点击显示 收起
1 From the Division of Gastroenterology and Nutrition (AMT) and Biostatistics (AFJ and NOS), The Childrens Hospital of Philadelphia, and the Tulane Center for Cardiovascular Health, New Orleans (AE, SRS, and GSB).
2 Supported by grant HL38844 from the National Heart, Lung, and Blood Institute of the US Public Health Service and grant HD-32194 from the National Institute of Child Health and Development. 3 Address reprint requests to AM Tershakovec, Division of Gastroenterology and Nutrition, The Childrens Hospital of Philadelphia, 324 South 34th Street, Philadelphia, PA 19104-4399.
ABSTRACT
Background: Obesity is associated with cardiovascular disease (CVD) risk factors. Cross-sectional data suggest that hypercholesterolemia is associated with the development of childhood obesity.
Objective: The objective was to assess age-related changes in relative weight and the association between relative weight and CVD risk factors in hypercholesterolemic and nonhypercholesterolemic children who were nonobese at baseline.
Design: Data on relative weight and CVD risk factors were extracted from the Bogalusa Heart Study for nonobese 56-y-old black and white hypercholesterolemic (LDL cholesterol > 75th percentile; n = 58) and nonhypercholesterolemic (LDL cholesterol < 60th percentile; n = 215) children (41% black, 52% girls) who were also assessed 3 and 6 y later. Changes in body mass index (BMI) and CVD risk factors were assessed as a function of age, sex, race, and cholesterol concentration.
Results: BMI increased more in the hypercholesterolemic (n = 31) than in the nonhypercholesterolemic (n = 111) girls during the 6 y of follow-up but was not significantly different between hypercholesterolemic (n = 27) and nonhypercholesterolemic (n = 104) boys aged 512 y. Associations between BMI and the risk factors blood pressure, insulin, and blood lipids were observed to be stronger with increasing age and, in some cases, stronger in hypercholesterolemic children and girls.
Conclusions: Hypercholesterolemia is associated with increased relative weight in girls. The increased relative weight, even at an early age, is associated with a deleterious effect on blood lipids and other CVD risk factors in hypercholesterolemic children, although the strength of these associations is sex dependent.
Key Words: Hypercholesterolemia obesity children Bogalusa Heart Study girls
INTRODUCTION
Obesity is strongly associated with other cardiovascular disease (CVD) risk factors (18). To explain this association, it has been postulated that obesity and higher visceral abdominal fat set off a cascade, which induces or exacerbates hyperlipidemia, hypertension, and insulin resistance (4,5,914). Furthermore, the associations among adiposity and other CVD risk factors in children seem to be age-related (eg, the associations between these risk factors are stronger in older children and adults than in younger children) (1,15). However, most of the studies of these associations have evaluated older children, adolescents, and adults, limiting the understanding of these relations in young children. In addition, most previous studies evaluating the association between adiposity and CVD risk factors have studied obese children or children from the general population (1,4,1528).
An association between adiposity and other CVD risk factors in 410-y-old hypercholesterolemic and nonhypercholesterolemic children was described previously (29). This cross-sectional analysis showed an age-related increase in adiposity in hypercholesterolemic children, which was not observed in nonhypercholesterolemic children. However, the cross-sectional nature of these data limits the conclusions. To further examine the potential association between hypercholesterolemia and the development of obesity in children, we evaluated longitudinal data relating to the relative weight of 56-y-old hypercholesterolemic and nonhypercholesterolemic children from the Bogalusa Heart Study. The associations between relative weight and other CVD risk factors were also evaluated as the children grew older.
