Relation of serum lipoprotein lipids and apolipoproteins to obesity in children: The Bogalusa heart study

PREVENTIVE

MEDICINE

21, 177-190 (1992)

Relation of Serum Lipoprotein Lipids and Apolipoproteins Obesity in Children: The Bogalusa Heart Study’ DAVID

to

A. KIKUCHI, PH.D., M.P.H. ,*‘-I SATHANUR R. SRINIVASAN, PH.D. ,*,$ DAVID W. HARSHA, PH.D. ,* LARRY S. WEBBER, PH.D. ,**t THOMAS A. SELLERS, PH.D.,? AND GERALD S. BERENSON, M.D.*T2

Departments of *Medicine, tBiometty and Genetics, and #Biochemistry, Louisiana State University Medical Center, 1542 Tulane Avenue, New Orleans, Louisiana 70112-2822 Background. The relationship of serum lipoprotein lipids and apolipoproteins to obesity was studied in a biracial sample of 2,816 children of ages 5-17 in Bogahtsa, Louisiana. Methods. Two measures of obesity were used: fatness (subscapular skinfold thickness) and fat centrality (the ratio of subscapular to triceps skinfold thickness). Plasma insulin and glucose were included as metabolic markers related to obesity. Results. The obesity associations were relatively strong with insulin (rs = 0.29, P < 0.001, skinfold; rs = 0.15, P < 0.001, skinfold ratio) and triglycerides (rs = 0.25, P < 0.001, skinfold; rs = 0.19, P < 0.001, skinfold ratio). The relationships of serum low-density lipoprotein cholesterol (LDL-C) (rS = 0.17, P < 0.001, skinfold; rs = 0.12, P < 0.001, skinfold ratio) and apolipoprotein (apo) B (rS = 0.16, P < 0.001, skinfold; rs = 0.13, P < 0.001, skinfold ratio) with the obesity measures were of lesser magnitude, but persisted after adjustment for insulin and triglycerides. The inverse association of obesity to serum highdensity lipoprotein cholesterol (HDL-C) (rs = - 0.13, P < 0.001, both skinfold and skinfold ratio) and apo A-I (rs = -0.04, P = 0.03, skinfold; rs = -0.05, P = 0.004, skinfold ratio) was significant only before adjustment for insulin and serum triglycerides. Multiple linear regression of obesity measures showed that, like insulin, serum triglycerides had consistently higher standardized coefftcients than LDL-C, HDL-C, apo B, and apo A-I. Apo A-I and apo B added only a small amount (~2%) of information to the relationship of serum lipoproteins with obesity measures. Conclusion. These results indicate that serum very-low-density lipoprotein (VLDL) levels are directly and independently related to obesity. The well-known inverse association between obesity and serum HDL-C is not independent, but secondary to the elevated VLDL or triglyceride levels associated with obesity. While associations of obesity and lipoprotein cholesterol are found, far fewer occur with apolipoproteins, especially Apo A-I. Interesting race and sex differences in the relationship of obesity to serum lipoproteins and apoproteins are noted, being greater among white children and more consistent in white males. 0 1992 Academic

Press. Inc.

INTRODUCTION

The association between obesity and risk for cardiovascular disease (CVD) is well established (22, 27, 42). The influence of obesity on cardiovascular risk is thought to be mediated by hyperlipidemia, hyperinsulinemia, and hypertension (11, 30, 42, 46). Several studies have shown that obesity has adverse effects on serum lipids and lipoproteins (1, 9, 20, 23, 24). ’ This research is supported by Research Grant ROl-HL-38844 from the National Heart, Lung, and Blood Institute of the United States Public Health Service. * To whom reprint requests should be addressed. 177 0091-7435192 $3.00 Copyright 0 1992 by Academic Press, Inc. All rights of reproduction in any form reserved.

178

KIKUCHI

ET

AL.

