Association of Apolipoprotein A Phenotypes and Oxidized Low-Density Lipoprotein Immune Complexes in Children

Background  Small apolipoprotein A (apo[A]) phenotypes and oxidized low-density lipoprotein immune complexes (oxLDL-ICs) are known to be associated with the development of atherosclerosis in adults. Presence of these factors in children and their relationships with other known cardiovascular risk factors have not been well documented.

Objective  To examine the relationship of oxLDL-ICs with apo(A) phenotypes and other known cardiovascular risk factors in children.

Design  A survey of asymptomatic 9- to 11-year-old children, randomly selected from a cohort of children stratified based on family history of premature coronary artery disease.

Setting  A preventive medicine research institute.

Participants  Thirty-five children with or without a family history of premature coronary artery disease who are participating in a longitudinal cardiovascular health study.

Main Outcome Measures  The influence of apo(A) phenotypes on plasma levels of oxLDL-ICs after controlling for lipid/lipoprotein levels, percentage of body fat, and physical fitness.

Results  Oxidized low-density lipoprotein immune complexes were significantly correlated with the levels of total cholesterol (r = 0.56, P≤.05), low-density lipoprotein cholesterol (r = 0.64, P≤.01), and low-density lipoprotein cholesterol/high-density lipoprotein cholesterol (r = 0.54, P<.05). Oxidized low-density lipoprotein immune complexes were also correlated with total cholesterol high-density lipoprotein cholesterol (r = 0.49, P≤.06) and percentage of body fat (r = 0.48, P≤.06). However, they achieved only a borderline level of statistical significance after adjustment for multiple comparisons. Multiple regression analysis demonstrated that small apo(A) phenotypes, levels of low-density lipoprotein cholesterol, and family history of premature coronary artery disease explained 54% of the variation of oxLDL-ICs using a parsimonious model (P=.001).

Conclusions  Significant correlations exist between oxLDL-ICs and known cardiovascular risk factors in children. The association of oxLDL-ICs with the genetically controlled small apo(A) phenotype suggests that the genetic predisposition to immune complex formation may be an important determinant of future coronary artery disease.

BOTH APOLIPOPROTEIN A (apo[A]) small phenotypes and higher levels of lipoprotein(a) (Lp[a]), a particle formed by bonding of apo(A) glycoprotein and apolipoprotein B-100, have been reported to be independent risk factors for premature coronary artery disease (CAD).^1-8 Higher levels of oxidized low-density lipoprotein antibodies (oxLDL-Ab) and oxLDL–immune complexes (oxLDL-ICs) also have been reported to be associated with premature atherosclerosis.^9-12 To date, most of the studies published have failed to demonstrate correlations of apo(A) phenotypes or Lp(a) levels with common cardiovascular risk factors such as obesity, hypertension, diabetes mellitus, high cholesterol levels, smoking, and sedentary lifestyle.^13-17 Most of the studies, however, supported the role of the apo(A) gene in explaining the variability of Lp(a)^18,19 or apo(A) phenotypes.^1-3 In contrast to Lp(a) levels, various environmental factors contribute to the oxidative modification of low-density lipoproteins (LDL),^20-23 and currently there is no compelling evidence to support the role of any genetic influence on LDL oxidation. However, if apo(A) phenotypes or the levels of Lp(a) influence the levels of oxLDL-ICs and oxLDL-Ab, then it will not only establish the role of the apo(A) gene on the formation of oxLDL-ICs or oxLDL-Ab, but it will also help in explaining the apo(A)- or Lp(a)-initiated pathophysiologic mechanisms in the development of atherosclerosis.

During the oxidative modification of LDL, lipid and protein oxidation products are generated that induce LDL modification, regardless of the environmental exposures. The modified LDL is not only chemotactic for monocytes, but also triggers the production of autoantibodies.⁹ Subsequently, these autoantibodies will associate with the antigen and lead to the formation of ICs. An example of immune complex formation would be formation of antiphospholipid antibody-LDL-ICs in the serum of patients with systemic lupus erythematosus.²⁴ These ICs are thought to contribute to the pathogenesis of early atherosclerosis in patients with systemic lupus erythematosus.²⁴ Similarly, LDL-ICs associated with modified LDL have been found to contribute to the development of atherosclerosis.²⁵ Several mechanisms can contribute to the atherogenic potential of oxLDL-IC.

