Assessment of atherosclerotic risk factors and endothelial function in children and young adults with pediatric-onset systemic lupus erythematosus




To characterize atherosclerotic risk factors and endothelial function in pediatric-onset systemic lupus erythematosus (SLE).


Lipoproteins, oxidized state, and autoantibodies to oxidized low-density lipoprotein (Ox-LDL) were assessed. Endothelial function was evaluated using brachial artery reactivity.


Thirty-three SLE patients and 30 controls were studied. SLE subjects had significantly decreased mean high-density lipoprotein (HDL) cholesterol (41 mg/dl versus 51 mg/dl; P = 0.002) and apolipoprotein A-I (97 mg/dl versus 199 mg/dl; P = 0.0004). There was no difference between groups in markers of oxidized state (including nitric oxide metabolites, isoprostanes, and Ox-LDL) or in endothelial function. However, SLE subjects had increased median anti-Ox-LDL IgG (2,480 relative light units [RLU] versus 1,567 RLU; P = 0.0007) and IgG immune complexes with LDL (4,222 RLU versus 2,868 RLU; P = 0.002).


Pediatric SLE patients had significantly decreased levels of HDL cholesterol and apolipoprotein A-I and elevated titers of autoantibodies to Ox-LDL. Despite these atherosclerotic risk factors, SLE patients had normal measures of oxidized state and endothelial function.


Atherosclerosis is a major cause of morbidity and mortality in adults with systemic lupus erythematosus (SLE) (1, 2). The pathogenesis of premature cardiovascular disease in SLE is likely multifactorial, related to vasculitis, corticosteroid use, renal disease, hypertension, hyperlipidemia, and thrombosis associated with antiphospholipid antibodies (1). Although complications of atherosclerosis generally are not seen until adulthood, the atherosclerotic process begins in childhood (3). Therefore, interventions directed at reducing or preventing atherosclerosis should begin in childhood.

Endothelial injury, oxidative modification of lipids, and inflammation all contribute to the development of atherosclerosis (4). Endothelial dysfunction is hypothesized to be an early event in atherogenesis, preceding the formation of plaques (5), and is predictive of cardiovascular events in adults (6). Brachial artery reactivity (BAR) is a noninvasive measure of endothelial function, measuring endothelium-dependent vascular relaxation (flow-mediated dilation [FMD]). FMD is thought to be dependent on adequate production, release, and activity of endothelial-derived relaxing factor or nitric oxide (NO), which may be impaired in atherosclerosis (7).

Endothelial cells, smooth muscle cells, and macrophages are all capable of oxidative modification of low-density lipoprotein (LDL) cholesterol. Oxidized LDL (Ox-LDL) is more susceptible to uptake by macrophages, thus resulting in foam cell formation and promoting atherogenesis (4). Ox-LDL may also contribute to atherosclerosis through direct injury to the endothelium, decreased expression of endothelial NO synthase, and inactivation of NO after it is released from endothelial cells (8, 9), thereby contributing to endothelial dysfunction. In addition to measuring levels of Ox-LDL, oxidized state can be assessed by quantifying levels of NO metabolites (10) and isoprostanes (11), markers of lipid peroxidation.

Evidence suggests that atherosclerosis is an inflammatory disease (12, 13). Inflammatory cells are key components of atheromatous plaques. Systemic inflammation, as evidenced by elevated serum C-reactive protein, is a significant predictor of future cardiovascular events (13). Ox-LDL may also play a significant role in inducing inflammation by its provision of a variety of proinflammatory oxidized lipids and by its ability to stimulate both a humoral and a cell-mediated immune response (14, 15). Furthermore, autoantibodies to Ox-LDL contribute to formation of immune complexes (IC) that can damage arterial walls and increase the uptake of Ox-LDL by macrophages (14, 15).

Although several investigators have studied atherosclerosis in adults with SLE, little work has been performed in children with this disease. Our aims were 1) to characterize atherosclerotic risk factors in children with SLE, including lipoprotein abnormalities, measures of oxidized state, and autoantibodies directed at Ox-LDL and 2) to measure FMD by BAR and evaluate for endothelial dysfunction as an early marker for atherosclerosis.