Since pathological precursors of cardiovascular disease begin in childhood (3,4, 31), recent studies have examined serum lipid and lipoprotein levels and their interrelationship to obesity in children (13, 18, 28,45). In general, the lipoprotein changes related to obesity have been measured in terms of lipoprotein cholesterol, which assumes this as a measure of lipoprotein particles. Comparable information is lacking on apolipoproteins (apo), which are an integral part of lipid transport and metabolism. Among the apolipoproteins, apo B and apo A-I are of particular interest. Apo B is the predominant protein constituent of very-low-density lipoprotein (VLDL) and low-density lipoprotein (LDL) and serves as the principal carrier of lipids to peripheral tissues. On the other hand, apo A-I is the major protein constituent of high-density lipoprotein (HDL), involved in the reverse transport of cholesterol from peripheral tissue (21). The measurement of these apolipoproteins reflect more accurately changes of the LDL and HDL molecular concentrations. Therefore, a divergence in the relationship of obesity to levels of serum lipoprotein lipids and apolipoproteins might be expected. This report examines the relation of serum lipoprotein lipids and apolipoproteins to measures of obesity in a biracial (black-white) sample of 2,816 5- to 17-year-old children. Previous analyses of apo A-I and apo B levels in this study sample have examined race-, sex-, and age-related differences (40). Since levels of plasma insulin and glucose are strongly related to obesity and serum lipoproteins (2, 6, 12, 14, 34, 36, 42, 45, 46), these variables were also included as metabolic markers in the present study. The following studies provide an opportunity to explore the relationship between obesity and serum lipoprotein cholesterol versus apolipoproteins in a total biracial population of children and adolescents. METHODS Population. The Bogalusa Heart Study is a prospective epidemiologic study of CVD risk factors from birth to early adulthood (3,4). The sample of 2,854 fasting children that forms the basis for the current analyses consists of children, ages 5-17, screened during 1981-1982 (40). Thirty-eight (1.3%) children were excluded due to missing laboratory values. General examinations. The Louisiana State University Medical Center Institutional Review Board reviewed the studies on human subjects. Informed consent (written) was obtained from a parent or guardian of each child. Standardized examination protocols have been used during the entire span of the project (4). Subjects were instructed to fast for 12 hr before venipuncture, with compliance confirmed by interview. There were two manual measurements each of height to within 0.1 cm and weight to within 0.1 kg. Obesity measures such as subscapular skinfold (SSSF) and right-arm triceps skinfold (TRSF) thicknesses were recorded three times each to within 1 mm by Lange skinfold calipers (Cambridge Scientific, Cambridge, MA). The means of these multiple anthropometric measurements were used for all analyses. Ponderal index (kg/m3) was used rather than the Quetelet index (weight/height2) for body mass index because of a smaller correlation with height (44). The central (SSSF)-peripheral (TRSF) fat pattern

LIPOPROTEINS

AND

OBESITY

179

was estimated by the ratio SSSF/TRSF. Sexual maturation scored from 1 (prepubescence) to 5 (full development) was assessed by a physician based on female breast or male genitalia development according to the method of Tanner (43). Collection of blood specimens. Antecubital venous blood samples were collected in sterile vacuum tubes and allowed to clot for about 1% hr. After centrifugation, the sera were collected in tubes containing Thimerosal (Aldrich Chemical Co., Milwaukee, WI) and sent in a cold-packed box to the New Orleans Core Lipid Laboratory. Every screening day, independent blind duplicate blood samples were collected on a 10% random subsample of the children to estimate measurement error (16, 40). Laboratory analyses. Concentrations of serum total cholesterol (TC) and triglycerides (TG) were measured with an AutoAnalyzer II (Technicon Instrument Corp., Tarrytown, NY) with the use of protocols developed by the Lipid Research Clinics (29). Our Core Lipid Laboratory has been standardized by the Centers for Disease Control (Atlanta, GA) and is monitored by a surveillance program. Serum levels of VLDL-C, LDL-C, and HDL-C were analyzed with a combination of heparin-calcium precipitation and agar-agarose gel electrophoresis (41). The levels of serum apo B and apo A-I were measured by electroimmunoassay (40). Plasma glucose was measured with a Beckman glucose analyzer, Model ERA2001 (Beckman Instruments, Inc., Fullerton, CA) and plasma insulin with the radioimmunoassay of the Phadebas insulin test (Pharmacia Diagnostics, Piscataway, NJ). Statistical methods. Means of the study variables were computed using the Statistical Analysis System, with adjustment for concomitant variables made by computing population marginal means (38). Adjustment for age was up to cubic power due to its nonlinear relationship with the other study variables. Spearman rank correlation coefficients (rs) were computed for the obesity measures and physiologic variables. They were calculated from the residuals of linear regressions on race, sex, age (to cubic power), and sexual maturity in order to remove the confounding effects of these covariates. When the race-sex groups were studied separately, the only adjustment was for age (to cubic power) and sexual maturity. We also adjusted for insulin and serum TG as additional covariates in a similar manner. Age-, race-, and sex-specific quintile rankings for subscapular skinfolds and subscapular/triceps skinfold ratios were obtained. The data were then divided into three age ranges (5-9 years, l&14 years, and 1.5-17 years). For each race, sex, and age group the percentage differences in the levels of various lipids, lipoproteins, and apolipoproteins for the second to fifth quintile were compared with the levels in the first quintile. In this way, for each age group the mean age was the same for each subscapular skinfold or subscapular/triceps skinfold ratio quintile. For each obesity measure, we compared its relationships with the various serum and plasma variables by means of standardized coefficients (i.e., the parameter estimates divided by their standard errors) in multiple linear regression models. This definition of standardized coefficients differs from those involving standard deviations of the independent and dependent variables in a linear regression (19). The significance of these coefficients was computed from standard normal

180

KIKUCHI

ET AL.