Accumulation of cholesterol esters in human macrophages exposed to oxLDL-IC is markedly increased and results from a combination of increased uptake and altered intracellular metabolism of the LDL taken up as part of LDL-ICs.²⁵ The amount of LDL taken up as part of LDL-ICs by human macrophages at the end of 5 hours is 7-fold higher than that of the native form. In addition, the degradation of LDLs complexed with IgG is impaired, being only 1.09-fold higher than that of native LDL, which leads to a 200-fold increase in the amount of LDL accumulation intracellularly.²⁵ Furthermore, the ICs associated with LDL activate macrophages that release active oxygen radicals. These free radicals promote further oxidation of LDL in the subendothelial space, creating a self-perpetuating cycle of events responsible for the progression of atherosclerosis.^26,27 The LDL-ICs have the ability to promote overexpression of LDL receptors in macrophages, leading to a persistent influx of LDL into a cell that has an impaired ability to process accumulated cholesterol. Finally, the LDL-ICs stimulate the release of various cytokines, which may lead to endothelial damage and smooth muscle cell proliferation.²⁸

Since atherosclerosis begins in childhood, it is reasonable to expect that some of the markers of atherosclerosis may be detectable among healthy children who are at high risk of future premature CAD. The influence of specific genes on the levels of oxLDL-ICs or oxLDL-Ab is unknown. To our knowledge, no study to date has evaluated whether oxLDL-ICs are present in children and whether genetic markers such as apo(A) phenotypes or the levels of Lp(a) influence the levels of oxLDL-ICs or oxLDL-Ab in children. Therefore, in this study, we have evaluated the influence of apo(A) phenotypes or Lp(a) levels in predicting the levels of oxLDL-ICs and oxLDL-Ab. In addition, we have evaluated the relationships of oxLDL-ICs and oxLDL-Ab with various known cardiovascular risk factors.

Thirty-five children, 9 to 11 years old, were randomly selected from a cohort of children who were participating in a longitudinal study of the biobehavioral antecedents of CAD.²⁹ Information on ethnicity was provided by the parents, and information on family history (FamHx) of premature CAD was collected through a short questionnaire and was verified through physicians' reports. There were 16 children (10 white, 6 black) with a positive family history (FamHx+) for premature CAD, and 19 children (13 white, 6 black) without such family history (FamHx−) at the time of data collection. Positive family history was defined as occurrence of myocardial infarction and/or CAD (angioplasty, coronary bypass surgery, angina, or receiving medication for angina) among any parents or grandparents before the age of 55 years. All children and their parents signed informed consent forms in accord with the procedures of the Medical College of Georgia, Augusta, Human Assurance Committee. In addition to demographic information such as age, sex, race, and FamHx, all children were assessed for anthropometric factors that could be related to lipids and lipoproteins, fasting levels of plasma total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), triglycerides, body-weight adjusted physical fitness, Lp(a), apo(A) phenotypes, oxLDL-Ab, and oxLDL-ICs.

Children came to the Medical College of Georgia, Augusta, phlebotomy laboratory after fasting for at least 12 hours. Venous blood was collected by a pediatric phlebotomist by venipuncture into vacuum-sealed tubes containing EDTA. Except for apo(A) phenotypes, oxLDL-Ab, and oxLDL-ICs, all analyses were performed in the lipid analytic laboratory of the Department of Pediatrics, Medical College of Georgia, which has been accredited by the College of American Pathologists and participates in the Centers for Disease Control and Prevention lipid standardization program. Total cholesterol was determined enzymatically in a coupled, spectrophotometric assay using the chromogens 4-aminoantipyrine and p-hydroxybenzenesulfonate. The intensity at 500 nm of the quinoneimine dye was measured. Only 40 µL of plasma was required for this measurement performed in triplicate.^30,31 High-density lipoprotein cholesterol was measured after using dextran sulfate/magnesium as the precipitating agent for very low-density lipoprotein and LDL.³² Triglycerides were analyzed by enzymatic hydrolysis to glycerol and free fatty acids, conversion of the glycerol to glycerol-1-phosphate, and measurement of nicotinamide adenine dinucleotide (indirectly as 500-nm absorbance of the formazan dye) produced in the subsequent formation of dihydroyacetone phosphate.³³ This assay required 40 µL of plasma for triplicate determinations. Low-density lipoprotein cholesterol was calculated using the Friedewald equation,³³ as all of the samples measured had triglyceride levels less than 5.16 mmol/L (<400 mg/dL).

