Dysfunctional proinflammatory high-density lipoproteins confer increased risk of atherosclerosis in women with systemic lupus erythematosus

Authors


Abstract

Objective

Women with systemic lupus erythematosus (SLE) have an increased risk of atherosclerosis. Identification of at-risk patients and the etiology underlying atherosclerosis in SLE remain elusive. The antioxidant capacity of normal high-density lipoproteins (HDLs) is lost during inflammation, and these dysfunctional HDLs might predispose individuals to atherosclerosis. The aim of this study was to determine whether dysfunctional proinflammatory HDL (piHDL) is associated with subclinical atherosclerosis in SLE.

Methods

Carotid artery ultrasound was performed in 276 women with SLE to identify carotid plaques and measure intima-media thickness (IMT). The antioxidant function of HDL was measured as the change in oxidation of low-density lipoprotein after the addition of HDL cholesterol. Two antiinflammatory HDL components, paraoxonase 1 and apolipoprotein A-I, were also measured.

Results

Among the SLE patients, 48.2% were determined to have piHDL on carotid ultrasound, while 86.7% of patients with plaque had piHDL compared with 40.7% of those without plaque (P < 0.001). Patients with piHDL also had a higher IMT (P < 0.001). After multivariate analysis, the only factors found to be significantly associated with plaque were the presence of piHDL (odds ratio [OR] 16.1, P < 0.001), older age (OR 1.2, P < 0.001), hypertension (OR 3.0, P = 0.04), dyslipidemia (OR 3.4, P = 0.04), and mixed racial background (OR 8.3, P = 0.04). Factors associated with IMT measurements in the highest quartile were the presence of piHDL (OR 2.5, P = 0.02), older age (OR 1.1, P < 0.001), a higher body mass index (OR 1.07, P = 0.04), a cumulative lifetime prednisone dose ≥20 gm (OR 2.9, P = 0.04), and African American race (OR 8.3, P = 0.001).

Conclusion

Dysfunctional piHDL greatly increases the risk of developing subclinical atherosclerosis in SLE. The presence of piHDL was associated with an increased prevalence of carotid plaque and with a higher IMT. Therefore, determination of piHDL may help identify patients at risk for atherosclerosis.

Premature atherosclerosis is a major comorbid condition in systemic lupus erythematosus (SLE), with a 10–50-fold increased risk of myocardial infarction (MI) (1, 2). Women with SLE also have an increased risk of subclinical atherosclerosis, measured as the presence of plaque on carotid artery ultrasound (3, 4) or calcification in the coronary arteries (5, 6). Traditional cardiac risk factors explain some of this increased risk (2, 7), but there is still an up to 10-fold increased risk of MI after these traditional factors have been taken into account (2). Thus, standard measures alone cannot be reliably used to predict which SLE patients are at risk for cardiovascular morbidity.

Plasma levels of high-density lipoprotein (HDL) cholesterol are inversely related to cardiovascular disease in the general population (8). This relationship is actually quite complex and involves not only the quantity but also the function of HDL cholesterol (9). Total HDL cholesterol, a group of particles of varying size and content, is antiinflammatory in the basal state, through at least 2 mechanisms. First, normal HDL prevents oxidation of low-density lipoprotein (LDL) cholesterol, and thus reduces the ability of oxidized LDL to attract monocytes into arterial tissue, wherein they form foam cells (10). These antioxidant effects depend, in large part, on the HDL content of apolipoprotein A-I (Apo A-I) and of the enzyme paraoxonase 1 (10). During acute-phase responses, HDLs lose this antioxidant capacity and can even promote increased oxidation of LDLs, thus becoming proinflammatory in function (10–12). Second, normal HDL cholesterol mediates reverse transport of cholesterol by removing the cholesterol from the artery walls (10).

We hypothesized that the chronic inflammation of SLE may continually promote conversion of HDL cholesterol into a pro-oxidant, and thus proinflammatory, state, thereby increasing the risk of atherosclerosis. In support of this hypothesis, we previously reported that HDL function was abnormal and proinflammatory in 45% of women with SLE, compared with 20% of patients with rheumatoid arthritis and 4% of healthy control subjects. The odds ratio (OR) for the likelihood of having dysfunctional proinflammatory HDL (piHDL) in patients with SLE, compared with healthy controls, was 19.3 (95% confidence interval [95% CI] 4.4–85.3). The present study was undertaken to determine if the presence of this dysfunctional, pro-oxidant, and proinflammatory form of HDL (piHDL) can be used to predict subclinical atherosclerosis in women with SLE.