tables. Serum VLDL-C was not included as an independent variable to avoid a singular design matrix arising from having TC, VLDL-C, LDL-C, and HDL-C in the model. Since the obesity measures were skewed to the right, they were logtransformed using natural logarithms to approach normality. The ponderal index was not reported, since it is less reflective of central fatness than subscapular skinfold. The independent variables were the serum lipoprotein lipids and apolipoproteins, glucose, insulin, age, and sexual maturity. Each of these independent variables (except sexual maturity) was fit to cubic power to account for possible curvilinear relationships. The quadratic and cubic terms for these variables (except age and sexual maturity) were subsequently deleted if not found to be significant. In alternate analyses, partial correlation coefftcients of the dependent variable were compared for each independent variable, with adjustment for the other independent variables. RESULTS Distributions and mean levels of study variables by race and sex (Table I). ?he distributions, mean levels, and race- and sex-related differences are provided as background information, although these study variables of this population have been discussed in depth previously (15, 17, 29,40). As expected, males were less obese than females, and black males less than white males. Nevertheless, the ratio of trunk to limb fat (SSSF/TRSF) was greater among blacks than whites and greater among males than females for each race. Females had greater levels of serum TC, TG, VLDL-C, and LDL-C, and lower levels of HDL-C than males. Blacks had greater levels of serum TC and HDL-C but lower levels of TG and VLDL-C than whites. Females had greater levels of apo B than males, and blacks greater levels of apo A-I than whites. Interrelationships of serum and plasma variables (Table 2). The associations between serum and plasma variables are in the expected direction and have the order of magnitude of correlations previously reported in this population of children (6, 40). Of particular interest is the observation that TG and VLDL-C were correlated positively with both glucose and insulin, and HDL-C and apo A-I were correlated negatively with insulin. Both LDL-C and apo B were associated positively with insulin. The negative correlations of TG and VLDL-C were stronger with HDL-C than with apo A-I. While LDL-C was strongly related to apo B, HDL-C was only moderately related to apo A-I. The associations occurred in each of the four race-sex groups and age groups (5-9, l&14, and 15-17 years) (data not shown). Interrelationships of obesity measures (Table 3). The obesity measures were highly correlated with each other, with the exception of SSSF/TRSF. The correlation coefficients involving SSSF/TRSF were relatively low and masked a wide range in values between the four race-sex groups and the ages 5-9, 10-14, and 15-17 years (data not shown). For example, the Spearman correlation between SSSF/TRSF and TRSF for black males ages 15-17 was rs = -0.55, while for white females ages 5-9 it was rs = 0.33. We selected SSSF and SSSFiTRSF as obesity measures in further analyses with serum lipoprotein lipids and apolipoproteins. These measures correspond approximately to fatness and fat centrality,

LIPOPROTEINS

MEAN

LEVELS

OF STUDY

VARIABLES

AND

TABLE 1 BY RACE AND SEX: THE BOGALLJSA

White Variables

Males (n = 911)

181

OBESITY

Black Females (n = 896)

Males (n = 475)

HEART

STUDY

Females (n = 535)

Race differences (P <)

Sex differences (P <)

11.1 f 0.1

0.005’

NS

0.002c 0.001 0.004 NS

0.001 0.001 0.002 O.OOld

13.2 k 0.1

0.026

0.001’

1.2 0.3 1.1 0.9

0.001 0.001 NS 0.001

0.05 O.OOP 0.004 O.OWd

0.001

0.04

NS O.OOP 0.001 0.001’

0.02 NS 0.001 0.001

mean + SE Age (years)” Skinfolds SSSF (mm) TRSF (mm) SSSFITRSF Weight (kg) Ponderal index Wm3) Cholesterol (mp/dl) Total VLDL LDL HDL Triglycerides (mddl) Apolipoproteins (mddl) Apo B Apo A-I Glucose (mg/dl) Insulin (pU/ml)

10.8 * 0.1 8.9 14.8 0.58 40.4

2 r + 2

0.2 0.2 0.01 0.3

13.1 ‘- 0.1 156.6 7.6 90.6 58.4

+ f + +

0.9 0.2 0.8 0.7

64.6 2 1.0 82.3 139.3 81.5 8.7

-t t -t +

0.7 0.7 0.2 0.2

10.6 ? 0.1 10.8 18.7 0.56 38.0

k -t f f

0.2 0.2 0.01 0.3

13.2 + 0.1 159.7 9.0 95.3 55.4

+ t f 2

0.9 0.2 0.8 0.7

71.5 2 1.0 85.8 139.8 79.7 10.0

+ + + f

0.7 0.7 0.2 0.2

10.8 2 0.2 7.8 11.8 0.67 39.8

‘f t +

0.3 0.3 0.01 0.4

12.8 t 0.1 163.3 5.7 91.4 66.2

+ + t 2

1.2 0.3 1.1 0.9

53.1 + 1.4 83.0 144.2 79.8 9.4

f + r 2

1.0 1.0 0.3 0.3

10.2 16.6 0.59 38.6

166.6 6.3 95.9 64.4

+ -c t it

f 2 It +

0.3 0.3 0.01 0.4

51.2 + 1.3

86.2 141.7 77.9 11.3

f 2 + r

0.9 0.9 0.3 0.3

Note. Abbreviations: SSSF, subscapular skinfold thickness; TRSF, triceps skinfold thickness; VLDL, very-lowdensity lipoprotein; LDL, low-density lipoprotein; HDL, high-density lipoprotein; NS, not significant. To express cholesterol and apolipoprotein values in mmohhters, triglyceride levels in mmohliter, glucose levels in mmohhter, and insulin levels in pmohhter, multiply by 0.02586, 0.1129, 0.05551, and 7.175, respectively. a Unadjusted means-the means of all other variables are adjusted for age (to cubic power) and sexual maturity. b Mates only. ’ Females only. d White only. e Blacks only.

