点击显示 收起
1 From the Division of Nutrition, Centers for Disease Control and Prevention, Atlanta (DSF and BAB); LipoMed, Inc, Raleigh, NC (JDO); the Department of Biochemistry, North Carolina State University, Raleigh (JDO); and the Tulane Center for Cardiovascular Health, School of Public Health and Tropical Medicine, Tulane University, New Orleans (SRS and GSB).
2 Supported by NIH grants HL 38844 and AG-16592. 3 Address reprint requests to DS Freedman, CDC MS-K26, 4770 Buford Highway NE, Atlanta, GA 30341-3717. E-mail: dfreedman{at}cdc.gov.
ABSTRACT
Background: Obese children and adults, particularly those with abdominal obesity, have an elevated serum triacylglycerol concentration. Furthermore, triacylglycerol concentrations are generally higher in whites than in blacks, and the relation of obesity to triacylglycerol concentrations may be stronger in whites. However, there is little information on the relation of obesity to the metabolically distinct subclasses of VLDL in children.
Objective: The objective was to examine possible differences between blacks (n = 367) and whites (n = 549) in mean concentrations of triacylglycerols, in mean concentrations of small and large VLDL, and in the relation of waist circumference to concentrations of triacylglycerol and VLDL subclasses.
Design: We measured VLDL subclass concentrations and assessed the relation of various obesity indexes to triacylglycerols in a cross-sectional study of 10- to 17-y-olds.
Results: The mean triacylglycerol concentration was 0.3 mmol/L (25 mg/dL) higher in white than in black children, primarily because of a 0.2-mmol/L (140%) difference in mean concentrations of large VLDL. In contrast, the mean concentrations of small VLDL differed by only 0.05 mmol/L (29%). In addition, the relations of waist girth to concentrations of triacylglycerol and large VLDL were 2- to 6-fold stronger among white children than among black children. Although white children had higher concentrations of large VLDL than did black children, this difference increased from 0.1 to 0.4 mmol/L across quintiles of waist circumference. Waist circumference was not significantly related to concentrations of small VLDL.
Conclusion: These contrasting associations with obesity, which differ between white and black children, suggest that information on VLDL subclasses could provide additional information on the risk of obesity-related ischemic heart disease.
Key Words: Blacks • whites • children • lipids • lipoprotein subclasses • nuclear magnetic resonance spectroscopy • triacylglycerol • VLDL • the Bogalusa Heart Study
INTRODUCTION
Obese children and adults, particularly those with a central or abdominal distribution of fat, have elevated concentrations of serum triacylglycerol—a surrogate measure of VLDL cholesterol—and low concentrations of HDL cholesterol (1, 2). The major lipoprotein classes, however, are composed of subclasses that vary in size, composition, and atherogenicity (3–5), and it is possible that these subclasses are differentially associated with obesity. For example, several obesity indexes have been found to be more strongly associated with concentrations of the larger VLDL subclasses than with the smaller VLDL subclasses (6–9). Obese persons also have relatively high concentrations of small HDL subclasses (6, 9, 10) and a smaller mean size of LDL particles (6, 8, 9, 11). The results of some studies indicate that these obesity-related differences in lipoprotein subclasses may be important in atherosclerosis (12, 13).
Mean triacylglycerol concentrations are higher in whites than in blacks (14, 15), and some evidence suggests that this contrast may arise, in part, from differences in the relation of obesity to triacylglycerol concentrations. Increases in skinfold thicknesses are associated with a 2- to 3-fold greater increase in triacylglycerol concentrations in white than in black adults (16). A similar racial difference was observed in early life, with white children (particularly boys) showing a stronger association between relative weight and triacylglycerol concentrations than black children (17). It has been suggested that the clearance of plasma triacylglycerol may be enhanced in blacks, and black men were found to have a higher lipoprotein lipase activity than were white men (18–20).
Additional information on the role of obesity in ischemic heart disease can be obtained by examining the relation of obesity to VLDL subfractions and by determining whether these relations differ between blacks and whites. Previous studies of obesity and lipoprotein subclasses focused on adults. However, because atherosclerosis is known to begin in early life (21), we examined these associations in 10- to 17-y-olds. We were particularly interested in determining whether the differences in the relation of obesity to triacylglycerol concentrations between blacks and whites could be attributed to specific VLDL subclasses.
