 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
* Clinical Diabetes and Nutrition Section
Diabetes and Arthritis Epidemiology Section, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, Maryland
Pennington Biomedical Research Center, Louisiana State University, Baton Rouge, Louisiana
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
Methods. During the summer months of 1992 to 1995 and again 5 years later, 138 (65 boys and 73 girls) 5-year-old Pima Indian children were studied. Height; weight; body composition; parental obesity; and fasting plasma insulin, glucose, and leptin concentrations were determined at baseline and follow-up. Linear regression models were used to assess the effect of the baseline variables on the development of obesity.
Results. At both 5 and 10 years of age, Pima Indian children were heavier and fatter than an age- and gender-matched reference population. All anthropometric and metabolic variables tracked strongly from 5 to 10 years of age (r 
0.70). The most significant determinant of percentage of body fat at 10 years of age was percentage of body fat at 5 years of age (R2 = 0.53). The combined effect of high maternal body mass index, elevated fasting plasma leptin concentrations, and low fasting plasma insulin concentrations at baseline explained an additional 4% of the total variance in adiposity at follow-up.
Conclusions. Although parental obesity and metabolic variables such as insulinemia and leptinemia at baseline account for a small percentage of the variance in adiposity at follow-up, early childhood obesity is the dominant predictor of obesity 5 years later. These results suggest that strategies to prevent childhood obesity must be initiated at a very early age.
Key Words: childhood obesity • growth and development • parent-child relationship • overweight tracking • Pima Indians
Abbreviations: BMI, body mass index • DEXA, dual-energy x-ray absorptiometry
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
The Pima Indians of Arizona are particularly prone to the development of obesity19; as a result, this Native American population has the highest reported prevalence of type 2 diabetes in the world.18 Since 1982, the National Institutes of Health has conducted prospective studies in this population, aimed at identifying risk factors for weight gain in adulthood (reviewed in 20). Because most Pima Indians develop obesity before reaching adulthood, however, the determination of risk factors for childhood and adolescent obesity takes on great importance. In white individuals, rare monogenic defects resulting in leptin deficiency or disruptions in the melanocortin pathway have been shown to result in severe early onset obesity in a handful of children.21 However, in the many longitudinal studies of childhood obesity that have been published, parental obesity is the only consistent demographic variable identified as a risk factor in large numbers of subjects.22–25
This report presents the findings of a comprehensive longitudinal investigation aimed at assessing anthropometric characteristics, parental obesity, and leptin and insulin concentrations as potential risk factors for obesity in children between 5 and 10 years of age. The aims of this study were 1) to assess body weight, body composition, and fasting plasma leptin and insulin concentrations in a large cohort of Pima Indian children at 5 and 10 years of age; 2) to examine predictors of obesity between childhood and the onset of adolescence; and 3) to determine whether these characteristics track over time.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
40 miles southeast of Phoenix. Pima Indian children were of full Indian and at least 75% Pima-Papago heritage. Children arrived at the clinic at 8:00 AM in the fasted state, accompanied by at least 1 parent, and their health status was assessed by medical history and physical examination. Because children were participating in a study of total energy expenditure,28 which required 2 clinic visits, they were seen again 1 week later. Because diabetes during pregnancy is known to affect the risk of obesity in the offspring,29 children whose mothers were known to have had diabetes before or during the pregnancy of interest were excluded from the analysis, reducing the number of children in this report to 138. Before participation, volunteers and their parents were fully informed of the nature and purpose of the study and written informed consent/assent was obtained. The experimental protocol was approved by the Institutional Review Board of the National Institute of Diabetes and Digestive and Kidney Diseases and by the Tribal Council of the Gila River Indian Community.
Anthropometry
Anthropometric measurements were performed during both clinic admissions, and results represent the means of these 2 measurements. Height was measured without shoes. Body weight was measured while the children were wearing light summer clothing. Children were stratified into BMI classes using age- and gender-matched BMI percentile data from the recently published National Center for Health Statistics growth charts.30 Accordingly, children with a BMI 95th percentile for the reference population were considered to be overweight, whereas children with a BMI <95th percentile but 85th percentile were considered to be "at risk of overweight."30 Relative weight was defined as the percentage median weight of a gender- and age-matched reference population.30 Birth weight was obtained from the medical record or reported by the parent. Parental weight and height were either measured at the time of the clinic visit or reported. However, there was no indication in the clinic chart as to the method of assessment. Because more children were accompanied on these visits by their mothers, it is likely that more maternal values were measured and more paternal values were recorded.