PATIENTS AND METHODS

Study population.

Two hundred eighty-one patients with SLE were enrolled in the study between February 2004 and November 2008. All subjects were seen in the rheumatology practices at the University of California, Los Angeles (UCLA) or Cedars-Sinai Medical Center in Los Angeles. In all patients, the diagnosis fulfilled at least 4 of the American College of Rheumatology (ACR) 1997 revised classification criteria for SLE (13). Because statins are known to alter HDL inflammatory function (9), subjects were excluded if they had taken statins within the prior 3 months or if they had renal failure (defined as a creatinine level >2.0 mg/dl), which also alters HDL function (14). Five subjects who completed all study procedures were found, on chart review, to be taking statins and were therefore removed from the final analysis. After all exclusions had been made, the final group consisted of 276 patients with SLE.

The patients' clinical data relevant to SLE, the frequency of cardiovascular disease events and risk factors, and the distribution of racial/ethnic groups are shown in Tables 1 and 2. The definitions used for cardiac risk factors, such as hypertension, diabetes, or family history of cardiac disease, are also given in Table 1. This study was approved by the Institutional Review Boards at UCLA and Cedars-Sinai Medical Center, and all participants gave their written informed consent.

Table 1. Demographic characteristics and cardiac risk factors at baseline in SLE patients with or without plaque*
 SLE patientsP
With plaque (n = 45)Without plaque (n = 231)
  • *

    Except where indicated otherwise, values are the mean ± SD. Dyslipidemia was defined as any of the following, either alone or in combination: levels of low-density lipoprotein (LDL) cholesterol ≥130 mg/dl, total cholesterol ≥200 mg/dl, high-density lipoprotein (HDL) cholesterol ≤40 mg/dl, and/or triglycerides ≥150 mg/dl. Hypertension was defined as use of antihypertensive medication or a systolic blood pressure >140 mm Hg or a diastolic blood pressure >90 mm Hg. Coronary artery disease (CAD) was defined as a history of myocardial infarction (MI) with appropriate documentation or CAD documented on angiogram or stress test. Cerebrovascular events included transient ischemic attacks (confirmed by a physician) and stroke (confirmed by appropriate imaging). Diabetes mellitus was defined as the presence of a fasting glucose level ≥7.0 mmoles/liter (126 mg/dl) or treatment with insulin or an oral hypoglycemic agent at the time of study entry. Smoking was reported if subjects had smoked any cigarettes within the 3 months prior to study entry. Family history of cardiovascular disease (CVD) was defined as a record of Ml, stroke, or sudden death in any first-degree relative before the age of 60 years. Mixed or other race/ethnicity was defined as a self-reported mixture of racial/ethnic backgrounds, including Asian/Caucasian (n = 3), African American/Caucasian (n = 3), Caucasian/Hispanic (n = 5), Native American/Caucasian (n = 1), Hispanic/Native American (n = 2), and Hispanic/African American (n = 2). SLE = systemic lupus erythematosus; NS = not significant; CRP = C-reactive protein.

Age, years55.7 ± 9.539.8 ± 12.0<0.001
Total cholesterol, mg/dl209.6 ± 42.4180.4 ± 41.8<0.001
HDL, mg/dl57.5 ± 17.656.0 ± 16.5NS
LDL, mg/dl124.7 ± 37.4102.2 ± 33.3<0.001
Triglycerides, mg/dl126.7 ± 68.7110.3 ± 71.8NS
Presence of dyslipidemia, no. (%)28 (62.2)61 (26.4)<0.001
Presence of hypertension, no. (%)27 (60.0)61 (26.4)<0.001
Systolic blood pressure, mm Hg118.0 ± 18.1112.1 ± 13.50.02
Diastolic blood pressure, mm Hg71.8 ± 12.470.9 ± 9.3NS
History of CAD, no. (%)3 (6.7)00.04
History of cerebrovascular events, no. (%)5 (11.1)14 (6.1)NS
High-sensitivity CRP, mg/liter4.0 ± 7.42.6 ± 6.7NS
Body mass index, kg/m227.0 ± 6.524.9 ± 6.90.05
Presence of diabetes, no. (%)5 (11.1)9 (3.9)0.04
History of smoking, no. (%)   
 Current4 (8.9)16 (6.9)NS
 Ever15 (33.3)64 (27.7)NS
Family history of CVD, no. (%)16 (35.6)47 (20.3)0.03
Race/ethnicity, no. (%)   
 Caucasian21 (46.7)115 (49.8)NS
 Asian or Pacific Islander4 (8.9)32 (13.9)NS
 African American12 (26.7)23 (10.0)0.002
 Hispanic4 (8.8)49 (21.2)NS
 Mixed or other4 (8.9)12 (5.2)NS
Table 2. Disease characteristics of SLE patients with or without plaque*
 SLE patientsP
With plaque (n = 45)Without plaque (n = 231)
  • *