The levels of lipoprotein Lp(a) were measured with an enzyme-linked immunosorbent assay kit (Macro Lp, Terumo Corporation Diagnostic Division, Somerville, NJ) in the Clinical Nutrition Laboratory at the Medical College of Georgia. This kit included standards (6 levels) containing Lp(a) in human plasma in buffered solution that were used to plot a standard curve from which controls and unknowns were calculated. An external source of Lp(a) controls (2 levels) assayed by Northwest Lipid Research Center in Seattle, Wash, was also included in the controls. All Lp(a) analyses were done in batches within 1 month and all standards, controls, and plasma samples were run in duplicate. The reproducibility coefficient of a subsample (n = 19) was 0.93.

Frozen plasma samples were sent to the Department of Human Genetics at the School of Public Health, University of Pittsburgh (Dr Kamboh), Pittsburgh, Pa, by overnight mail in plastic foam boxes cooled with dry ice. The apo(A) phenotypes were determined using a high-resolution sodium dodecyl sulfate–agarose gel electrophoresis method that resolved at least 30 apo(A) phenotypes.³⁴ Ten to 15 µL of plasma was mixed with 30 µL of reducing buffer (1:2:10 ratio of β mercaptoethanol-0.5% bromophenol blue in 5% glycerol–5% sodium dodecyl sulfate) and the mixture was heated for 5 minutes at 100°C. Electrophoresis was performed on 1.5% agarose submarine gels (90-mmol/L Tris, 90-mmol/L boric acid, 2-mmol/L EDTA, and 0.1% sodium dodecyl sulfate) in the Hoefer submarine gel unit with power supply. Electrophoresis was carried out in a tank buffer containing 45-mmol/L Tris, 45-mmol/L boric acid, 2-mmol/L EDTA, and 0.1% sodium dodecyl sulfate for 7 to 8 hours at a constant 25 W. Each gel contained a mixture of 5 known apo(A) phenotypes as internal controls. After electrophoresis, proteins were transferred to a 0.45-µm nitrocellulose membrane by electroblotting overnight using a Hoefer Transphor cell at 90 V in 10-mmol/L Tris, 40-mmol/L glycine, and 5% methanol. After protein transfer, the membrane was incubated with 5% powdered skim milk for 1 hour followed by incubation overnight with rabbit antihuman apo(A) antiserum,³⁵ and finally with goat antirabbit IgG conjugated with alkaline phosphatase for 3 hours. Subsequently, apo(A) bands were visualized by histochemical staining. For reliability, each sample was run twice.

Frozen plasma samples were sent to the Department of Medicine, Medical University of South Carolina, Charleston, for measurement of autoantibodies against oxLDL. Details of the laboratory methods have been published previously.²⁵ Briefly, a solid-phase enzyme immunoassay for antioxidized LDL was carried out. Flat-bottomed plates (Immulon type 1 plates) were coated with oxLDL in 0.25-mmol/L sodium carbonate buffer at pH 9.6 (100 µL per well of a 5-µg/mL oxLDL preparation) and were incubated overnight at 4°C. The unbound oxLDL was washed off and plates were blocked with 5% bovine serum albumin in phosphate-buffered saline pH (200-µL well, 45-60 minute incubation at 37°C). After blocking, the plates were washed 3 times in phosphate-buffered saline with 0.05% polysorbate (Tween 20, Sigma Chemical, St Louis, Mo). Serum samples were run both absorbed with oxLDL (100 µg/mL) and unabsorbed.

Two milliliters of serum collected from each participant was mixed with 2 mL of a 7% solution of polyethylene glycol 6000 wt/vol prepared in borate buffered saline, pH 8.4. The sample containing a final polyethylene glycol concentration of 3.5% was incubated overnight at 4°C and then centrifuged at 3000 rpm for 20 minutes. The precipitates were washed twice with 10 mL of chilled 3.5% polyethylene glycol solution, centrifuged again, and resuspended in 1 mL of phosphate-buffered saline.