Body Composition
In all 5-year-old children, body composition was assessed using total body water calculated from 18O dilution spaces with the assumption that water is 75% of the fat-free mass in girls and 74% in boys.31 At follow-up, body composition was measured by 18O dilution in 53 of the 10-year-old children; in the 85 remaining children, body composition was determined using dual-energy x-ray absorptiometry (DEXA) as previously described.32 The following regression equation, developed using the percentage of body fat values of 64 10-year-old children who had both DEXA and 18O measurements, was used to convert percentage of body fat measured by DEXA to percentage fat measured by 18O when the latter measurements were lacking: %Body Fat 18O = 0.835 x %Body Fat DEXA + 9.13 (R2 = 0.95, standard error of the estimate = 2.1%, P < .0001).
Analytical Measurements
Blood samples were drawn while the child was in the fasted state. Plasma glucose concentrations were measured using the glucose oxidase method (Beckman Instruments Inc, Fullerton, CA). Plasma insulin concentrations were measured with an automated radioimmunoassay (ICN Biochemicals, Costa Mesa, CA). Plasma leptin concentrations were measured in duplicate by a solid-phase sandwich enzyme immunoassay using affinity-purified polyclonal and monoclonal antibodies (AMGEN, Inc, Thousand Oaks, CA).
Statistical Methods
All statistical analyses were performed using software of the SAS Institute (Cary, NC). Throughout the text, the data are expressed as means ± standard deviation. However, the median and 5th and 95th centiles are given for fasting plasma insulin and leptin concentrations because these variables were not normally distributed. In addition, fasting plasma insulin and leptin concentrations were log transformed (log10) to normalize the distributions before analysis.
Spearman rank correlation coefficients were used to quantify the relationships among the variables of interest and to assess the tracking of anthropometric and metabolic variables. Linear regression models were used to assess the effect of baseline variables on the development of obesity; percentage of body fat at age 10 years was used as the dependent variable and percentage of body fat at age 5 years was used as the independent variable in all models. Because the children were growing at the same time as obesity was developing and because weight has been used as an indicator of growth and/or obesity, models were also developed in which body weight at age 10 years was used as the dependent variable and body weight at age 5 years was used as the independent variable.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
|
There was a weak positive correlation between birth weight and body weight at 5 and 10 years of age (r = 0.24, P = .01; r = 0.19, P = .04, respectively); in contrast, the correlation between birth weight and percentage of body fat was not significant at 5 or 10 years of age (Table 2). At both 5 and 10 years of age, percentage of body fat and body weight were highly correlated with parental BMI (Table 2). This was true for both boys and girls separately (data not shown) as well as for both genders combined. Moreover, the finding was consistent for maternal as well as paternal BMI, for the combined effect of both parents, and whether using the BMI values of the parents obtained when the child was 5 years of age (Table 2) or 10 years of age (data not shown).
|
There was no effect of gender on unadjusted plasma insulin concentrations at baseline; at follow-up, girls had higher unadjusted insulin concentrations than boys (P = .02; Table 1). The mean fasting plasma insulin concentration adjusted for percentage of body fat at 5 and 10 years of age was not different in boys and girls. Although the unadjusted mean fasting plasma leptin concentration was higher in girls than in boys at both 5 and 10 years of age (P = .001; Table 1), there were no gender differences in adjusted leptin concentrations at either age, and there were no gender differences in fasting glucose concentrations at either age.
Longitudinal Analysis: Tracking of Growth, Adiposity, and Metabolic Variables
The correlation coefficients for the relationships between the anthropometric characteristics at baseline and follow-up were 0.75 for both percentage of body fat and body weight (Fig 1). Moreover, with the use of the recently published National Center for Health Statistics growth charts to establish BMI cutoff criteria for overweight in children and adolescents,30 BMI risk status was found to have tracked in 59% of the children from 5 to 10 years of age: 43 children who were within the normal range (<85th percentile), 2 children who were classified as "at risk of becoming overweight" (95th > BMI 85th percentile), and 36 who were already considered overweight (95th percentile) at baseline maintained these same risk status classifications at follow-up. Of the 57 children whose risk factor classification did not track, 54 became more obese whereas only 3 improved their risk factor class.
|
Prospective Analysis
The results of the prospective analyses are shown in Table 3. Baseline percentage of body fat and gender accounted for 53% of the variance in follow-up adiposity, whereas baseline body weight and gender accounted for 72% of the variance in follow-up weight. The addition of birth weight to these models had no significant effect on the variance. Maternal BMI accounted for approximately 1% to 4% of the variance in percentage of body fat and/or body weight; paternal BMI was not significant in these models, and the combined effect of parental BMI seems to be attributable primarily to maternal BMI (data not shown).
|
The fasting plasma leptin concentration at age 5 years was a positive determinant of subsequent percentage of body fat (r = 0.20, P = .01) and body weight (r = 0.20, P = .06), ie, children with higher fasting plasma leptin concentrations at baseline were both fatter and heavier at follow-up.
