhD dissertation research under Professor Youfa Wang's guidance at Johns Hopkins University. The twin study was supported in part by multiple research grants including Grant P60 DK0079637 from National Institutes of Health (NIH)/National Institute of Diabetes and Digestive and Kidney Diseases, the Faculty Innovation Fund, Procter & Gamble Fellowship from the Johns Hopkins Bloomberg School of Public Health, the Nestle Foundation, and by Grant U54HD070725 from the Eunice Kennedy Shriver National Institute of Child Health & Human Development (NICHD). The U54 project is co-funded by the NICHD and the Office of Behavioral and Social Sciences Research (OBSSR) at NIH. The content is solely the responsibility of the authors and does not necessarily represent the official views of the funders.

Supplementary Material

To view supplementary material for this article, please visit http://dx.doi.org/10.1017/thg.2016.61.

APPENDIX Food Included in Each of the 23 Food Groups That We Created for Our Dietary Pattern Analysis

References

Beydoun, M. A., & Wang, Y. (2009). Parent-child dietary intake resemblance in the United States: Evidence from a large representative survey. Social Science & Medicine, 68, 21372144.Google Scholar
Breen, F. M., Plomin, R., & Wardle, J. (2006). Heritability of food preferences in young children. Physiology & Behavior, 88, 443447.Google Scholar
Cameron, M. A., Paton, L. M., Nowson, C. A., Margerison, C., Frame, M., & Wark, J. D. (2004). The effect of calcium supplementation on bone density in premenarcheal females: A co-twin approach. Journal of Clinical Endocrinology & Metabolism, 89, 49164922.CrossRefGoogle Scholar
Castro, D. J., Saxton, R. E., Haghighat, S., Reisler, E., Plant, D., & Soudant, J. (1993). The synergistic effects of rhodamine-123 and merocyanine-540 laser dyes on human tumor cell lines: A new approach to laser phototherapy. Otolaryngology – Head and Neck Surgery, 108, 233242.Google Scholar
Chen, W. J., Chang, H. W., Wu, M.-Z., Lin, C. C., Chang, C., Chiu, Y.-N., Soong, W.-T. (1999). Diagnosis of zygosity by questionnaire and polymarker polymerase chain reaction in young twins. Behavior Genetics, 29, 115123.Google Scholar
Chunming, C. (2000). Fat intake and nutritional status of children in China. American Journal of Clinical Nutrition, 72, 1368S–1372S.Google Scholar
Darnton-Hill, I., & Coyne, E. T. (1998). Feast and famine: Socioeconomic disparities in global nutrition and health. Public Health Nutrition, 1, 2331.Google Scholar
Dehghan, M., Akhtar-Danesh, N., & Merchant, A. T. (2011). Factors associated with fruit and vegetable consumption among adults. Journal of Human Nutrition and Dietetics, 24, 128134.CrossRefGoogle ScholarPubMed
Ding, D., Sallis, J. F., Norman, G. J., Saelens, B. E., Harris, S. K., Kerr, J., . . . Glanz, K. (2012). Community food environment, home food environment, and fruit and vegetable intake of children and adolescents. Journal of Nutrition Education and Behavior, 44, 634638.Google Scholar
Dunger, D. B., Ahmed, M. L., & Ong, K. K. (2006). Early and late weight gain and the timing of puberty. Molecular and Cellular Endocrinology, 254–255, 140145.Google Scholar
Faith, M. S., Rhea, S. A., Corley, R. P., & Hewitt, J. K. (2008). Genetic and shared environmental influences on children's 24-h food and beverage intake: Sex differences at age 7 y. American Journal of Clinical Nutrition, 87, 903911.Google Scholar
Falconer, D., & Mackay, T. (1996). Introduction to quantitative genetics (4th ed.). Harlow, UK: Longmans Green.Google Scholar
Field, A. E., Camargo, C. A. Jr., Taylor, C. B., Berkey, C. S., & Colditz, G. A. (1999). Relation of peer and media influences to the development of purging behaviors among preadolescent and adolescent girls. Archives of Pediatrics and Adolescent Medicine, 153, 11841189.Google Scholar
Field, A. E., Camargo, C. A. Jr., Taylor, C. B., Berkey, C. S., Roberts, S. B., & Colditz, G. A. (2001). Peer, parent, and media influences on the development of weight concerns and frequent dieting among preadolescent and adolescent girls and boys. Pediatrics, 107, 5460.Google Scholar
Formica, J. V., & Regelson, W. (1995). Review of the biology of Quercetin and related bioflavonoids. Food and Chemical Toxicology, 33, 10611080.Google Scholar
Giskes, K., van Lenthe, F. J., Kamphuis, C. B., Huisman, M., Brug, J., & Mackenbach, J. P. (2009). Household and food shopping environments: Do they play a role in socioeconomic inequalities in fruit and vegetable consumption? A multilevel study among Dutch adults. Journal of Epidemiology and Community Health, 63, 113120.Google Scholar
Hansell, N. K., Halford, G. S., Andrews, G., Shum, D. H., Harris, S. E., Davies, G., . . . Wright, M. J. (2015). Genetic basis of a cognitive complexity metric. PLoS One, 10, e0123886.Google Scholar
Hasselbalch, A. L., Heitmann, B. L., Kyvik, K. O., & Sorensen, T. I. (2008). Studies of twins indicate that genetics influence dietary intake. Journal of Nutrition, 138, 24062412.Google Scholar
Hasselbalch, A. L., Heitmann, B. L., Kyvik, K. O., & Sorensen, T. I. (2010). Associations between dietary intake and body fat independent of genetic and familial environmental background. International Journal of Obesity, 34, 892898.Google Scholar
Heitmann, B. L., Harris, J. R., Lissner, L., & Pedersen, N. L. (1999). Genetic effects on weight change and food intake in Swedish adult twins. American Journal of Clinical Nutrition, 69, 597602.Google Scholar
Jahns, L., Raatz, S. K., Johnson, L. K., Kranz, S., Silverstein, J. T., & Picklo, M. J. (2014). Intake of seafood in the US varies by age, income, and education level but not by race-ethnicity. Nutrients, 6, 60606075.CrossRefGoogle Scholar
Kar, B. R., Rao, S. L., & Chandramouli, B. A. (2008). Cognitive development in children with chronic protein energy malnutrition. Behavioral and Brain Functions, 4, 31.Google Scholar
Kemm, J. R. (1987). Eating patterns in childhood and adult health. Nutrition and Health, 4, 205215.CrossRefGoogle ScholarPubMed
Keskitalo, K., Silventoinen, K., Tuorila, H., Perola, M., Pietilainen, K. H., Rissanen, A., & Kaprio, J. (2008). Genetic and environmental contributions to food use patterns of young adult twins. Physiology & Behavior, 93, 235242.Google Scholar