    Except where indicated otherwise, values are the number (%) of patients. SELENA–SLEDAI = Safety of Estrogens in Lupus Erythematosus: National Assessment version of the SLE Disease Activity Index; SDI = Systemic Lupus International Collaborating Clinics/American College of Rheumatology Damage Index (see Table 1 for other definitions).

  • History of anticardiolipin antibody positivity included findings of positivity for IgG, IgM, or IgA on at least 2 separate occasions, 6 weeks apart.

History of glomerulonephritis (ever)13 (28.9)59 (25.5)NS
Disease duration, mean ± SD years16.8 ± 10.511.4 ± 7.80.002
SELENA–SLEDAI, mean ± SD2.9 ± 3.24.2 ± 4.10.06
SDI, mean ± SD2.0 ± 2.01.2 ± 1.60.01
History of lupus anticoagulant positivity3 (6.7)36 (15.6)NS
History of anticardiolipin antibody positivity11 (24.4)93 (40.3)NS
Current medications   
 Mycophenolate mofetil9 (20.0)53 (22.9)NS
 Hydroxychloroquine25 (55.6)152 (65.8)NS
 Cyclophosphamide0 (0)3 (1.3)NS
 Methotrexate2 (4.4)16 (6.9)NS
 Azathioprine4 (8.8)29 (12.6)NS
 Nonsteroidal antiinflammatory drugs23 (51.1)89 (38.5)0.1
 Glucocorticoids20 (44.4)104 (45.0)NS
Current prednisone dosage, mean ± SD mg/day3.1 ± 4.54.8 ± 8.50.06
Cumulative lifetime prednisone dose ≥20 gm18 (40.0)64 (27.8)0.07

Sample collection.

Blood samples were processed with sucrose cryopreservation. HDL cholesterol was isolated from the plasma using a magnetic bead dextran sulfate reagent (Reference Diagnostics, Bedford, MA) as previously described (9). Plasma lipid concentrations, levels of high-sensitivity C-reactive protein (hsCRP) and serum complement, the erythrocyte sedimentation rate, and autoantibodies against DNA and cardiolipin were measured in the UCLA clinical laboratory, using standard methods. On the day of plasma sampling, the disease activity in patients with SLE was assessed using a validated instrument, the Safety of Estrogens in Lupus Erythematosus: National Assessment (SELENA) version of the SLE Disease Activity Index (SLEDAI) (15). Organ damage was determined using the Systemic Lupus International Collaborating Clinics/ACR Damage Index (SDI) (16). The lifetime cumulative dose of prednisone in each patient was determined from chart review.

Carotid ultrasound.

Brightness-mode grey-scale, color, and spectral Doppler techniques were used to investigate the carotid arteries according to a standardized protocol (17). The same radiologist (NR) interpreted all ultrasound studies in a blinded manner, and the same ultrasound unit (Iu22; Philips Medical Systems, Bothell, WA) was used for scanning of all participants.

Various anatomic sites were examined for the presence of atherosclerotic plaque, defined as focal protrusion (intima-media thickening) into the arterial lumen with a thickness exceeding that of the surrounding wall by at least 50%; the anatomic sites examined were the common carotid arteries bilaterally, the carotid bulbs, the internal carotid arteries bilaterally, the external carotid arteries bilaterally, and the vertebral arteries bilaterally. The number, location, and sonographic appearance of each plaque was recorded. The intima-media thickness (IMT) of the far wall of the distal common carotid artery was measured 1 cm proximal to the flow divider, at the end diastole on cineloop real-time playback, and the results were assessed using automated QLAB software (Philips Medical Systems, Bothell, WA). IMT was never measured at the level of a plaque. The results from carotid ultrasound are presented as the mean of 3 values of the left and right segments.

Cell-free assay for measurement of HDL cholesterol function.