Multivariate Prediction Models
Results of the stepwise multiple linear regression models are shown in Table 4. The major determinants of percentage of body fat at age 10 years were percentage of body fat, maternal BMI, and fasting plasma insulin and leptin concentrations at age 5 years, explaining 57% of the variance. Baseline body weight, maternal BMI, and fasting plasma insulin concentrations at age 5 years explained 78% of the variance in body weight at 10 years of age.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
Cross-Sectional Findings
Applying the recently published BMI standards of overweight in children (BMI 95th percentile30) to this cohort, we found that by the age of 5 years, almost 1 (28%) of 3 Pima Indian children was overweight; at the age of 10 years, 1 (53%) of every 2 Pima Indian children was overweight. Pima Indian children are heavier and fatter compared not only with age- and gender-matched white children1,2,26 but also with age- and gender-matched Native American children of other tribal affiliations.5–7 A recent investigation from our laboratory found that from 1955 to 1994, the average BMI increased in Pima Indian children by 1.1 kg/m2 and in Pima Indian adolescents by 3.6 kg/m2,33 increases that are reflected in the BMI of the children in the current study.
When comparing the body composition results of this cohort with those of a reference population,31 it is apparent that the bulk of the excess weight in these Pima Indian children is attributable to excess fat mass. Although 5- and 10-year-old Pima Indian children are taller and heavier than the reference child,31 their percentage of body fat is almost twice as high. Not only is this marked prepubertal adiposity likely to have long-term adverse effects, but it also can have immediate health consequences in childhood, including orthopedic, pulmonary, and gastroenterological problems.34 Finally, and perhaps most important, childhood and adolescent obesity convey a substantially increased risk of developing type 2 diabetes,35 an effect of major concern in a population already burdened by the highest reported prevalence of this disease worldwide.18
Although many investigators have reported a significant effect of gender on leptin concentrations in early childhood, once leptin was adjusted for body composition, we did not find that to be true in this or a previous study from our laboratory that included white as well as Pima Indian children.27
Tracking of Variables
The repeated assessment of body weight and body composition at 5 and 10 years of age in the present study allowed us to evaluate how these variables tracked during this critical period of development. Guo and Chumlea36 recently reviewed the definition of tracking and suggested that it is the "prediction of future measures from earlier values, as determined by correlations between values at pairs of ages for the same individual." Canalization, in contrast, is the "tendency for all of an individual’s serial measurements to be in the same canal of the population distribution."36 In this study, the children’s anthropometric variables tracked strongly, with correlation coefficients ranging from r = 0.75 for percentage of body fat to r = 0.83 for weight. Canalization of BMI risk factor status proved to be robust as well, with 59% of the cohort remaining in the same relative BMI population percentiles from 5 to 10 years of age. However, these results also show that in an obesity-prone population such as the Pima Indians, one must be cautious in assigning categorical variables. Even those children who were categorized as within normal limits at age 5 years proved to be actually at risk of developing obesity because almost 1 of every 3 (27%) became overweight at age 10 years. Previous data suggest that the probability of having a BMI 26 kg/m2 in women or 28 kg/m2 in men at the age of 35 years is between 20% and 30% if childhood BMI is 95th percentile but 40% to 80% if adolescent BMI is 95th percentile.36 On the basis of these data, many of the children in this cohort are at risk of adult obesity.
Although (not surprising) insulin concentrations increased with increasing adiposity, the tracking of insulinemia was not simply a consequence of the tracking of adiposity because relative hyperinsulinemia (insulinemia adjusted for percentage of body fat) also tracked. This suggests that metabolic abnormalities such as insulin resistance and basal insulin hypersecretion track in and of themselves, which is important given that both predict diabetes in adulthood.37 Several authors38,39 have recently begun assessing insulin secretion and insulin sensitivity in obesity-prone pediatric populations and have found important racial differences, independent of weight and activity level, in black compared with white children, suggesting that genetic factors may play a role.