respectively. Their degree of association varied between the four race-sex groups, as indicated by the range in the correlation coefftcients between them: us = 0.56, 0.77, 0.11, and 0.44 for white males, white females, black males, and black females, respectively. Correlations of serum and plasma variables with obesity measures. Partial correlation coefftcients of serum and plasma variables with obesity after adjustment for age, race, sex, and sexual maturity are presented in Fig. 1. The largest correlation coefficients of obesity measures were with insulin, followed by serum TG, LDL-C, and apo B. The inverse association with HDL-C was stronger than with apo A-I. The correlation coefficients for serum VLDL-C and TC were somewhat smaller than TG and LDL-C, respectively, but followed a similar pattern to them (data not shown). The associations of plasma and serum variables with SSSF/TRSF were significant but somewhat weaker than those with SSSF. The above correlations of LDL-C, HDL-C, apo B, and apo A-I with obesity measures were further adjusted for insulin and TG. Adjustment for insulin and serum TG eliminated the associations among SSSF, HDL-C, and apo A-I and those between SSSF/TRSF and apo A-I.

182

KIKUCHI

ET AL.

TABLE

2

SPEARMAN CORRELATION COEFFICIENTS BETWEEN SERUM LIWPROTEIN APOLIPOPROTEINS, GLUCOSE, AND INSULIN: THE BOGALIJSA HEART

TC TG VLDL-C LDL-C HDL-C Apo B Apo A-I Glucose

LIPIDS AND STUDY

TG

VLDL-C

LDL-C

HDL-C

Apo B

Apo A-I

Glucose

Insulin

0.24”

0.15” 0.77”

0.76” 0.41” 0.34”

0.31” - 0.43” -0.51” - 0.29”

0.74” 0.46” 0.42” 0.92” - 0.27”

0.36” - 0.09” -0.11” 0.01 0.50” 0.01

0.03 0.16” 0.11” -0.00 0.00 0.00 0.00

0.02 0.26” 0.21” 0.06b -0.12” 0.11” -0.12” 0.22”

Note. Coefficients are adjusted for race, sex, age (to cubic power), and sexual maturity. Abbreviations: see Table 1. * P < 0.001. b P -=c0.01.

The strength of the associations did not differ by race-sex group. Among black males, however, there was a decrease in the correlation coefficients of serum LDL-C from 0.02 to - 0.06 and apo B from 0.00 to - 0.03 with SSSFITRSF from SSSF. This relationship persisted after adjustment for insulin and serum TG. To explore a possible nonlinear relationship of SSSF and SSSF/TRSF with the serum lipoprotein lipids and apolipoproteins, we computed the percentage change in levels of the serum variables from the first quintile at each subsequent quintile of the obesity measures (Figs. 2-5). As noted in Figs. 2 and 3, with quintiles of subscapular skinfold, changes were noted especially in the top two quintiles. Greater changes were noted after adoTABLE SPEARMAN

CORRELATION

TRSF

3

COEFFICIENTS BETWEEN DIFFERENT MEASURES AND SEX: THE BOCALUSA HEART STUDY

S/-f

PI

wt

TRSF

0.73 0.69 0.35 0.68

0.80

White males SSSF TRSF SIT PI

0.80

0.56 0.07”

0.80

0.11” -0.40

S/T

PI

BY RACE

wt

White females 0.75 0.75 0.31

Black males SSSF TRSF S/T PI

OF OBESITY

0.77 0.30

0.78 0.76 0.50

0.75 0.70 0.49 0.71

Black females 0.67 0.69 -0.07

0.66 0.58 0.076 0.56

0.83

0.44 - o.03b

0.79 0.79 0.20

0.70 0.70 0.20 0.71

Note. All correlation coefficients are significant at P < 0.001 except where noted. All coefficients are adjusted for age (to cubic power) and sexual maturity. Abbreviations: see Table 1. S/T, subscapular/triceps ratio; PI, ponderal index; Wt, weight. = P < 0.05. b Not significant.

LIPOPROTEINS

AND

183

OBESITY

0.2 0

*

-0.2 wg$-

E9lnsulin

RHTG

1

I yFJ$

q$i-

*P
rape

A-l

FIG, 1. Spearman correlation coefficients of obesity measures with serum and plasma variables: The Bogalusa Heart Study. Abbreviations: see Table 1. All coefficients are adjusted for age (to cubic power), race, sex, and sexual maturity. In the right graph, they are also adjusted for insulin and triglycerides, showing a marked effect on HDL-C and Apo A-l.

lescence at 15-17 years, young adult ages. Much with earlier findings (3, increases and negative 5-9

Years

which suggest an increasing effect of obesity into the smaller effects were noted in black children, consistent 4). Major positive effects were reflected in serum TG effects on HDL-C levels. Serum LDL-C and apo B lo-14

Years

15-17

Years

123451234512345

Subscapular

Skinfold

Quintile

Fto. 2. Percentage change in serum lipoprotein lipids and apolipoproteins ular skinfold, race, and age for males: The Bogalusa Heart Study.

by quintiles of subscap-

184

KIKUCHI

ET AL.