Study population.

SLE subjects meeting the revised criteria of the American College of Rheumatology (16) were recruited from the Pediatric Rheumatology Clinic at the University of California, San Francisco (UCSF). Inclusion criteria were disease onset at age <18 years and current age <22 years. Exclusion criteria were neonatal or drug-induced lupus, pregnancy, and current smoking. Patients with a history of cardiovascular disease were not excluded from the study.

The control subjects included healthy siblings of the SLE subjects, siblings of other Pediatric Rheumatology Clinic patients, and healthy children and adolescents recruited from the UCSF campus. The SLE subjects and controls were group matched for sex, age, and ethnicity. Exclusion criteria for controls were diagnosis of a chronic illness, pregnancy, current use of medications that alter lipid metabolism, and current smoking.

The UCSF Committee on Human Research approved the study. A parent and/or the subject (if >18 years of age) provided written consent. Subjects were evaluated during a single outpatient visit to UCSF after a 12-hour fast.

Atherosclerotic risk factors.

Race, history of hypertension, and use of antihypertensives was determined by questionnaire and chart abstraction. Weight, height, body mass index, pulse, and blood pressure (BP) were measured. Dietary intake of calories, fat, and cholesterol were determined by self-reported 3-day dietary records (17) that were analyzed utilizing The Food Processor Plus (version 6.0; ESHA Research Inc, Salem, OR) (18).

Lipid and lipoprotein measurements.

Plasma was isolated from venous blood by centrifugation at 4°C. Total cholesterol (TC) was determined by an automated technique using the cholesterol oxidase reaction and triglycerides (TG) were measured by the glycerokinase method (19). High-density lipoprotein (HDL) cholesterol was determined after precipitation of other lipoproteins by dextran sulfate and manganese. LDL cholesterol was calculated by the Friedewald formula (20). Plasma Lp(a) and apolipoprotein A-I (apo A-I) were quantified by enzyme-linked immunosorbent assay (ELISA) (21, 22).

SLE disease activity and severity assessments.

Information about disease onset, clinical manifestations, and medications was obtained through questionnaires and chart reviews. The following laboratory measurements were completed: a complete blood count, erythrocyte sedimentation rate (Westergren method), double-stranded DNA antibody (dsDNA; by ELISA), complement (C3, C4, CH50; by rate nephelometry), urinalysis, spot urine protein, and creatinine. The SLE Disease Activity Index (SLEDAI) was calculated (16).

Measures of oxidized state.

Nitric oxide. Plasma was stored at −70°C. In solution, NO reacts with molecular oxygen to form nitrite, and with oxyhemoglobin and superoxide anion to form nitrate. The nitrite and nitrate were reduced using vanadium and hydrochloric acid at 90°C. NO was then purged from solution resulting in a peak of NO, representing total NO, nitrite, and nitrate. This peak was then detected by chemiluminescence (NOA 280; Sievers Instruments Inc, Boulder, CO). The detection limit is 1 nM/ml of nitrate.


Blood was collected into an EDTA tube containing butylated hydroxytoluene (2.25 μmol) and indomethacin (10 μmol) and immediately placed on ice. Plasma was stored at −80°C and levels of F2-isoprostanes were quantified employing stable isotope dilution methodology utilizing gas chromatography/mass spectrometry with negative ion chemical ionization (23).

Determination of Ox-LDL epitopes.

The EO6 epitope concentration on apo B-100–containing particles was measured as previously described (24, 25) using a chemiluminescent modification that has been described in detail elsewhere (26). Data are expressed as EO6/MB24 multiplied by 100. All samples were measured in a single assay, with intraassay coefficients of variation of low and high standards of 6–8%.

Measures of autoantibodies to Ox-LDL.

Autoantibody binding to Ox-LDL.