The cholesterol content of these isolated ICs was determined by gas chromatography after a Folch lipid extraction.³⁶ To ensure that only LDL that are complexed with antibodies were precipitated by polyethylene glycol, native LDL (1 mg) was added to the serum, and the precipitation of ICs was performed in the samples with or without added LDL. No differences between the amount of cholesterol contained in the ICs were observed between the samples with or without added LDL. Further details about the performance and specificity of this assay have been reported previously.³⁷

Body fat was measured with dual energy x-ray absorptiometry (QDR 2000, Hologic Inc, Waltham, Wash). This technique has been shown to be reliable and valid for determining the percentage of body fat in children.³⁸

Physical fitness was determined by measuring heart rate (HR) response to supine cycling; a lower HR denoted a higher level of fitness.³⁹ To integrate the HR values from 3 workloads, the regression of HR on power output was calculated and fitness was expressed as the power output at an HR of 150 beats per minute with greater output signifying greater physical fitness. To account for the effects of maturity and size on HR, the power output was regressed on body weight and the residuals around the regression line were determined. These residuals represent individual differences in physical fitness, adjusted for body weight.

Initial descriptive analyses for oxLDL-Ab and oxLDL-ICs were computed by race, sex, and FamHx. For variables with standardized skew and kurtosis exceeding the ±2 range, nonparametric analyses were used. The classification of apo(A) phenotypes was based on apparent molecular weight and was categorized as small (mean molecular weight, <454.9 kd) and others (large or medium) (molecular weight, ≥454.9 kd). Spearman rank correlation coefficient was used to examine the correlations between various anthropometric, lipid, and lipoprotein variables. A variable selection method known as backwards elimination using multiple linear regression analysis was carried out to evaluate the degree to which fatness, fitness, Lp(a) levels, or apo(A) phenotypes and lipid and lipoprotein levels explained the variations in levels of oxLDL-Ab or oxLDL-ICs and to assess the individual importance of certain predictors in the presence of other predictor variables. The method is directed toward the development of a relatively simple (parsimonious) model, which predicts the response variable.⁴⁰

The descriptive sample characteristics are given in Table 1. Data on oxLDL-Ab and oxLDL-ICs were available for all 35 children selected initially. Data on Lp(a) and apo(A) phenotypes were available for 33 and 31 of the children, respectively. Due to inadequate processing of some part of blood samples or unavailability of the children for some measurements, some information is missing for a few children. However, the overall results did not change when data were reanalyzed, excluding the children with missing data. There were 19 boys (54%) and 16 girls (46%), 23 white subjects (66%) and 12 black subjects (34%). Sixteen children (46%) had a FamHx+ for premature CAD and 19 (54%) had a FamHx− for premature CAD. The mean levels of antibodies as measured from absolute differences in absorption seem to be within the normal range observed in adults. Table 2 gives Spearman rank correlations between fitness, fatness, insulin, lipoprotein-related variables, and oxLDL-Ab or oxLDL-ICs. The correlational analysis revealed an interesting contrast between oxLDL-Ab and oxLDL-ICs. Oxidized LDL antibodies were negatively correlated with percentage of body fat (r = −0.39), TC (r = −0.32), and TC/HDL-C ratio (r = −0.31). In contrast, oxLDL-ICs were positively correlated with TC (r = 0.56), LDL-C (r = 0.64), LDL-C/HDL-C ratio (r = 0.54), TC/HDL-C ratio (r = 0.49), and percentage of body fat (r = 0.48). Correlations between oxLDL-ICs and TC, LDL-C, or LDL-C/HDL-C were all significant (P<.05). Because of significant collinearity between various lipid or lipoprotein variables, we carried out linear regression analyses with 1 lipid or lipoprotein variable at a time along with other predictor variables. Table 3 gives the result of "backwards elimination" regression search for a parsimonious model. Sex and percentage of body fat accounted for 26% of the variation in oxLDL-Ab, while LDL-C, apo(A) phenotype categories (small vs others), and FamHx explained 54% of the variation in oxLDL-ICs levels. Race, sex, physical fitness, and other lipids/lipoprotein variables did not contribute to the models in the presence of the factors included in the best-fitted model. However, these analyses were made on observational data and the conclusions drawn must be tentative until confirmed by subsequent studies.

Ample evidence exists to suggest that the atherosclerotic process begins during childhood.^41,42 The progression of lesions varies,⁴³ and it is thought that this variability relates to genetic susceptibility that selectively modifies the effect of environmental exposures on disease risk. Small apo(A) phenotypes and high Lp(a) levels have been associated with premature coronary atherosclerosis in adults. We previously reported that small apo(A) phenotypes are associated with FamHx of premature CAD in children.⁴⁴