1O-l 4 Years White Females

5-9 Years 100

15-17 Years l

TG

50 s 5

0

.- .r.f:

-.r.:j(4 .

.:.:.

Y-k -50 k Y= f3 z

Black Females

loo

s 250

bb

0 -50 12

3

4

5

Subscapular

(“1 Skinfold

5

1

5

Quintile

FIG. 3. Percentage change in serum lipoprotein lipids and apolipoproteins ular skinfold, race, and age for females: The Bogalusa Heart Study.

by quintiles of subscap-

showed positive increases with increasing central obesity, again being more exaggerated in the older age group and white adolescents, especially white males. In contrast to the negative influence on serum HDL-C, little or no effect was found on apo A-I. With regard to studies of the subscapular/triceps skinfold ratio, similar findings were noted, although more uniform (linear) changes as described above were found in the third, fourth, and fifth quintiles, as well as even in the second quintile for ages 15-17 years. The most dramatic evidence for nonlinearity in these graphs occurred for serum TG among white females of ages 5-14. Serum TG appeared to have a sharp increase of about 50% over the first quintile from the fourth to fifth quintiles of obesity measure. Regression of obesity measures on serum and plasma variables. To analyze the relationships of the serum and plasma variables simultaneously with SSSF and SSSFITRSF, we performed a multiple linear regression of the obesity measures (Table 4). For both SSSF and SSSF/TRSF, insulin and TG showed consistently higher standardized coefficient than LDL-C, HDL-C, apo B, and apo A-I, although there was considerable fluctuation between the four race-sex groups. Multiple squared correlation coefficients for the serum and plasma variables were in the range 0.25-0.40 (data not shown). The serum apolipoproteins added only a small amount of information to the relationship of serum and plasma variables to obesity measures, increasing the multiple squared correlation coefficients by <0.02. For SSSF these increases were significant for both white and black males, and for SSSF/TRSF they were significant only for white females.

LIPOPROTEINS

5-9 100

Years

AND

185

OBESITY

1O-l 4 Years White Males

15-17 Years . TG

Black Males

Subscapular/Triceps

Skinfold

Quintile

FIG. 4. Percentage change in serum lipoprotein lipids and apolipoproteins triceps skinfold ratio, race, and age for males: The Bogalusa Heart Study.

by the subscapular/

DISCUSSION

The results of the present study in a free-living pediatric population show the close relationship of obesity in children to carbohydrate and lipid metabolism. The observations suggest that the distribution of fat, central compared with peripheral, has an adverse impact on serum lipids and perhaps on carbohydrate metabolism, extending over the entire range of body fat distribution, and is a powerful determinant of lipoprotein lipid levels, one which increases in importance as children mature into adulthood. For example, in the 15 to 17-year-old males, who demonstrate the greatest impact from obesity, TG levels increased approximately 75% in the top quintile compared with those in the first quartile; HDL-C decreased approximately 45%. Several studies have considered the impact of obesity on serum lipoprotein lipids (13, 18, 28,47) and carbohydrate intolerance (5,25, 26, 37). The serum and plasma variables were ranked in order of the strength of their association with obesity to indicate that, like insulin, serum VLDL (triglycerides) is strongly and independently related to measures of obesity reflecting both fatness and fat centrality. In contrast, serum apo B and A-I contributed little to this association, and HDL-C changes only marginally among white males. In the present study, both insulin and glucose were related positively to serum TG and VLDL-C, as noted before (3,4). In addition, insulin was related positively to LDL-C and apo B and negatively to HDL-C and apo A-I, but of lower order. The associations between measures of carbohydrate metabolism and lipoprotein

186

KIKUCHI

ET AL.

lo-14 Years White Females

5-9 Years

1

15-l 7 Years l

1

TG

0 LDL-c

Black Females

Subscapular/Triceps

Skinfold

Quintile

FIG. 5. Percentage change in serum lipoprotein lipids and apolipoproteins triceps skinfold ratios, race, and age for females: The Bogalusa Heart Study.

by the subscapularl

variables are consistent with our previous findings (14, 36). Indeed, the efficiency of the insulin-glucose homeostatic mechanism may be a major determinant of obesity-related lipoprotein changes (35). The current finding that measures of obesity were strongly related to insulin levels supports this concept. Increased levels of insulin and its attendant insulin resistance may stimulate hepatic triglyceride synthesis and VLDL secretion without enhancing the catabolism of triglycerides (2, 33, 34). Since degradation of VLDL is probably the major (if not exclusive) source of LDL, there is a net increase in LDL production in relation to increased VLDL synthesis and hyperinsulinemia (10,39). The inverse association of insulin with HDL may arise from the negative association between VLDL and HDL (10, 32). This study found that the inverse associations of TG and VLDL-C were stronger with HDL-C than with apo A-I. Donahue et al. (9) reported that weight/height* was inversely related to levels of HDL-C but not apo A-I in young men. The divergent relationship of obesity to HDL-C and apo A-I probably arises from the lipid-transfer protein mediated exchange of cholesteryl esters for TG between HDL and VLDL (8). Obesity-related increases in TG levels seem to reflect more accurately decreases in levels of HDL-C rather than apo A-I. Moreover, the observed (inverse) associations of obesity with HDL-C were stronger than those with apo A-I. Controlling for insulin and TG levels eliminated the inverse associations of obesity with HDL-C and apo A-I. Likewise, the association of obesity with LDL-C and apo B diminished (if they did not disappear completely) after adjust-