LDL was isolated from pooled plasma of healthy donors by sequential preparative ultracentrifugation under conditions to minimize oxidation and proteolysis and subsequently oxidized by copper or modified by malondialdehyde (MDA) as described previously (27). The chemiluminescent assay was performed with modifications as described previously (24). Data are expressed as relative light units per millisecond (RLU/msec). Each determination was done in triplicate and all samples were measured in a single assay with coefficients of variation for low and high standards of 6–8%.

Immune complexes with LDL.

Apo B-IC containing IgG or IgM were determined as described previously (25).

Brachial artery reactivity studies.

All BAR studies were performed on the right arm by a single investigator using high-resolution 2-dimensional ultrasound with a 15-MHz linear array vascular transducer and a Sequoia C256 system (Acuson, Mountain View, CA) using a previously described protocol (28). Following basal scans, hyperemia was induced by occluding the artery with a pneumatic tourniquet around the right forearm for 5 minutes. To determine endothelium-dependent dilation (FMD), recordings were taken continuously from 30 seconds prior to and for 120 seconds following tourniquet release. To measure endothelium-independent (nitroglycerin [NTG]-induced) dilation, a final scan was recorded 2–3 minutes following the administration of sublingual NTG (300 μg). Scans were recorded digitally and on Super VHS tape for off-line analyses using the Brachial Ultrasound Workstation (Medical Imaging Applications, Iowa City, IA). FMD was expressed as the peak change in arterial diameter from baseline within 2 minutes of hyperemia.

Statistical analysis.

STATA (version 6.0; Stata Corporation, College Station, TX) was used to perform all statistical analyses. To test for differences between study groups, 2-sample, unpaired t-tests were performed for normally distributed variables and Wilcoxon's rank sum test was used for nonparametric variables. Spearman correlations were used to describe the relationships between variables. P values < 0.05 were considered statistically significant. The study was designed to have >80% power to detect a clinically significant difference in TC and endothelium-dependent dilation with enrollment of 30 subjects per group.


Thirty-three SLE subjects and 30 controls were evaluated. The general characteristics of the subjects are summarized in Table 1. Subjects ranged in age from 8 to 21 years. SLE patients and controls were well matched for age, ethnicity, and sex and had similar weight, body mass index, and systolic BP. There was a small, statistically significant elevation in diastolic BP among the SLE subjects compared with controls. There was no difference between groups for intake of fat, saturated fat, or cholesterol, with both groups reporting diets that approximated an American Heart Association recommended diet (29).

Table 1. General characteristics of cohort*
 SLE n = 33Controls n = 30P
  • *

    SLE = systemic lupus erythematosus; BP = blood pressure.

Age, mean ± SD (range), years16.1 ± 4 (9–21)16.5 ± 4 (8–21)0.65
Female, no. (%)30 (91)24 (80)0.29
Ethnicity, no. (%)  0.99
 White7 (21)12 (40) 
 African American6 (18)3 (10) 
 Hispanic6 (18)4 (13) 
 Asian11 (34)9 (30) 
 Multiethnic3 (9)2 (7) 
Weight, mean ± SD, kg59.7 ± 2261.7 ± 190.70
Body mass index, mean ± SD, kg/m224.7 ± 723.6 ± 50.47
Systolic BP, mean ± SD, mm Hg115 ± 13112 ± 90.26
Diastolic BP, mean ± SD, mm Hg65 ± 1260 ± 80.05
Calories from fat, mean ± SD, % total calories33 ± 930 ± 80.25
Calories from saturated fat, mean ± SD, % total calories11 ± 410 ± 40.39
Cholesterol intake, mean ± SD, mg259 ± 151240 ± 1700.74


SLE subjects had significantly lower HDL cholesterol and apo A-I levels. However, there was no significant difference between groups for TC, LDL cholesterol, TG, and Lp(a) levels (Table 2). There was no correlation between steroid dosage and lipid measurements. Decreased HDL cholesterol was associated with increased markers of disease activity, including SLEDAI (r = −0.46, P = 0.01) and antibodies to dsDNA (r = −0.43, P = 0.009).

Table 2. Lipid and BAR results*
 SLE n = 33Controls n = 30P
  • *

    Data presented as mean ± SD. BAR = brachial artery reactivity; SLE = systemic lupus erythematosus; HDL = high-density lipoprotein; LDL = low-density lipoprotein.