Relative leptinemia tracked from 5 to 10 years of age in boys but not in girls, possibly an indication of a pubertal shift in the girls in this study; a number of studies indicate that leptin concentrations increase during puberty.40–42 Although pubertal development was not assessed in this cohort, it is likely that a number of these children, especially the girls, were pubertal at the time of study. Ahmed et al40 reported that differences in the acquisition of adult body composition primarily account for the pubertal divergence in leptin levels between boys and girls. In the current study, both genders had similar increases in percentage of body fat from 5 to 10 years of age, a change that is expected in girls as they approach puberty, but not in boys.
Prospective Findings
In addition to documenting the tracking of growth and adiposity during this formative childhood period, prospective analysis of our data has allowed us to assess the association between baseline factors and adiposity and body weight at 10 years of age. The results of the multivariate models suggest that prospective analyses perform better with weight than with percentage of body fat. This is most likely because weight incorporates factors such as height, which we found to have a correlation coefficient of 0.85 for measurements at 5 and 10 years of age. Moreover, the minor increase in explanatory power of the multivariate models is most likely attributable to the inclusion of baseline weight or fat into the models. In so doing, the genetic and family influences that affect the variable at age 5 years are lost at age 10 years. Despite these caveats, however, we did find a predictive effect of maternal BMI as well as insulin and leptin.
Birth Weight
Once baseline body fat percentage and body weight were entered into the respective regression models, birth weight was no longer a determinant of adiposity or growth in childhood. Both low and high birth weights have been implicated as risk factors for childhood and adult obesity; however, the results are inconclusive. A recent investigation in black children who were followed from birth to young adulthood found that birth weight adjusted for gestational age was no longer associated with young adult weight once factors such as maternal prepregnancy weight were accounted for.22 Frisancho23 found that heavier newborns became heavier adolescents only when the parents were also overweight. In the Fels Longitudinal Study, birth weight was only a weak predictor of adult overweight when childhood weight and lifestyle factors were considered.43 Furthermore, in a study of monozygotic and dizygotic twin pairs, Allison et al44 found that birth weight had an enduring effect on adult height independent of weight but not on adult weight independent of height, suggesting that the intrauterine environment entrains height but not weight.
Parental Obesity
In the current study, parental BMI was significantly associated with adiposity and body weight at both 5 and 10 years of age. The relationship was stronger for maternal than for paternal BMI, although this may have been biased by the fact that much of the data in fathers was reported whereas that in the mothers was measured. Maternal BMI had a significant effect on the tracking of both growth and adiposity when it was the only covariate in the model as well as when the models were adjusted for the child’s baseline metabolic characteristics (plasma insulin and leptin concentrations).
These results are in agreement with findings by Whitaker et al,24 who reported that parental obesity significantly increased the odds of an obese child’s becoming an obese adult, especially when the child is obese before the age of 10 years. In the present study, mean parental BMI was 33 kg/m2 when the child was 5 years of age and 35 kg/m2 when the child was 10 years of age, indicating that the parents themselves became more obese during the 5-year study period. Parental perception of overweight in children has been cited as a detrimental factor in obesity prevention efforts.45 In this cohort, however, 67% of parents correctly perceived that their child was already overweight at the age of 5 years when they responded in the affirmative to that query. In addition, parents were informed of their children’s weight status and the risks thereof during the course of the physical examination and appropriate referrals were made for follow-up care. However, only in 1 child did this result in decreased risk at age 10 years. Among Pima Indians who are predisposed to obesity19 and who consume diets fairly high in fat and saturated fat,46 the effect of parental BMI may be as a result of shared genes47 or shared environmental effects48 or both; the perception that a child is overweight apparently is not sufficient to overcome these factors.
Metabolic Variables
Insulin
In contrast to a previous report from this laboratory,49 childhood hyperinsulinemia was not found to predict adiposity or body weight at 10 years of age. On the contrary, fasting plasma insulin concentrations in childhood, either adjusted or unadjusted for percentage of body fat, were actually negatively related to adiposity and body weight at age 10 years. Several factors may account for this discrepancy: In the current study, offspring of mothers with diabetes were carefully excluded and all children were 5 years of age at baseline and 10 years of age at follow-up. In contrast, the previous investigation may have contained data from offspring of mothers with diabetes, and baseline data were collected on children aged 5 to 9 years and follow-up data were collected >9 years later, carefully excluding the pubertal period. Consequently, the findings between the 2 studies may not necessarily be conflicting. However, even when we included the offspring of diabetic pregnancies in this analysis, the correlation between insulin and adiposity was still negative and consistent with those in Pima Indian adults, in whom reduced insulin secretion was found to reflect greater insulin sensitivity and was predictive of weight gain.50
Leptin
In a cross-sectional study, Caprio et al51 reported that leptin concentrations are elevated in obese compared with nonobese children, adolescents, and young adults, suggesting that hyperleptinemia is an early sign of juvenile obesity. The results of this longitudinal study confirm that finding; however, they conflict with a previous report on adults from our laboratory in which relatively low leptin concentrations predicted weight gain.52 Although these results in adults have not been confirmed by others,53 follow-up investigations in the present cohort will allow us to assess the significance of this finding.