LIPOPROTEINS

TABLE STANDARDIZED ON SERUM

187

AND OBESITY 4

REGRESSION COEFFICIENTS FROM MULTIPLE REGRESSION OF OBESITY MEASURES AND PLASMA VARIABLES BY RACE AND SEX: THE BOGALUSA HEART STUDY

log(SSSF)

log(SSSF/TRSF)

Whites

Insulin Linear Triglycerides Linear Quadratic Cubic LDL-C Linear Quadratic Cubic HDL-C Linear Quadratic Cubic Apo B Linear Quadratic Cubic Apo A-I Linear Glucose Linear

Blacks

Males

Females

7.05

8.29

-0.97 2.12 -2.28

Whites

Blacks

Females

Males

Females

5.32

7.06

2.66

7.59

0.05

3.80

3.87 -

3.74 -

3.16 -

3.25 -

3.46 -

0.60 -

2.32 -

-

1.00 -

1.95 -

2.59 -

1.17 -

3.10 -2.92 2.81

-0.84 -

- 2.35 2.84 -3.01

-0.63 -

1.25 -

- 1.44 -

-0.23 -

-2.07 -

1.20

-2.56

3.29

0.41

0.40

Males

Males

-1.36 -

-2.27 -

-1.99 2.21 -2.33

0.83 -

- 0.63 -

-2.98 2.82 -2.69

1.29 -

0.86

-0.76

1.67

- 1.56

- 1.72

0.75

-0.02

- 1.12

- 1.07

-0.00

Females

1.70 0.38 -0.74 1.55 - 1.58

Note. Logarithms are natural logarithms. Abbreviations: see Tables 1 and 2. Critical values of a two-sided test of significance for P = 0.05, 0.01, and 0.01 are 1.960, 2.576, and 3.291, respectively.

ment for insulin and TG levels. Furthermore, the multiple linear regression on obesity measures showed that, like insulin, serum TG (VLDL) had consistently higher standardized coefficients than any other lipoprotein variables. As known in adults, serum TG and VLDL-C levels are fundamentally linked to the status of obesity. The impact of obesity on HDL-C appears to be mediated by elevated insulin and TG levels. Since the metabolic associations among VLDL-C, LDL-C, and HDL-C are strong (10, 21), the apparent influence of obesity on LDL and HDL may be due to an increased production and concentration of VLDL. In the present study, centrality of fat deposition as gauged by either the level of SSSF or the ratio SSSF/TRSF showed significant interrelationship among central fat, insulin, and glucose just as found in adults (25, 27, 47). Such research is only recently being investigated in children (14,26, 45). In this respect, it is interesting to note that, although blacks, especially males, tend to be leaner than whites, they disproportionately deposit fat centrally. But, the effect is much less and may not begin to occur until adult ages or at more extreme levels of obesity. It should be mentioned that the interrelationship among dyslipidemia, glucose

188

KIKUCHI

ET AL.

tolerance, and central obesity may be stronger with deep abdominal obesity rather than abdominal obesity per se. Sophisticated tools such as computed tomography and magnetic resonance images are needed to determine deep abdominal obesity accurately. Nonetheless, the present study shows the usefulness of simple obesity measures such as SSSF and SSSF/TRSF in evaluating the relation of central fat deposition to carbohydrate-lipid metabolism in population-based studies. CONCLUSION From a public health viewpoint, the common occurrence of the associations we found in childhood underscores the need for preventive measures (e.g., interventions on diet and physical activity) to begin early in life. From a practical point of view, the uppermost quintile of subscapular skinfold measurement may be considered obesity-associated with adverse lipid and lipoprotein changes. Therefore, it is recommended that subscapular skinfold should be included as part of a physical examination. Interestingly, obese adult black females begin to manifest obesity around 10 years of age (3, 4). Despite the low order of relationships in blacks noted in this study, it will be important to follow such changes into adulthood. The impact of obesity and the unusually high insulin levels in black girls may presage the greater incidence of diabetes mellitus seen in black middle-aged females (46). From a practical standpoint primary care physicians should be alert to the importance of even subtle obesity and its effect on carbohydrate-lipid metabolism. In addition to height and weight, truncal obesity should be followed in children as part of cardiovascular risk evaluation. ACKNOWLEDGMENTS The Bogahtsa Heart Study is a joint effort of numerous people. We thank Mrs. Bettye Seal for her work as community coordinator, the nurses of the Bogalusa project staff, teachers and staff of the Bogalusa school system, Bogalusa community volunteers, and, finally, the children of Bogalusa and their parents for making this study possible.