Khan, Y., & Bhutta, Z. A. (2010). Nutritional deficiencies in the developing world: Current status and opportunities for intervention. Pediatric Clinics of North America, 57, 14091441.Google Scholar
Kitzman-Ulrich, H., Wilson, D. K., St George, S. M., Lawman, H., Segal, M., & Fairchild, A. (2010). The integration of a family systems approach for understanding youth obesity, physical activity, and dietary programs. Clinical Child and Family Psychology Review, 13, 231253.CrossRefGoogle ScholarPubMed
Larson, N. I., Story, M. T., & Nelson, M. C. (2009). Neighborhood environments: Disparities in access to healthy foods in the U.S. American Journal of Preventive Medicine, 36, 7481.CrossRefGoogle ScholarPubMed
Levine, M. P., Smolak, L., Moodey, A. F., Shuman, M. D., & Hessen, L. D. (1994). Normative developmental challenges and dieting and eating disturbances in middle school girls. International Journal of Eating Disorders, 15, 1120.Google Scholar
Li, J., & Wang, Y. (2008). Tracking of dietary intake patterns is associated with baseline characteristics of urban low-income African-American adolescents. Journal of Nutrition, 138, 94100.Google Scholar
Li, L., Li, S., Ali, M., & Ushijima, H. (2003). Feeding practice of infants and their correlates in urban areas of Beijing, China. Pediatrics International, 45, 400406.Google Scholar
Li, S., Moore, C. A., Li, Z., Berry, R. J., Gindler, J., Hong, S. X., . . . Erickson, J. D. (2003). A population-based birth defects surveillance system in the People's Republic of China. Paediatric and Perinatal Epidemiology, 17, 287293.Google Scholar
Ma, G., Jin, Y., Li, Y., Zhai, F., Kok, F. J., Jacobsen, E., & Yang, X. (2008). Iron and zinc deficiencies in China: What is a feasible and cost-effective strategy? Public Health Nutrition, 11, 632638.Google Scholar
Moore, L. L., Singer, M. R., Bradlee, M. L., Djousse, L., Proctor, M. H., Cupples, L. A. Ellison, R. C. (2005). Intake of fruits, vegetables, and dairy products in early childhood and subsequent blood pressure change. Epidemiology, 16, 411.Google Scholar
Neale, M. C., Cardon, L. R., & North Atlantic Treaty Organization, Scientific Affairs Division. (1992). Methodology for Genetic Studies of Twins and Families. Dordrecht: Kluwer Academic Publishers.Google Scholar
Newby, P. K., Weismayer, C., Akesson, A., Tucker, K. L., & Wolk, A. (2006). Long-term stability of food patterns identified by use of factor analysis among Swedish women. Journal of Nutrition, 136, 626633.Google Scholar
Ooki, S., & Asaka, A. (2004). Zygosity diagnosis in young twins by questionnaire for twins’ mothers and twins’ self-reports. Twin Research, 7, 512.CrossRefGoogle ScholarPubMed
Paul, A. A., Muller, E. M., & Whitehead, R. G. (1979). The quantitative effects of maternal dietary energy intake on pregnancy and lactation in rural Gambian women. Transactions of the Royal Society of Tropical Medicine and Hygiene, 73, 686692.Google Scholar
Popkin, B. M., & Du, S. (2003). Dynamics of the nutrition transition toward the animal foods sector in China and its implications: A worried perspective. Journal of Nutrition, 133, 3898S–3906S.CrossRefGoogle Scholar
Popkin, B. M., Lu, B., & Zhai, F. (2002). Understanding the nutrition transition: Measuring rapid dietary changes in transitional countries. Public Health Nutrition, 5, 947953.Google Scholar
Rasmussen, M., Krolner, R., Klepp, K. I., Lytle, L., Brug, J., Bere, E., & Due, P. (2006). Determinants of fruit and vegetable consumption among children and adolescents: A review of the literature. Part I: Quantitative studies. International Journal of Behavioral Nutrition and Physical Activity, 3, 22.Google Scholar
Rietveld, M. J., van Der Valk, J. C., Bongers, I. L., Stroet, T. M., Slagboom, P. E., & Boomsma, D. I. (2000). Zygosity diagnosis in young twins by parental report. Twin Research, 3, 134141.Google Scholar
Saxton, J., Carnell, S., van Jaarsveld, C. H., & Wardle, J. (2009). Maternal education is associated with feeding style. Journal of the American Dietetic Association, 109, 894898.Google Scholar
Silventoinen, K., Hasselbalch, A. L., Lallukka, T., Bogl, L., Pietilainen, K. H., Heitmann, B. L., . . . Kaprio, J. (2009). Modification effects of physical activity and protein intake on heritability of body size and composition. American Journal of Clinical Nutrition, 90, 10961103.Google Scholar
Smolak, L., Levine, M. P., & Schermer, F. (1999). Parental input and weight concerns among elementary school children. International Journal of Eating Disorders, 25, 263271.Google Scholar
Stoll, B. A. (1998). Western diet, early puberty, and breast cancer risk. Breast Cancer Research and Treatment, 49, 187193.Google Scholar
Teucher, B., Skinner, J., Skidmore, P. M., Cassidy, A., Fairweather-Tait, S. J., Hooper, L., . . . MacGregor, A. J. (2007). Dietary patterns and heritability of food choice in a UK female twin cohort. Twin Research and Human Genetics, 10, 734748.Google Scholar
Wang, W. (2009). Reproducibility and validation of a food frequency questionnaire among children and adolescents in Beijing. Beijing, China: Capital Institute of Pediatrics.Google Scholar
Wang, Y., Bentley, M. E., Zhai, F., & Popkin, B. M. (2002). Tracking of dietary intake patterns of Chinese from childhood to adolescence over a six-year follow-up period. Journal of Nutrition, 132, 430438.Google Scholar
Wang, Y., Li, J., & Caballero, B. (2009). Resemblance in dietary intakes between urban low-income African-American adolescents and their mothers: The healthy eating and active lifestyles from school to home for kids study. Journal of the American Dietetic Association, 109, 5263.Google Scholar
Yang, Y. X. (2004). China Food Composition. Beijing, China: Peking University Medical Publisher.Google Scholar
Yang, Y. X., Wang, G. Y., Pan, X. C., & Chinese Center for Disease Control and Prevention. (2002). China Food Composition. Beijing, China: Peking University Medical Publisher.Google Scholar
Zive, M. M., Berry, C. C., Sallis, J. F., Frank, G. C., & Nader, P. R. (2002). Tracking dietary intake in white and Mexican-American children from age 4 to 12 years. Journal of the American Dietetic Association, 102, 683689.Google Scholar
Figure 0