Total cholesterol, mg/dl170.4 ± 40166.9 ± 260.69
Triglyceride, mg/dl132.9 ± 9594.1 ± 510.07
HDL, mg/dl40.6 ± 1350.7 ± 120.002
Apolipoprotein A-I, mg/dl97 ± 24118.7 ± 210.0004
LDL, mg/dl103.2 ± 3297.4 ± 190.39
Lipoprotein(a), nM/liter64.3 ± 9072 ± 960.79
Endothelium-dependent dilation, % dilation9.32 ± 3.99.2 ± 3.40.9
Endothelium-independent dilation, % dilation18.03 ± 5.521.1 ± 80.11

Brachial artery reactivity.

Measurements of endothelium-dependent and -independent dilation did not differ between the 2 groups (Table 2). There was no correlation between the BAR results and use of prednisone, hydroxychloroquine, folic acid, calcium channel blockers, angiotensin-converting enzyme inhibitors, or aspirin.

Oxidized state and autoantibodies to Ox-LDL.

Measurements of oxidized state in SLE subjects, including NO metabolites, isoprostanes, and Ox-LDL (EO6), were not significantly different from controls (Table 3). SLE subjects had significantly increased levels of anti-Ox-LDL IgG antibodies (P = 0.0007; Figure 1) and anti-MDA-LDL IgG antibodies (P = 0.0003). In addition, IgG IC with LDL were also significantly increased in SLE subjects (P = 0.002; Figure 2). When abnormal values are defined as >95th percentile for the controls, 30% (9 of 30) of our SLE subjects had elevated anti-Ox-LDL IgG antibodies and 23% (7 of 30) had elevated levels of IgG IC with LDL. There was no difference in anti-Ox-LDL and anti-MDA-LDL IgM antibodies between the 2 groups. Interestingly, the controls had significantly increased IgM IC with LDL compared with the SLE subjects (P = 0.017; Table 3). Antibody titers to dsDNA correlated with anti-Ox-LDL IgG antibodies (r = 0.68, P < 0.0001) and IgG IC with LDL (r = 0.63, P = 0.0001). Among SLE subjects, NO metabolites correlated with markers of disease activity, including ESR (r = 0.45, P = 0.001) and CH50 (r = −0.43, P = 0.03).

Table 3. Measures of oxidized state and autoantibodies to Ox-LDL*
 SLE n = 33Controls n = 30P
  • *

    Data presented as median (range). SLE = systemic lupus erythematosus; Ox-LDL = oxidized low-density lipoprotein; SLE = systemic lupus erythematosus; NO = nitric oxide; apoB = apolipoprotein B-100; RLU = relative light units; MDA = malondialdehyde; IC = immune complexes.

NO metabolites, μM3.9 (2.1–19.4)3.9 (2.2–9.4)0.38
Isoprostanes, ng/ml0.06 (0.03–0.14)0.07 (0.03–0.1)0.19
Ox-LDL (EO6)/apoB × 1005.6 (0.6–107)4.8 (0–68)0.31
Anti-Ox-LDL IgG antibodies, RLU2,480 (912–20,308)1,567 (445–5,345)0.0007
Anti-Ox-LDL IgM antibodies, RLU8,175 (3,547–22,172)7,106 (2,599–22,397)0.42
Anti-MDA-LDL IgG antibodies, RLU8,985 (2,071–80,723)4,028 (868–13,732)0.0003
Anti-MDA-LDL IgM antibodies, RLU22,159 (12,333–54,125)21,181 (11,393–67,533)0.63
IgG IC with LDL, RLU4,222 (897–14,213)2,868 (1,052–8,465)0.002
IgM IC with LDL, RLU5,212 (2,434–14,219)6,397 (3,090–12,580)0.017
Figure 1.

Anti-oxidized low-density lipoprotein IgG antibodies. Horizontal lines represent median values. SLE = systemic lupus erythematosus; Ox-LDL = oxidized low-density lipoprotein; RLU = relative light units.