SUBJECTS AND METHODS
Subjects
The Bogalusa Heart study is a cross-sectional and longitudinal epidemiologic study of the early natural history of arteriosclerosis, coronary atherosclerosis, and essential hypertension. Children and young adults in the semirural, biracial (black and white) community of Bogalusa, LA, were surveyed between 1973 and 1991. The surveys included school-aged children and young adults (posthigh school). In the original study, 3179 children aged 514 y (49% girls, 37% black) completed an initial lipoprotein assessment (30). Data were extracted for black and white nonobese (ponderal index < 85th percentile at baseline) children who were 56-y-old at the first evaluation and who had been evaluated at the next 2 evaluation points 3 and 6 y later (at ages 89 and 1112 y, respectively). Children with age-, race-, and sex-specific elevations in LDL-cholesterol concentrations (> 75th percentile) at all 3 evaluations were classified as hypercholesterolemic. Children with LDL-cholesterol concentrations persistently less than the 60th percentile were considered nonhypercholesterolemic. The extracted data included age, height, weight, blood pressure, lipid profile, and insulin and glucose concentrations. Heights, weights, blood lipid values, and blood pressure measurements were available for all children in the analysis, whereas information relating to blood glucose and insulin concentrations was available for a subsample of the group.
We obtained the consent of the parents or guardians of all children participating in the Bogalusa Heart Study. The Bogalusa Heart Study protocols were approved by the Louisiana State University Institutional Review Board.
General and anthropometric assessment
All examinations followed the same protocols described previously (30). Trained personnel collected information and samples. Subjects were instructed to fast for 12 h before the screening, and compliance was determined by interview on the morning of the examination. Blood was drawn by antecubital venipuncture to obtain serum and plasma. Height and weight were measured twice to ± 0.1 cm and ± 0.1 kg, respectively. Blood pressure was measured in 6 replicates by 2 randomly assigned observers on the right arm of subjects while they were in a relaxed sitting position. Systolic blood pressure was recorded at the first Korotkoff phase, and diastolic blood pressure was measured at the fourth and fifth phases. The fourth phase was used in the analyses.
Laboratory analyses
From 1973 to 1986, serum total cholesterol and triacylglycerol concentrations were measured with the use of chemical procedures on a Technicon AutoAnalyzer II (Technicon Instrument Corp, Tarrytown, NY) according to the protocol developed by the Lipid Research Clinics Program (31). After 1986, these variables were determined with the use of enzymatic procedures (32,33) on an Abbott VP instrument (Abbott Laboratories, North Chicago). Serum concentrations of VLDL, LDL, and HDL cholesterol were analyzed by a combination of heparin-calcium precipitation and agar-agarose gel electrophoresis procedures (34). Both chemical and enzymatic procedures met the performance requirements of the Lipid Standardization Program sponsored by the Centers for Disease Control and Prevention (CDC), Atlanta. Starting in 1981, CDC-assigned quality-control samples were used to examine the bias in the laboratory analysis over time. For example, the average bias in concentrations of total cholesterol on CDC control samples ranged from -0.002 to -0.04 mmol/L (-0.1 to -1.6 mg/dL) between different cross-sectional surveys, with no consistent pattern over time within or between surveys.
Plasma immunoreactive insulin concentrations were measured with a commercial radioimmunoassay kit (Phadebas; Pharmacia Diagnostics, Piscataway, NJ). From 1981 to 1986, plasma glucose concentrations were measured with a glucose analyzer (Beckman Instrument Corp, Fullerton, CA) according to a glucose oxidase (EC 1.1.3.4) method. From 1987 to 1991, plasma glucose concentrations were determined as a part of a multiple chemistry profile.
Statistical analysis
Baseline characteristics of the children were compared by using one-way analysis of variance with sex and cholesterol status and their interaction as factors in the model. To assess longitudinally the change in relative weight in the 2 groups, repeated-measures analysis of variance was used with BMI as the dependent variable and age, sex, race, and hypercholesterolemia status (hypercholesterolemia compared with nonhypercholesterolemia) as the independent variables. Age at 3 levels (56, 89, and 1112 y) was the within-subjects factor. Hypercholesterolemia status, sex, and race were the between-group factors (each with 2 levels).
The associations between relative weight (BMI) and blood pressure, glucose, insulin, and lipid concentrations for hypercholesterolemic and nonhypercholesterolemic children were examined cross-sectionally for the 3 age groups with the use of Spearmans correlations. These associations were further assessed longitudinally with the use of SAS PROC MIXED, with BMI, cholesterol status, time (first measurement, 3 y, and 6 y follow-up), and the resultant higher interaction terms as factors. Statistical significance was defined as a P value 0.05. Analyses were completed by using SAS (35) and SPSS (36) software.