REFERENCES 1. Albrink MJ, Meigs JW, Granoff MA. Weight gain and serum triglycerides in normal men. New Engl J Med 1%2; 266:4&l-489. 2. Bagdade JD, Bierman EL, Porte D Jr. Influence of obesity on the relationship between insulin and triglyceride levels in endogenous hypertriglyceridemia. Diabetes 1971; 20564-672. 3. Berenson GS (Ed.). Causation of Cardiovascular Risk Factors in Children: Perspectives on Cardiovascular Risk in Early Life. New York: Raven Press, 1986. 4. Berenson GS, McMahon CA, Voors AW, Webber LS, Srinivasan SR, Frank GC, Foster TA, Blonde CV. Cardiovascular Risk Factors in Children: The Early Natural History of Atherosclerosis and Essential Hypertension. New York: Oxford Univ. Press, 1980. 5. Brook CGD, Lloyd JK. Adipose cell size and glucose tolerance in obese children and effects of diet. Arch Dis Child 1973; 48:301-304. 6. Burke GL, Webber LS, Srinivasan SR, Radhakrishnamurthy B, Freedman DS, Berenson GS. Fasting plasma glucose and insulin levels and their relationship to cardiovascular risk factors in children: The Bogahtsa Heart Study. Metabolism 1986; 354414l6. 7. Cambien F, Wamet JM, Eschwege E, Jacqueson A, Richard JL, Rosselin G. Body mass, blood pressure, glucose, and lipids: Does plasma insulin explain their relationships? Arteriosclerosis 1987; 7:197-202.

LIPOPROTEINS

AND

OBESITY

189

8. Deckelbaum RJ, Granot E, Oschry Y, Rose L, Eisenberg S. Plasma triglyceride determines structure-composition in low and high density lipoproteins. Arteriosclerosis 1984; 4~225-231. 9. Donahue RP, Orchard TJ, Stein EA, Kuller LH. Apolipoproteins AI, AH, and B in young adults: Associations with CHD risk factors: The Beaver County experience. .I Chronic Dis 1986; 39823-830. 10. Eisenberg S. Plasma lipoprotein conversions: The origins of low-density and high-density lipoproteins. Ann NY Acad Sci 1980; 348:30-47. 11. Epstein FH. The epidemiology of coronary heart disease: A review. J Chronic Dis 1965; 18:735774. 12. Ford S, Bozian RC, Knowles HC Jr. Interactions of obesity, and glucose and insulin levels of hypertriglyceridemia. Am J Clin Nutr 1968; 21904-910. 13. Freedman DS, Burke GL, Harsha DW, Srinivasan SR, Cresanta JL, Webber LS, Berenson GS. Relationship of changes in obesity to serum lipid and lipoprotein changes in childhood and adolescence. JAMA 1985; 254~515-520. 14. Freedman DS, Srinivasan SR, Burke CL, Shear CL, Smoak CG, Harsha DW, Webber LW, Berenson GS. Relation of body fat distribution to hyperinsulinemia in children and adolescents: The Bogalusa Heart Study. Am J Clin Nutr 1987; 46~403-410. 15. Freedman DS, Srinivasan SR, Webber LS, Berenson GS. Divergent levels of high density lipoprotein cholesterol and apolipoprotein A-I in children: The Bogalusa Heart Study. Arteriosclerosis 1987; 7~347-353. 16. Foster TA, Berenson GS. Measurement error and reliability in four pediatric cross-sectional surveys of cardiovascular disease risk factor variables-The Bogalusa Heart Study. J Chronic Dis 1987; 40~13-21. 17. Frerichs RR, Srinivasan SR, Webber LS, Berenson GS. Serum cholesterol and triglyceride levels in 3,446 children from a biracial community: The Bogalusa Heart Study. Circulation 1976; 54~302-309. 18. Frerichs RR, Webber LS, Srinivasan SR, Berenson GS. Relation of serum lipids and lipoproteins to obesity and sexual maturity in white and black children. Am J Epidemiol 1978; 108:48ti96. 19. Greenland S, Schlesselman JJ, Criqui MH. The fallacy of employing standardized regression coefftcients and correlations as measures of effect. Am J Epidemiol 1986; 123:203-208. 20. Harlan WR Jr, Oberman A, Mitchell RE, Graybiel A. Constitutional and environmental factors related to serum lipid and lipoprotein levels. Ann Intern Med 1967; 66:540-555. 21. Have1 RJ, Goldstein JL, Brown MS. Lipoproteins and lipid transport. In: Bondy PK, Rosenberg LE, Eds. Metabolic Control and Disease. 8th ed. Philadelphia; Saunders, 1980:393494. 22. Hubert HB, Feinleib M, McNamara PM, Castelli WP. Obesity as an independent risk factor for cardiovascular disease: A 26-year follow-up of participants in the Framingham Heart Study. Circulation 1983; 67:968977. 23. Kannel WB, Gordon T, Castelli WP. Obesity, lipids, and glucose intolerance: The Framingham Study. Am J Clin Nutr 1979; 32:1,238-1,245. 24. Khoury P, Morrison JA, Mellies M. Weight change since age 18 years in 30- to 55-year-old whites and blacks: Associations with lipid values, lipoprotein levels, and blood pressure. JAMA 1983; 250:3,179-3,187. 25. Kissebah AH, Vydelingum N, Murray R, Evans DJ, Hartz AJ, Kalkhoff RK, Adams PW. Relation of body fat distribution to metabolic complications of obesity. J Clin Endocrinol Metab 1982; 54~254-260. 26. Kolterman OG, Insel J, Saekow M, Olefsky JM. Mechanisms of insulin resistance in human obesity: Evidence for receptor and postreceptor defects. J Clin Invest 1980; 65:1,272-l ,284. 27. Lapidus L, Bengtsson C, Larsson B, Pennert K, Rybo E, Sjostrom L. Distribution of adipose tissue and risk of cardiovascular disease and death: A 12-year follow up of participants in the population study of women in Gothenburg, Sweden. Br Med J 1984; 2891,257-l ,261. 28. Laskarzewski P, Morrison JA, Mellies MJ, Kelly K, Gartside PS, Khoury P, Glueck CJ. Relationships of measurements of body mass to plasma lipoproteins in schoolchildren and adults. Am J Epidemiol 1980; 111:395-406. 29. Lipid Research Clinics Program. Manual of Laboratory Operations, Vol. 1, Lipid and Lipoprotein