Dysfunctional HDL has historically been diagnosed using a cell-based assay that requires the presence of endothelial cells, smooth muscle cells, and monocytes; however, this assay is not practical for large-scale studies (18). A cell-free assay has been developed that rapidly detects dysfunctional HDL and that yields results that are highly comparable with those produced by the cell-based assay (9, 19). The cell-free assay is based on the ability of normal HDL to prevent oxidation of LDL. The presence of oxidized LDL leads to the conversion of normally nonfluorescent dichlorofluorescein diacetate (DCF-DA) into a fluorescent form (DCFH). DCFH is then measured on a plate reader (Spectra Max, Gemini XS; Molecular Devices, Sunnyvale, CA) set at an excitation wavelength of 485 nm and an emission wavelength of 530 nm. In this way, the change in fluorescence intensity resulting from oxidation of DCF-DA by LDL in the presence or absence of test HDL can be quantitated.

Dysfunctional HDL is unable to prevent the oxidation of LDL that occurs spontaneously in vitro, and actually increases oxidation, and thus can be considered pro-oxidant and proinflammatory. LDL cholesterol was prepared from normal plasma as previously described (19). Twenty microliters of the normal LDL-cholesterol solution (final concentration of 50 μg/ml) and 90 μl of test HDL cholesterol (at a final concentration of 10 μg/ml cholesterol) were incubated in quadruplicate in 96-well plates for 1 hour. Ten microliters of DCF-DA solution (0.2 mg/ml) was added to each well, and the mixture was incubated for 2 hours. The fluorescence intensity was then determined with a plate reader.

Values for the fluorescence intensity of DCFH activated by LDL cholesterol alone were normalized to 1.0. In addition to preventing the oxidation of LDL, the presence of dysfunctional HDL in the assay often amplified LDL oxidation and the subsequent DCFH formation. Therefore, fluorescence intensity values equal to or greater than 1.0 after the addition of test HDL cholesterol (HDL function scores) indicated the presence of dysfunctional piHDL; values less than 1.0 indicated the presence of normal, antiinflammatory HDL.

Measurement of Apo A-I and paraoxonase 1 activity.

Plasma paraoxonase 1 activity was measured using paraoxon as a substrate, according to the method previously described by Eckerson et al (20). Levels of Apo A-I were measured using an enzyme-linked immunosorbent assay (Mabtech, Cincinnati, OH).

Statistical analysis.

Data were analyzed using SPSS software, version 13.0 (SPSS, Chicago, IL). Skewed continuous variables were logarithmically transformed to attain a normal distribution. For variables that did not attain a normal distribution by logarithmic transformation, nonparametric tests were used. Study groups were compared using analysis of variance and Student's t-test for continuous parametric variables, Mann-Whitney test for nonparametric variables, and the chi-square test for categorical variables. Either Pearson's or Spearman's rank correlation was calculated, depending on whether the variable was normally distributed. The significance level was set at P values less than 0.05. Multiple regression analysis was used to build models identifying risk factors associated with the presence of plaque and with the highest quartile of IMT in patients with SLE.

RESULTS

Association of plaque with traditional cardiac and SLE-specific risk factors in SLE patients.

Forty-five (16.3%) of the 276 women in the SLE group had ≥1 area of plaque on carotid ultrasound. As determined by univariate analysis, SLE patients with plaque, compared with those without plaque, were more likely to have hypertension (defined as present or not present) (P < 0.001), a higher mean systolic blood pressure at the time of study entry (continuous variable) (P = 0.02), diabetes (defined as present or not present) (P = 0.004), a documented history of coronary artery disease (P = 0.04), a family history of cardiovascular disease (P < 0.001), older age (continuous variable) (P < 0.001), higher total cholesterol levels (P < 0.001), higher LDL cholesterol levels (P < 0.001), and a higher body mass index (BMI) (P = 0.05). They were also more likely to have any dyslipidemia (defined as levels of LDL cholesterol ≥130 mg/dl, total cholesterol ≥200 mg/dl, HDL cholesterol ≤40 mg/dl, and triglycerides ≥150 mg/dl) as measured in the blood sample obtained at study entry (P < 0.001 versus those without plaque). In addition, patients of African American descent were more likely to have plaque than were patients in the other racial/ethnic groups. There were no associations between the presence of plaque and a history of smoking (now or ever) or a documented history of cerebrovascular disease. There were also no associations between the presence of plaque and levels of HDL cholesterol, triglycerides, or hsCRP (Table 1).