|
CONCLUSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
We thank the members of the Gila River Indian Community for their support of this study. Most of all, we thank the children and families who participated.
|
FOOTNOTES |
|---|
Reprint requests to (A.D.S.) NIH/NIDDK/CDNS, 4212 North 16th St, Rm 541, Phoenix, AZ 85016. E-mail: arline_salbe{at}nih.gov
Dr Weyer’s current affiliation is Amylin Pharmaceuticals, San Diego, California.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  A. F. Fleisch, N. Agarwal, M. D. Roberts, J. C. Han, K. R. Theim, A. Vexler, J. Troendle, S. Z. Yanovski, and J. A. Yanovski Influence of Serum Leptin on Weight and Body Fat Growth in Children at High Risk for Adult Obesity J. Clin. Endocrinol. Metab., March 1, 2007; 92(3): 948 - 954. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. E Saelens, R. J Seeley, K. van Schaick, L. F Donnelly, and K. J O'Brien Visceral abdominal fat is correlated with whole-body fat and physical activity among 8-y-old children at risk of obesity Am. J. Clinical Nutrition, January 1, 2007; 85(1): 46 - 53. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Zephier, J. H. Himes, M. Story, and X. Zhou Increasing Prevalences of Overweight and Obesity in Northern Plains American Indian Children Arch Pediatr Adolesc Med, January 1, 2006; 160(1): 34 - 39. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. C. Bunt, P. A. Tataranni, and A. D. Salbe Intrauterine Exposure to Diabetes Is a Determinant of Hemoglobin A1c and Systolic Blood Pressure in Pima Indian Children J. Clin. Endocrinol. Metab., June 1, 2005; 90(6): 3225 - 3229. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
U. Ekelund, L. B Sardinha, S. A Anderssen, M. Harro, P. W Franks, S. Brage, A. R Cooper, L. B. Andersen, C. Riddoch, and K. Froberg Associations between objectively assessed physical activity and indicators of body fatness in 9- to 10-y-old European children: a population-based study from 4 distinct regions in Europe (the European Youth Heart Study) Am. J. Clinical Nutrition, September 1, 2004; 80(3): 584 - 590. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Hakanen, T. Ronnemaa, S. Talvia, L. Rask-Nissila, M. Koulu, J. Viikari, M. Bergendahl, and O. Simell Serum Leptin Concentration Poorly Reflects Growth and Energy and Nutrient Intake in Young Children Pediatrics, May 1, 2004; 113(5): 1273 - 1278. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. C. Bunt, A. D. Salbe, I. T. Harper, R. L. Hanson, and P. A. Tataranni Weight, Adiposity, and Physical Activity as Determinants of an Insulin Sensitivity Index in Pima Indian Children Diabetes Care, September 1, 2003; 26(9): 2524 - 2530. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. C. Bunt, A. D. Salbe, M. H. Tschop, A. DelParigi, P. Daychild, and P. A. Tataranni Cross-Sectional and Prospective Relationships of Fasting Plasma Ghrelin Concentrations with Anthropometric Measures in Pima Indian Children J. Clin. Endocrinol. Metab., August 1, 2003; 88(8): 3756 - 3761. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 OTHER ARTICLES NOTED (Nov 01 to 18 Oct 02) Evid. Based Nurs., January 1, 2003; 6(1): e1 - 1. [Full Text] [PDF]
|
|
|
|
|
|
N. Stefan, J. C. Bunt, A. D. Salbe, T. Funahashi, Y. Matsuzawa, and P. A. Tataranni Plasma Adiponectin Concentrations in Children: Relationships with Obesity and Insulinemia J. Clin. Endocrinol. Metab., October 1, 2002; 87(10): 4652 - 4656. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. D. Salbe, C. Weyer, I. Harper, R. S. Lindsay, E. Ravussin, and P. A. Tataranni Assessing Risk Factors for Obesity Between Childhood and Adolescence: II. Energy Metabolism and Physical Activity Pediatrics, August 1, 2002; 110(2): 307 - 314. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