RESULTS
Nonhypercholesterolemic (n = 215) and hypercholesterolemic (n = 58) children were identified. Forty-one percent of the children were black and 52% were girls. Some characteristics of the children at baseline (56 y of age) are shown in Table 1. Triacylglycerol concentrations were significantly different between the sexes, and differences in height and in LDL-cholesterol and triacylglycerol concentrations were noted by cholesterol-status group. Hypercholesterolemic boys and girls had higher LDL-cholesterol concentrations than did nonhypercholesterolemic boys and girls, respectively.
View this table:
TABLE 1 . Baseline characteristics of hypercholesterolemic and nonhypercholesterolemic boys and girls
To assess the longitudinal changes in relative weight, a repeated-measures analysis of variance was completed, with BMI as the dependent variable. Significant effects were found for the following factors: age (P < 0.0001), the age x sex interaction term (P < 0.0001), the age x hypercholesterolemic status interaction term (P = 0.04), and the age x sex x hypercholesterolemic status interaction term (P = 0.01). As shown in Figure 1, the hypercholesterolemic girls had a greater increase in BMI from 5 to 12 y of age than did the nonhypercholesterolemic girls. At 1112 y of age, significantly more hypercholesterolemic girls (45.2%) than nonhypercholesterolemic girls (21.6%) were overweight (BMI > 85th percentile, P = 0.01; 37). Over this age range, the BMI for the hypercholesterolemic and nonhypercholesterolemic boys did not differ significantly. The relation between BMI and age, sex, and hypercholesterolemia status did not change after the adjustment for baseline triacylglycerol concentration. Similarly, the increased BMI expressed in the hypercholesterolemic girls was independent of race.
FIGURE 1. . BMI as a function of age in hypercholesterolemic () and nonhypercholesterolemic () girls and hypercholesterolemic () and nonhypercholesterolemic () boys. The BMI of the hypercholesterolemic girls increased significantly with age compared with the nonhypercholesterolemic girls. No significant differences between nonhypercholesterolemic and hypercholesterolemic boys were observed
The associations between BMI and blood pressure and glucose, insulin, and lipid concentrations were initially assessed cross-sectionally with the use of Spearmans correlations at the 3 initial evaluation points (ages 56 y, 89 y, and 1112 y) for the girls (Table 2) and boys (Table 3). The longitudinal association between these factors and BMI were assessed with the use of mixed-effect-model analysis by using SAS PROC MIXED. Age and cholesterol status were also included in this model. For the girls, BMI was significantly related to systolic and diastolic blood pressure and HDL-cholesterol and triacylglycerol concentrations (P < 0.0001). In addition, the relations between BMI and systolic blood pressure (P < 0.0001), diastolic blood pressure (P = 0.008), and triacylglycerol concentration (P < 0.0001) increased with age in the girls. The relations between BMI and systolic blood pressure (P = 0.02) and between BMI and triacylglycerol concentration (P = 0.01) were also stronger with age in the hypercholesterolemic than in the nonhypercholesterolemic girls (P < 0.0001). In the boys, BMI was associated with insulin concentration (P = 0.01) and systolic blood pressure (P = 0.0004). In addition, the relation between BMI and systolic blood pressure (P = 0.01) and diastolic blood pressure (P = 0.0003) changed with age in the boys; the relation between BMI and diastolic blood pressure also differed with age between the hypercholesterolemic and nonhypercholesterolemic boys (P = 0.03).
View this table:
TABLE 2 . Spearmans correlation coefficients (r) between BMI and other cardiovascular disease risk factors for girls by age group1
View this table:
TABLE 3 . Spearmans correlation coefficients (r) between BMI and other cardiovascular disease risk factors for boys by age group1
DISCUSSION
In this longitudinal assessment of changes in relative weight in young hypercholesterolemic children, we found an increase in BMI with age in the hypercholesterolemic girls. These findings contrast with those in the literature, which suggest that obesity precedes the development of dyslipidemia (4,5,914,16). However, these current longitudinal findings are somewhat consistent with a previous assessment of cross-sectional data relating to hypercholesterolemic and nonhypercholesterolemic children, which suggests that hypercholesterolemic boys and girls developed increased adiposity between 6 and 10 y of age (29).