The relationship between carotid plaque and disease activity or damage was also examined in the patients with SLE. A higher SDI was positively associated with the presence of plaque on carotid ultrasound (P = 0.01), as was a longer disease duration (P = 0.002). However, plaque was not significantly associated with previous renal disease or renal transplantation, active renal disease, a documented history of positivity for antiphospholipid antibodies, a higher SELENA–SLEDAI score at the time of blood sampling, current use of any SLE-related medications, including prednisone, hydroxychloroquine, cyclophosphamide, mycophenolate mofetil, azathioprine, methotrexate, or nonsteroidal antiinflammatory drugs, or the 6-month cumulative prednisone dose (Table 2).

Association of IMT with traditional cardiac risk factors in SLE patients.

The mean ± SD IMT in the SLE group was 0.55 ± 0.14 mm2. The relationship between cardiac risk factors and IMT measurements in patients with SLE was determined using a bivariate analysis stratified by IMT values in the highest quartile (those with IMT ≥0.65 mm2) compared with IMT values in the lowest 3 quartiles of IMT measurements. Patients in the highest quartile of IMT measurements, compared with the lowest 3 quartiles, were more likely to have hypertension (P < 0.001), older age (P < 0.001), a higher BMI (P < 0.001), and higher levels of total cholesterol (P < 0.001) and LDL cholesterol (P < 0.001). They were also more likely to have any dyslipidemia at the time of blood sampling at study entry (P < 0.001), to have a history of coronary artery disease (P = 0.02), and to be of African American descent (P < 0.001). In addition, SLE patients in the highest quartile of IMT were more likely to have a longer disease duration (P = 0.001), to have a cumulative lifetime prednisone dose ≥20 gm (P = 0.04), and to have a higher mean SDI (P = 0.01). Finally, SLE patients in the highest quartile of IMT were less likely to have a history of glomerulonephritis (P = 0.04) and were also less likely to be taking hydroxychloroquine (P = 0.002) than were SLE patients in the lowest 3 quartiles of IMT.

There were no associations between the highest quartile of IMT and current or past use of tobacco, a family history of cardiovascular disease, presence of diabetes, a history of cerebrovascular disease, levels of hsCRP, HDL cholesterol, or triglycerides, the SELENA–SLEDAI score at the time blood sampling, or a history of or current presence of antiphospholipid antibodies (Table 3). There were also no associations with current use of any SLE-related medications, including prednisone, cyclophosphamide, mycophenolate mofetil, azathioprine, methotrexate, or nonsteroidal antiinflammatory drugs, or with the 6-month cumulative prednisone dose (results not shown).

Table 3. Demographic characteristics and cardiac risk factors at baseline in SLE patients in the highest quartile of IMT measurements compared with those in the lowest 3 quartiles of IMT*
 SLE patientsP
Highest quartiles of IMT (n = 70)Lowest quartiles of IMT (n = 206)
  • *

    Except where indicated otherwise, values are the mean ± SD. The highest quartile of intima-media thickness (IMT) was defined as an IMT value ≥0.65 mm2. See Tables 1 and 2 for other definitions.

Age, years54.1 ± 9.838.3 ± 11.5<0.001
Total cholesterol, mg/dl201.0 ± 42.1180.0 ± 42.4<0.001
HDL, mg/dl57.1 ± 14.956.0 ± 17.2NS
LDL, mg/dl118.6 ± 35.6101.7 ± 33.8<0.001
Triglycerides, mg/dl126.8 ± 83.0108.4 ± 66.8NS
Presence of dyslipidemia, no. (%)32 (45.7)57 (27.6)<0.001
Presence of hypertension, no. (%)35 (50.0)53 (25.8)<0.001
History of CAD, no. (%)3 (4.3)0 (0)0.02
History of cerebrovascular events, no. (%)8 (11.4)11 (5.3)NS
High-sensitivity CRP, mg/liter2.8 ± 5.52.9 ± 7.2NS
Body mass index, kg/m228.6 ± 6.225.3 ± 6.2<0.001
Presence of diabetes, no. (%)6 (8.6)8 (3.8)NS
Current smoking, no. (%)3 (4.3)17 (8.3)NS
Family history of CVD, no. (%)21 (30)42 (20.4)0.14
History of glomerulonephritis, no. (%)12 (17.1)60 (29.1)0.04
Race/ethnicity, no. (%)   
 Caucasian33 (47.1)103 (50.5)NS
 Asian or Pacific Islander5 (7.1)31 (15.0)0.1
 African American21 (30.0)14 (6.8)<0.001
 Hispanic8 (11.4)45 (21.8)0.06
 Mixed or other4 (5.7)12 (5.2)NS
Disease duration, years15.5 ± 9.211.1 ± 7.90.001
SELENA–SLEDAI4.1 ± 4.43.9 ± 3.9NS
SDI1.8 ± 1.81.1 ± 1.60.01
History of lupus anticoagulant positivity, no. (%)6 (8.6)33 (16)NS
History of anticardiolipin antibody positivity, no. (%)21 (30)83 (40.3)NS
Current prednisone dosage, mg/day3.6 ± 7.04.8 ± 8.5NS
Cumulative lifetime prednisone dose ≥20 gm, no. (%)28 (40.0)54 (26.2)0.04
Current use of hydroxychloroquine, no. (%)35 (50)142 (68.9)0.002
Current use of mycophenolate mofetil, no. (%)11 (15.7)51 (24.8)0.11