Figure 2.

IgG immune complexes with low-density lipoprotein (LDL). Horizontal lines represent median values. SLE = systemic lupus erythematosus; RLU = relative light units; IC = immune complexes.

SLE characteristics.

The disease-specific characteristics and medications of the SLE patients are summarized in Table 4. Age at diagnosis ranged from 5 to 18 years (mean 12.8) and disease duration ranged from 0.05 to 9.6 years (mean 3.1). Surprisingly, increased disease duration was associated with a decrease in a number of the atherosclerosis risk factors measured. Specifically, disease duration correlated with increased HDL (r = 0.45, P = 0.009), decreased anti-Ox-LDL IgG antibodies (r = −0.42, P = 0.023), and decreased anti-MDA-LDL IgG antibodies (r = −0.41, P = 0.026). Nine SLE subjects had a history of hypertension but none had a history of myocardial infarction (MI) or angina. More than one-third of patients had received cyclophosphamide during the course of their disease, 82% were receiving prednisone at the time of the study, and many were being treated with potentially cardioprotective drugs, including hydroxychloroquine, folic acid, and aspirin (30–32).

Table 4. SLE patient characteristics and medication exposures (n = 33)*
  • *

    SLE = systemic lupus erythematosus; SLEDAI = systemic lupus erythematosus disease activity index.

  • Normal ranges: sedimentation rate 0–10 mm/hour men, 0–15 mm/hour women; double-stranded DNA <30 IU/ml; C3 71–159 mg/dl.

  • Past or present use of medication.

  • §

    Current use of medication.

Age at diagnosis, mean ± SD (range), years12.8 ± 3 (5–18)
Disease duration, mean ± SD (range), years3.1 ± 3 (0.05–9.6)
SLEDAI8 ± 8 (0–34)
Sedimentation rate, mean ± SD (range), mm/hour30 ± 27 (1–100)
Double-stranded DNA, mean ± SD (range), IU/ml373 ± 761 (3–3,142)
C3, mean ± SD (range), mg/dl92.5 ± 32 (31–180)
Urine protein/creatinine ratio, mean ± SD (range)0.54 ± 1 (0.005–3.6)
History of hypertension, no. (%)9 (30)
Medication exposure, no. (%) 
 Cyclophosphamide11 (33)
  Current use27 (82)
  Dosage, mean ± SD (range), mg/kg/day0.2 ± 0.2 (0–0.73)
 Hydroxychloroquine§25 (76)
 Folic acid§4 (12)
 Calcium channel blocker§3 (9)
 Angiotensin-converting enzyme inhibitor§5 (15)
 Aspirin§13 (39)


SLE is complicated by premature atherosclerosis, with rates of MI ranging from 9 to 50 times greater than expected (1, 2). A number of risk factors for atherosclerosis have been described in adults with SLE, including a dyslipoproteinemia characterized by elevated TG and very low-density lipoprotein and decreased HDL levels (33). In addition, there is evidence for elevated serum Lp(a) (34), decreased apo A-I, and increased antibodies to apo A-I (35) and to Ox-LDL (26, 36) in adults with SLE. Several studies have also demonstrated increased oxidized state, as measured by nitrates and nitrites (37), urinary isoprostanes (38), and Ox-LDL (36). Studies in adults with SLE have detected subclinical cardiovascular disease utilizing carotid ultrasound, myocardial single photon emission computed tomography, and BAR (39–41).

There is a paucity of information about atherosclerosis in pediatric patients with SLE. A dyslipoproteinemia has been described in children with SLE that is similar to that found in adults (42). There have been only 2 previous studies characterizing asymptomatic atherosclerosis in children with SLE. Utilizing thallium myocardial perfusion scans, radionuclide angiography, and echocardiography, one study was able to detect asymptomatic abnormalities of myocardial perfusion in 5 of 31 children, aged 10–19 years (43). Another study detected significantly increased intima-media wall thickness of the carotid artery in children and young adults (aged 6–25 years) with SLE (mean 0.57 mm versus 0.54 mm; P = 0.006) (44).