This previous cross-sectional analysis also showed that systolic blood pressure and insulin concentrations were positively correlated with the increased adiposity of the hypercholesterolemic children. The current longitudinal analysis showed a correlation between BMI and other CVD risk factors, but more of these relations were significant for the girls than for the boys. This is consistent with an earlier expression of obesity and CVD risk factors in girls, as indicated by the observation that, at presentation, the mean age of adolescent girls with type 2 diabetes was lower than the mean age of adolescent boys with new onset type 2 diabetes (38). It is possible that the dramatic changes in lipids and lipoproteins occurring in boys between the ages of 9 and 15 y or the differences in sex hormones and timing of onset of puberty between boys and girls may influence the sex differences observed in our analysis (39). However, such a sex difference was not observed in our previous study (29). Similarly, Bao et al (40) reported that an elevated LDL-cholesterol concentration in childhood was associated with obesity and dyslipidemia in adulthood in both sexes. Given these different results for boys and girls, an evaluation of these relations in young boys and girls that involves follow-up past the age of 12 y into adolescence should be considered. Other studies also defined ethnic differences in development and maturation, which affect body composition (41). However, the pattern of increased BMI expressed in the hypercholesterolemic girls in the current study was independent of race.
Studies that focus on the expression of other CVD risk factors, specifically in hyperlipidemic children, are limited. An association between age and blood lipid concentrations and an independent relation between relative weight and hyperlipidemia associated with familial combined hyperlipidemia was previously described in children with familial combined hyperlipidemia (42). Although data from the Bogalusa Heart Study are not specific for children with familial hyperlipidemia, these previous observations are consistent with the current findings, again suggesting that increased weight exacerbates hyperlipidemia in hyperlipidemic children, at least in girls. Also, the observation that adults with pure hypercholesterolemia did not have altered insulin sensitivity suggests that the expression of other CVD risk factors may differ among persons with different lipid disorders (43). This should be further evaluated in children and adults.
As opposed to these studies of hypercholesterolemic children, most published studies that evaluated the clustering of CVD risk factors in children focused on children from the general population without predefined risk factors (14, 7, 8, 16, 17, 2 3,4450) or on obese children (1822,26,51). Webber et al (1) and Freedman et al (52) described a greater than expected prevalence of CVD risk factors, including increased cholesterol concentrations, blood pressure and weight-height index, in 517-y-old children but not in preschool children. Boulton and Johnston (16) confirmed this lack of clustering of risk factors in 4-y-old children. These results are consistent with the age-related associations we described in hypercholesterolemic children. The potential long-term effect of these risk factors at a young age is suggested by the fact that risk factors, such as high insulin concentrations, track from childhood into early adulthood, and higher insulin concentrations cluster with higher relative weights, higher blood pressure, and adverse changes in lipids and lipoproteins (8,16).
Despite the consistency of our findings, it is not clear how hypercholesterolemia is linked to high adiposity. It seems more likely that hypercholesterolemia acts as a marker of altered metabolism, which results in excessive adiposity. This is supported by the higher triacylglycerol concentrations in the hypercholesterolemic children at baseline. The hypercholesterolemic children described here at baseline may have been a pediatric variant of metabolically obese, normal-weight adults, ie, adults with normal body weight but who express a cluster of obesity-related characteristics (53,54). However, because these hypercholesterolemic girls go on to express an increasing BMI, a more appropriate descriptive phrase for these children may be "metabolically obese, preobese." Ultimately, the increased relative weight these girls displayed was associated with increased blood pressure and insulin concentrations. In addition, the association between BMI and lipid concentrations suggests that obesity further exacerbates dyslipidemia. This is consistent with the hypothesis that hyperinsulinemia (associated with increased adiposity) also induces or exacerbates dyslipidemia (14). Because the study of older children and adults may be complicated by secondary changes or by adaptations that occur with age or time, it is clear that future studies should evaluate the longitudinal expression of and interaction between high adiposity, dyslipidemia, and hyperinsulinemia in young children.
ACKNOWLEDGMENTS
We are grateful to the children who participated in the Bogalusa Heart Study, without whom this research could not have been conducted, and to Deirdre MacLeod for preparing the manuscript.
REFERENCES