SUBJECTS AND METHODS
Sample
Children and adolescents in the current analyses participated in the Bogalusa Heart Study, an epidemiologic study of cardiovascular disease risk factors in early life (22). The surrounding community, Ward 4 of Washington Parish (Louisiana), is fairly typical of semirural towns in the South; the 1990 population of 43000 was 33% black. Cross-sectional examinations of schoolchildren have been conducted in Bogalusa every 3–5 y since 1973. Informed consent was obtained from all participants, and the study protocols were approved by the Human Subjects Review Committee of the Tulane University School of Public Health and Tropical Medicine.
The relations of various obesity indexes to triacylglycerol concentrations were assessed in 2 samples from the Bogalusa Heart Study. Sample 1 (n = 12524 observations from 7991 children) consisted of all 10- to 17-y-olds examined (after fasting) between 1973 and 1994 (23); this sample consisted of a fairly similar number of boys and girls and was 40% black. Sample 2 (n = 918) consisted of all 10- to 17-y-olds examined in 1992–1994 in whom lipoprotein subclass determinations were performed (24); the sample represented 55% of all age-eligible children. Because there is a tendency for a portion of large VLDL particles in hypertriglyceridemic plasma to be adversely affected by freezing, 2 of the children in the second sample with a triacylglycerol concentration >4.5 mmol/L were excluded from the analyses. The maximum triacylglycerol concentration in this sample was 3.6 mmol/L and the 99th percentile was 2.8 mmol/L.
Anthropometry
While subjects were wearing a gown, underpants, and socks, weight was measured to the nearest 0.1 kg with a balance-beam scale, and height was measured to the nearest 0.1 cm with a manual height board. Both the body mass index (BMI; in kg/m2) and the Rohrer index (in kg/m3) were used as measures of relative weight in the current analyses; height was found to be moderately correlated (r = 0.35) with BMI but not with the Rohrer index (r = 0.01). Sex- and age-specific BMI percentiles were calculated by using data collected in national surveys between 1963 and 1994 (25).
Skinfold thicknesses (subscapular and triceps) and waist circumference were each measured 3 times, and the mean value was used in the analyses. Skinfold thicknesses were measured to the nearest millimeter with Lange skinfold calipers (Cambridge Scientific Instruments, Cambridge, MD), and waist circumference was measured midway between the rib cage and the superior border of the iliac crest. Waist circumference was measured only in sample 2.
Chemical analyses of lipids and lipoproteins
All chemical analyses were performed in the Bogalusa Heart Study Core Laboratory. Serum concentrations of total cholesterol and triacylglycerol were determined with the use of enzymatic procedures (Abbott VP, North Chicago, IL). After the heparin-calcium precipitation of ß- and pre-ß-lipoproteins, the concentration of LDL cholesterol was measured from the densitometric (electrophoretic) ratio and the cholesterol content of the 2 lipoproteins (26). The laboratory met the performance requirements of the Centers for Disease Control and Prevention Lipid Standardization Program (27). All analyses of the triacylglycerol and cholesterol (total, LDL, and HDL) concentrations in the current study are based on these chemical, rather than nuclear magnetic resonance (NMR) spectroscopy, determinations.
Nuclear magnetic resonance spectroscopy
Plasma samples, which had been stored at -70°C, were sent to LipoMed, Inc (Raleigh, NC), for the determination of lipoprotein subclasses. The proton NMR spectra of freshly thawed aliquots (0.25 mL) were acquired in duplicate at 47°C with a dedicated 400-MHz NMR analyzer. Spectral deconvolution was performed to determine the amplitudes of the lipid methyl group signals broadcast by particles of different size (28, 29), and these amplitudes were used to quantify the concentrations of the various lipoprotein subclasses. Because all lipids (cholesterol, cholesterol esters, triacylglycerol, and phospholipids) contribute to the signal for each subclass, the concentrations are unaffected by variations in lipid composition. The NMR signal amplitudes of the VLDL subclasses were converted to mass concentration units (mmol/L) of triacylglycerol for reporting purposes (29).
Six VLDL subclasses were quantified with the use of these techniques. To simplify the presentation of the data, adjacent subclasses were grouped together to yield 2 subclasses: large VLDL (diameter range: >40–200 nm) and small VLDL (27–40 nm). The grouping of adjacent subclasses improves the reproducibility of the lipoprotein subclass measurements (24), and the analyses indicated that adjacent subclasses were similarly related to obesity.