190

KIKUCHI

ET AL.

Analyses. DHEW Publication No. (NIH) 75-628. Washington, DC: National Institutes of Health, 1974. 30. Modan M, Halkin H, Almog S, Lusky A, Eshkol A, Shefi M, Shitrit A, Fuchs Z. Hyperinsulinemia: A link between hypertension obesity and glucose intolerance. J C/in Invest 1985; 75:809817. 31. Newman WP III, Freedman DS, Voors AW, Gard PD, Srinivasan SR, Cresanta JL, Williamson GD, Webber LS, Berenson GS. Relation of serum lipoprotein levels and systolic blood pressure to early atherosclerosis: The Bogalusa Heart Study. N Engl J Med 1986; 314138-144. 32. Nichols AV. Human serum lipoproteins and their interrelationships. Adv Rio/ Med Phys 1967; 11:lW158.

33. Olefsky JM, Farquhar JW, Reaven GM. Reappraisal of the role of insulin in hypertriglyceridemia. AmJMed

1974;57:551-560.

34. Olefsky JM, Kolterman OG. Mechanisms of insulin resistance in obesity and noninsulindependent (type II) diabetes. Am .I Med 1981; 70:151-168. 35. Orchard TJ, Becker DJ, Bates M, Kuller LH, Drash AL. Plasma insulin and lipoprotein concentrations: An atherogenic association? Am J Epidemiol 1983; 118:32&337. 36. Radhakrishnamurthy B, Srinivasan SR, Webber LS, Dalferes ER Jr, Berenson GS. Relationship of carbohydrate intolerance to serum lipoprotein profiles in childhood: The Bogalusa Heart Study. Metabolism 1985; 34:850-860. 37. Salans LB, Knittle JL, Hirsch J. The role of adipose cell size and adipose tissue insulin sensitivity in the carbohydrate intolerance of human obesity. J C/in Invest 1968; 47:153-165. 38. SAS Institute, Inc. SAS User’s Guide: Statistics. Version 5. Cary, NC: SAS Institute, 1985:483484. 39. Sigurdsson G, Nicoll A, Lewis B. Conversion of very low density lipoprotein to low density lipoprotein: A metabolic study of apolipoprotein B kinetics in human subjects. J Clin Invest 1975;56:1,481-1,490.

40. Srinivasan SR, Freedman DS, Sharma C, Webber LS, Berenson GS. Serum apolipoproteins A-I and B in 2,854 children from a biracial community: Bogalusa Heart Study. Pediatrics 1986; 78:18%200.

41. Srinivasan SR, Frerichs RR, Webber LS, Berenson GS. Serum lipoprotein profile in children from a biracial community. Circulation 1976; 54:3tJ9319. 42. Stem MP, Haffner SM. Body fat distribution and hyperinsulinemia as risk factors for diabetes and cardiovascular disease. Arteriosclerosis 1986; 6:123-130. 43. Tanner JM. Growth at Adolescence. 2nd ed. Oxford: Blackwell Scientific, 1%2:28-39. 44. Voors AW, Harsha DW, Webber LS, Berenson GS. Relationship of blood pressure to stature in healthy young adults. Am .I Epidemiol 1982; 115:833-840. 45. Voors AW, Harsha DW, Webber LS, Radhakrishnamurthy B, Srinivasan SR, Berenson GS. Clustering of anthropometric parameters, glucose tolerance, and serum lipids in children with high and low beta- and pre-beta-lipoproteins: Bogalusa Heart Study. Arteriosclerosis 1982; 2:346-355. 46. West KM. Epidemiology of Diabetes and Its Vascular Lesions. New York: Elsevier, 1978. 47. Zimmerman J, Kaufmann NA, Fainaru M, Eisenberg S, Oschry Y, Friedlander Y, Stein Y. Effect of weight loss in moderate obesity on plasma lipoprotein and apolipoprotein levels and on high density lipoprotein composition. Arferiosclerosis 1984; 4:115-123. Received March 5, 1991 Revised July 17, 1991 Accepted August 21, 1991