Increased titers of anti-Ox-LDL antibodies are found in plasma of patients with cardiovascular disease. These antibodies are also found in atherosclerotic lesions, in part, as immune complexes with Ox-LDL (45). In murine and rabbit models of atherosclerosis, such anti-Ox-LDL antibody titers correlate with the progression and regression of atherosclerosis (46, 47). In human studies, they are predictive of MI and progression of carotid atherosclerosis (48). Among adult women with SLE, levels of Ox-LDL and autoantibodies to Ox-LDL discriminated those subjects with a history of coronary artery disease, suggesting that these indices are a marker of underlying atherosclerosis and that oxidation of LDL is possibly contributing to this process (26).

In adults, BAR correlates with coronary artery reactivity (49) and with angiographically diagnosed coronary artery disease (50), and is abnormal in the setting of many atherosclerotic risk factors in children and adults (28, 51, 52). Endothelial dysfunction is hypothesized to be an early event in atherogenesis, preceding the formation of plaques (5), and may serve as a marker for increased risk of early atherogenesis (28).

The SLE subjects included in this study had an average disease duration of 3.1 years (range 0.05–9.6 years) and a mean SLEDAI of 8 (range 0–34). Although many of them had historic risk factors for atherosclerosis, including hypertension and renal disease, none had a history of MI. Similar to previous studies, pediatric SLE subjects in the current study demonstrated decreased levels of HDL cholesterol and apo A-I and increased anti-Ox-LDL IgG antibodies. However, we did not detect significant abnormalities of oxidized state measurements, including NO metabolites, isoprostanes, and Ox-LDL.

Despite risk factors for atherosclerosis, including abnormal lipoprotein profiles and anti-Ox-LDL antibodies, our pediatric SLE subjects demonstrated normal endothelial function. This study had >80% power to detect a 5% difference in endothelium-dependent dilation between SLE subjects and controls. The normal endothelial function could reflect the fact that our patients are young and may still have resilient endothelium. Endothelial dysfunction has been reported in children of similar age who have familial hypercholesterolemia (28, 53), but generally these children have had longer disease duration than our SLE patients. It is also possible that our BAR technique was not sensitive enough to detect the earliest changes in endothelial function. Although we were not able to detect a significant influence of medications on BAR measurements, it is of note that many of our patients were taking aspirin and/or folic acid, both of which have vasculoprotective properties (31, 32). It is also possible that unidentified factors are present in our subjects that are attenuating the oxidative stress and protecting the endothelium.

One significant limitation to this study is that we cannot exclude the possibility that increased autoantibodies to oxidized lipoproteins reflect nonspecific B-cell stimulation associated with SLE. However, in an adult population, the anti-Ox-LDL antibody titers were increased in SLE patients with cardiovascular disease compared with age- and disease-matched SLE patients who did not have clinical evidence of cardiovascular disease (26). Thus, it is possible that the increased anti-Ox-LDL antibodies seen in our pediatric SLE subjects may be a very early marker of subclinical atherosclerosis. In addition, although the levels of anti-Ox-LDL IgM were similar to controls, it is interesting that we detected significantly decreased IgM IC with LDL in our SLE subjects. Studies in both healthy and diabetic adults have described an inverse relationship between carotid atherosclerosis and IgM titers against Ox-LDL and MDA-LDL, suggesting that these IgM antibodies may be protective against early atherosclerosis (54, 55).

In conclusion, although our cohort of pediatric SLE patients demonstrated normal measures of oxidized stress and endothelial function, they have dyslipidemia and enhanced titers of IgG autoantibodies to Ox-LDL that are associated with an increased risk of atherosclerosis. These patients need to be followed prospectively to determine if these risk factors predict future development of endothelial dysfunction and atherosclerosis.


The authors thank the following for their help with this project: Dr. Jason D. Morrow, Dr. Jeffrey R. Fineman, Dr. Boaz Ovadia, Elizabeth Miller, Dorthy Lee, Sharon Cohen, Cari DeLoa, David Glidden, Dr. Marguerite Engler, and the UCSF EARLY trial.