The diameters of the particles that made up each of the 6 VLDL subclasses were determined by electron microscopy of purified subfraction standards isolated from the plasma of a diverse group of normo- and hypertriglyceridemic individuals by ultracentrifugation (density < 1.006 kg/L) followed by gel chromatography on 2% agarose (30). No spectroscopic differences were discerned for particles of the same diameter isolated from donors of different ages or sexes. The accuracy and linearity of NMR-derived VLDL subclass concentrations were assessed by adding whole plasma specimens with VLDL subfractions of defined size obtained by agarose gel chromatography. Consistently, NMR correctly identified which subfraction was added and in what amount. The mean VLDL particle size in the current study, based on a mass-weighted average of the 6 VLDL subclasses, was 42 nm and the 99th percentile was 65 nm.
Statistical analyses
The initial analyses focused on the relation of the various obesity indexes to triacylglycerol concentrations in the first sample. Linear regression was used to compare the magnitudes of the associations across the various obesity indexes and race-sex groups. The interaction terms between the obesity indexes and race (eg, BMI x race) were included in these models to assess the statistical significance of the racial differences in the relation of waist circumference to concentrations of triacylglycerol and the VLDL subclasses. Similar techniques indicated that age did not modify the observed associations significantly.
The panel design resulted in children being eligible for examination in more than one study, and the 12524 observations used in these analyses were obtained from 7991 different children. Because serial measurements within an individual are not independent, PROC MIXED (with a first-order autoregressive covariance structure) of SAS (SAS Institute, Inc, Cary, NC) was used to compute all regression coefficients and P values. These regression coefficients, which account for the within-person correlations, typically differed by <5% from those obtained with the use of standard regression analyses.
The subsequent analyses focused on the 916 children examined in sample 2, in whom VLDL subclass determinations were made. Various percentiles of these subclasses were calculated, and associations with waist circumference were assessed by using Spearman rank-order correlation coefficients (rs), robust regression techniques (31), and lowess (locally weighted regression) curves with a robustness procedure. Lowess is a graphic technique that uses nearby data points to determine the functional form of the relation (32), rather than assuming a prespecified shape.
RESULTS
Some characteristics of the 10- to 17-y-olds in sample 1 are shown in Table 1. As indicated by the 3 relative weight indexes, black boys were thinner than white boys, whereas black girls were heavier than white girls. Although there was a difference in mean concentrations of total cholesterol between blacks and whites, differences in HDL cholesterol and triacylglycerol were more pronounced. Compared with the white children, the black children had a higher (by 0.2 mmol/L) mean HDL-cholesterol concentration and lower mean (by 0.16 mmol/L) and median (by 0.11 mmol/L) triacylglycerol concentrations.
View this table:
TABLE 1 . Characteristics of 10- to 17-y-olds: sample 1 from the Bogalusa Heart Study1
Associations with the various measures of obesity in the first sample are shown in Table 2; the regression coefficients represent the expected difference in triacylglycerol or HDL-cholesterol concentrations associated with a specified difference in each characteristic. For example, the estimated changes in triacylglycerol associated with a 1-kg/m2 difference in BMI ranged from 0.04 mmol/L in white boys to 0.01 mmol/L in black girls, whereas the corresponding changes associated with a 10-unit difference in BMI percentiles were 0.05 mmol/L in white boys and 0.01 mmol/L in black girls. (The relatively weak association between age and triacylglycerol concentrations was controlled in these analyses.) For each obesity measure, the magnitude of the association with triacylglycerol concentrations was approximately twice as large in white children than in black children. The racial differences in the associations with concentrations of HDL cholesterol were weaker, with P values for the interaction terms ranging from <0.01 to 0.80.
View this table:
TABLE 2 . Relation of various obesity indexes to triacylglycerol concentrations in 10- to 17-y-olds, by race and sex: sample 1 from the Bogalusa Heart Study1
The results in Figure 1 (and subsequent analyses) are limited to the 916 children in sample 2 in whom VLDL subclasses were determined. The strongest correlation was between triacylglycerol concentrations and waist circumference (r = 0.36). By regression analyses, the association was nonlinear, and the racial difference in the strength of the association resulted in differences in mean triacylglycerol concentrations between blacks and whites that increased with the level of obesity. For example, whereas the mean triacylglycerol concentration was 0.2 mmol/L higher in the white children in the lowest quintile of waist circumference, the difference was 0.6 mmol/L higher in the highest quintile of waist circumference. The black children with waist circumferences above the 90th percentile (91 cm) had an estimated triacylglycerol concentration (0.9 mmol/L) that was not different from that of the relatively thin white children.