The importance of oxLDL in the pathogenesis of atherosclerosis is well established in adults. However, little is known about their presence in children. If a genetically determined factor like high Lp(a) levels or small size apo(A) phenotypes influences the formation of oxLDL-ICs in children, then it would explain one of the mechanisms of accelerated atherosclerosis associated with high Lp(a) levels or small size apo(A) phenotypes observed in adults. Furthermore, one can speculate that children or adults with both higher levels of Lp(a) and oxLDL-ICs will be at high risk of developing premature coronary artery disease in the future. To our knowledge, this is the first study to evaluate the relationship of oxLDL-Ab and oxLDL-ICs with known cardiovascular risk factors in children, including a well-established genetic factor such as Lp(a) or apo(A) phenotypes. The small apo(A) phenotypes, rather than Lp(a) levels (0.3 g/L [30 mg/dL]), were a significant predictor of variability of oxLDL-ICs. In a previous report, we observed that pathogenic apo(A) phenotypes are a better predictor of FamHx of premature CAD than absolute levels of Lp(a).⁴⁴ This study found either negative correlation or no relationship between oxLDL-Ab and known cardiovascular risk factors. In contrast, soluble LDL-ICs were significantly correlated with known cardiovascular risk factors. In addition, oxLDL-Abs were significantly negatively correlated with soluble LDL-ICs. Such contrasting correlations can be explained by the fact that oxLDL-ICs are generally found in subjects with high levels of oxLDL. As oxLDL-ICs are formed, the binding sites of oxLDL-Ab become saturated and free anti–oxLDL-Abs are not detected in the serum. In addition, the free antibodies are generally a reflection of activation of humoral or cellular immune response. Apart from the antigen burden, humoral immune response may depend on genetic background, rate of antibody production, and catabolism. However, there are no compelling reasons to suggest that there is significant variability in the rate of antibody production and catabolism in the children studied. The most likely explanation of the correlations observed is that soluble ICs are a marker for antigen excess in the subendothelial space and that increased formation of ICs in the vascular wall containing oxLDL and autoantibodies against oxLDL leads to a decrease in the level of free antibodies in the serum.

This study demonstrates that oxLDL-ICs may be an important marker of underlying atherosclerotic processes that may be influenced by genetic predispositions. The presence of these markers in children not only supports the idea that the atherosclerotic process begins early in life, it also indicates the possibility of identifying a genetically susceptible population during childhood. A potential benefit of identifying such children and young adults will be the development of prevention trials targeting oxLDL and oxLDL-ICs. The feasibility of such interventions is supported by the following facts: Overwhelming evidence exists that suggests that oxLDLs result from various environmental exposures.^20-24,45 In humans, it has been shown that dietary oxidized lipids absorbed by the small intestine are incorporated in chylomicrons and appear in the bloodstream, where they contribute to the total body pool of oxidized lipid.⁴⁶ Oxidized LDLs have been shown to be influenced by the presence of antioxidants and free radicals. Belcher et al⁴⁷ has shown that brief oral supplementation of vitamin E reduced the oxLDL. Another recent study also reported the effect of vitamin E supplementation on oxLDL.⁴⁸ The role of smoking in oxLDL has been documented previously.^20-23 Few studies indicated an effect of physical activity on the chemical composition of LDL,^49,50 and it is unknown whether physical activity influences oxLDL-Ab or oxLDL-ICs. Therefore, screening children and young adults for abnormal levels of oxLDL-ICs may set the stage for prevention trials using vitamin E supplementation.

Although we have identified a potential genetic influence on the levels of oxLDL-ICs in children, further studies are needed to evaluate the effect of the presence of both higher levels of oxLDL-ICs and small apo(A) phenotypes in predicting early atherosclerosis.

Accepted for publication July 13, 1998.

Supported by grants HL 35073 and HL 44672 from the National Heart Lung and Blood Institute, National Institutes of Health, Bethesda, Md; the Medical College of Georgia Research Institute, Augusta; and the American Heart Association, Georgia Affiliate, Atlanta.

We thank Gabriel Virella, MD, PhD, Department of Immunology-Microbiology, Medical University of South Carolina, Charleston, for his advice and support in the measurement of low-density lipoprotein immune complexes. We also thank Arshadul Haque, MBBS, West Virginia University, Morgantown, for organizing the data and helping in the data analysis.

Corresponding author: Syed Islam, MBBS, DrPH, Institute of Occupational and Environmental Health, Box 9190, West Virginia University School of Medicine, Morgantown, WV 26506.

Editor's Note: It's important to be aware of factors present in childhood that correlate with major illness in adulthood. However, the more important question is: "What are we going to do about it?"—Catherine D. DeAngelis, MD