TABLE 1 Average Daily Dietary Intakes by Sex in Twin Pairs of Chinese Children Aged 7–15 Years (n = 566)a, b

Figure 1

TABLE 2 Intraclass Correlation Coefficients (ICCs) and Heritability Estimates of Daily Nutrient and Food Group Intakes in Chinese MZ and DZ Twinsa,b

Figure 2

TABLE 3 Dietary Patterns and Loading Scores of Food Items From Factor Analysis in Pooled Sample and by Sexa

Figure 3

TABLE 4 Intraclass Correlation Coefficients and Heritability of Dietary Patterns by Sexa,b

Figure 4

FIGURE 1 Estimated heritabilities of dietary intakes (food groups and dietary patterns) in Chinese child twins, by maternal education and family income. Note: Heritability was estimated using variance components models. The model was fit on each food group or dietary pattern and adjusted for age, sex and daily energy intake using OpenMx. Higher education: senior high school or above; lower education: junior high school, elementary school or below; higher income: income ≥50,000 Yuan (≈7,457 USD) per year; lower income: income <50,000 Yuan per year.

Figure 5

APPENDIX Food Included in Each of the 23 Food Groups That We Created for Our Dietary Pattern Analysis

Supplementary material: File

Li supplementary material

Li supplementary material

Download Li supplementary material(File)
File 18.8 KB