FIGURE 1. . Relation of waist circumference to triacylglycerol concentrations by sex and race. Within each race-sex group, lowess (locally weighted regression) curves were constructed by using a robust linear regression fitting procedure and a neighborhood width of 50%. These smoothed levels, which are somewhat comparable with moving averages, roughly correspond to the median triacylglycerol concentration at various waist circumferences. The mean waist circumference ranged from 71 to 75 cm across race-sex groups. The relation of waist circumference to triacylglycerol concentrations was not linear, P < 0.001 (regression analysis).
The mean concentrations of both VLDL subclasses were significantly higher in the white than in the black children (Table 3). The 0.3-mmol/L difference in mean triacylglycerol concentrations, however, was primarily attributable to the 0.2-mmol/L (140%) higher mean concentration of large VLDL in the white children. In contrast, the comparable percentage difference in concentrations of small VLDL was only 29%. Despite the skewed distribution of large VLDL, the median concentration of large VLDL was 2.5-fold higher in the white than in the black children.
View this table:
TABLE 3 . Distribution of triacylglycerol and VLDL subclass concentrations in 10- to 17-y-olds, by race: sample 2 from the Bogalusa Heart Study1
The association between concentrations of triacylglycerol and large VLDL was strong (rs = 0.88); therefore, we assessed whether the racial difference in large VLDL could be attributable to triacylglycerol concentration. The control for triacylglycerol concentrations in the regression models reduced the racial differences, but the higher concentration of large VLDL in the white children remained significant (P < 0.001). These analyses also indicated that the racial difference was most evident at relatively high triacylglycerol concentrations, with the higher concentrations of large VLDL in whites increasing from 0.02 mmol/L (at the median triacylglycerol concentration) to 0.2 mmol/L (at the 95th percentile for triacylglycerol concentration).
The relations between waist circumference and VLDL subclass concentrations are shown in Table 4; the regression coefficients represent the expected differences in triacylglycerol and VLDL subclass concentrations between children whose waist circumferences differed by 10 cm. For example, a 10-cm difference in waist circumference was associated with a 0.15-mmol/L (in whites) and 0.05-mmol/L (in blacks) significantly higher triacylglycerol concentration. Concentrations of large VLDL were also significantly related to waist circumference; the estimated regression coefficient in the whites was 6 times that in blacks. Concentrations of small VLDL were not significantly associated with waist circumference in blacks or whites.
View this table:
TABLE 4 . Relation of waist circumference to VLDL subclass concentrations in 10- to 17-y-olds: sample 2 from the Bogalusa Heart Study1
As seen in Figure 2, the relation of waist circumference to concentrations of large VLDL among white children was not linear. At waist circumferences lower than 70 cm (close to the median value), concentrations of large VLDL in white children varied only slightly with changes in waist circumference. Concentrations of large VLDL, however, increased rapidly as waist circumferences increased, and the estimated concentration in white children with a waist circumference of 100 cm (the 95th percentile) was 0.6 mmol/L. These analyses also confirmed that waist circumference was only weakly associated with concentrations of large VLDL in black children and was not associated with concentrations of small VLDL in either race. Despite the increase in waist circumference with age, stratification and regression analyses indicated that these associations did not differ significantly by age (data not shown).
FIGURE 2. . Relation of waist circumference to concentrations of large and small VLDL by race. Within each sex group, lowess (locally weighted regression) curves were constructed by using a robust linear regression fitting procedure and a neighborhood width of 50%. The relation of waist circumference to concentrations of large VLDL among white children was not linear, P < 0.001 (regression analysis).
DISCUSSION
Our results indicate that the large VLDL subclass was primarily responsible for the observed racial differences in mean triacylglycerol concentrations and in the relation of obesity to triacylglycerol concentrations. Concentrations of large VLDL in white children were 2.5 times those in black children and showed a 6-fold stronger association with waist circumference in white children. In contrast, the concentration of small VLDL was not significantly associated with obesity and differed only slightly between black and white children. It is possible that the analysis of VLDL subclasses could provide additional information on the relation of obesity to ischemic heart disease.
Triacylglycerols, which are primarily transported by VLDL particles, play a central role in lipoprotein metabolism (4). When triacylglycerol concentrations are measured, they are typically measured as total concentrations; however, VLDL subclasses are metabolically distinct (33, 34), and the atherogenicity of VLDL particles can vary by size. Although some findings have emphasized the importance of small VLDL particles (including remnants) in the risk of atherosclerosis (35, 36), large VLDL particles may also play a role. Large hypertriglyceridemic VLDL can bind to LDL receptors and lead to the formation of cholesterol-rich foam cells (4, 33). In addition, occlusive disease is associated with concentrations of large VLDL (13) and with postprandial lipemia (37), a condition in which concentrations of larger VLDL subclasses are elevated (38). Although VLDL particles with a diameter >75 nm are unlikely to enter the arterial wall, slightly smaller particles are strongly retained in the intima (39), and a detailed breakdown of the NMR data indicated that >90% of the large VLDL particles in the current study had diameters of 40–80 nm. Large VLDL particles are also metabolically linked with small LDL and HDL particles (3, 7, 37), a profile that could promote atherosclerosis. Additional research, however, is needed to determine the importance of specific VLDL subclasses in atherosclerosis and ischemic heart disease.
Insulin-resistant persons have an altered VLDL structure and metabolism and elevated concentrations of the larger, triacylglycerol-rich VLDL subclasses (7). Although various obesity indexes also appear to be more strongly associated with concentrations of large than with small VLDL subclasses (6, 8, 9), these associations have not been examined in children or nonwhites. In agreement with our results concerning the nonlinear association with concentrations of large VLDL, Sattar et al (9) also found a threshold for waist circumference, with concentrations of both triacylglycerols and small LDLs increasing as obesity increases. Of the various anthropometric dimensions, we found waist circumference to show the most consistent and strongest associations with adverse risk factors (1), probably reflecting its ability to quantify both the amount and distribution of body fat. The observed, nonlinear associations with waist circumference may provide additional support for the use of cutoffs in health promotion programs (40).
There are well-documented racial differences in triacylglycerol concentrations among adults, with white adults and children having mean concentrations that are 25% higher than those of blacks (15, 22, 41). We found that the 0.3-mmol/L higher triacylglycerol concentration in whites was almost entirely attributable to the difference in large VLDL concentrations. In addition, previously reported racial differences in the relation of obesity to triacylglycerol concentrations appear to be attributable to different associations with large VLDL (14, 16, 17), which resulted in a larger difference in the concentration of large VLDL among obese children than among thinner children in the current study (Figure 2). This finding may be related to the racial differences in the relation of obesity to ischemic heart disease and total mortality that were reported in some studies (42, 43). In the current study we found no significant association between small VLDL and obesity.
Our findings are consistent with the suggestion that blacks have a greater clearance of plasma triacylglycerol than do whites and that the racial difference in the concentration of large VLDL is due, at least in part, to the activity of lipoprotein lipase—an enzyme involved in the hydrolysis of VLDL triacylglycerol (16). The activity of lipoprotein lipase, which is higher in blacks than in whites (18–20), may also be inversely associated with large VLDL concentrations and with obesity (7, 18, 44). Black men, however, have higher rates of lipoprotein lipase activity and epinephrine-stimulated lipolysis than do white men, even at similar levels of body fat (18). Differences in the activity of lipoprotein lipase may possibly account for the differences between blacks and whites in concentrations of HDL cholesterol (20) and large VLDL as well as for the increase in the magnitude of the difference in large VLDL concentrations between black and white children at larger waist circumferences.
The number of subjects in the current analysis was substantially greater than the number of subjects in other studies of VLDL subclasses (6–9, 38), but several points should be considered when interpreting our results, which are based on NMR spectroscopy. The methodologic differences between NMR and various separation techniques, such as cumulative flotation ultracentrifugation and analytic centrifugation, that have been used to quantify VLDL subclasses make it difficult to quantitatively compare results. However, the associations between VLDL subclasses and obesity determined with the use of NMR were found to be similar to those in other studies (6, 8, 9) and thus support the validity of the use of NMR. Although the current analyses focused on the relation of obesity to the concentration of triacylglycerol and VLDL subclasses, we showed previously that obesity is inversely associated with the size of LDL and HDL particles (24, 45).
We also showed previously that NMR-determined concentrations of large VLDL predict the extent of arteriographically documented occlusive disease, and that this association was independent of triacylglycerol concentrations (13). Our current findings suggest that the measurement of VLDL subclasses may provide a better understanding of the role of race, obesity, and triacylglycerol concentrations in the development of ischemic heart disease.
REFERENCES