Strong association of piHDL with subclinical atherosclerosis (both plaque and IMT) in SLE patients.

The HDL function score for the entire SLE cohort was indicative of the presence of piHDL, with a mean ± SD score of 1.09 ± 0.67. Using an HDL function score of ≥1.0 to define the presence of piHDL, 48.2% of SLE patients had piHDL. These results are similar to those in our previously reported cohort, in which the mean ± SD HDL function score was 1.02 ± 0.57 in SLE patients, compared with 0.68 ± 0.28 in controls (P < 0.001), and piHDL was found in 44% of the SLE patients in that cohort (21).

Among the 45 SLE patients with plaque in our current cohort, 39 (86.7%) had piHDL, compared with 94 (40.7%) of the 231 SLE patients without plaque (P < 0.001) (Figure 1). Patients with piHDL also had a higher mean number of plaques (mean ± SD 0.62 ± 1.2) than those with normal HDL function (mean ± SD number of plaques 0.10 ± 0.49; P < 0.001). In addition, the mean IMT was also significantly thicker in SLE patients with piHDL (mean ± SD IMT 0.57 ± 0.15 mm2) than in those with normal HDL function (mean ± SD IMT 0.53 ± 0.12 mm2; P = 0.001). Overall, the positive predictive value of a single positive finding of piHDL as being indicative of carotid plaque in SLE patients was only 29.3%. However, the negative predictive value of a single positive finding of piHDL as a measure of carotid plaque in SLE patients was 95.8%.

Figure 1.

Distribution of high-density lipoprotein (HDL) function scores in patients with systemic lupus erythematosus (SLE) with or without plaque. Each symbol represents HDL function scores for an individual SLE patient with or without plaque. The horizontal line at 1.0 indicates the dividing line for abnormal proinflammatory HDL (piHDL) scores (scores of ≥1.0 indicate the presence of piHDL). Horizontal lines above 1.0 indicate the mean scores for each group; values for the mean (SD) are also shown.

Lack of association between selected individual protective components of HDL and piHDL function and plaque.

We tested for associations between piHDL or subclinical atherosclerosis and plasma levels of Apo A-I or activity of paraoxonase 1, which are 2 known protective components of normal HDL. No associations were found between paraoxonase 1 activity or plasma levels of Apo A-I and the presence of plaque in SLE patients. There was an inverse correlation between IMT and paraoxonase 1 activity (r = −0.32, P = 0.001), but there was no correlation between IMT and levels of Apo A-I. Furthermore, there was no correlation between HDL function and levels of either paraoxonase 1 activity or Apo A-I.

Association of piHDL with atherosclerosis in SLE even after accounting for traditional cardiac and disease-specific risk factors.

Multivariate analysis determined which variables were most consistently associated with carotid plaque in SLE subjects. The model included traditional cardiac risk factors and SLE-associated factors that might affect the risk of developing cardiac disease. The factors that were significantly associated with carotid plaque were presence of piHDL (OR 16.1, 95% CI 4.3–59.6, P < 0.001), older age (OR 1.15, 95% CI 1.09–1.2, P < 0.001), hypertension (OR 3.0, 95% CI 1.05–8.8, P = 0.04), dyslipidemia (OR 3.4, 95% CI 1.06–10.7, P = 0.04), and mixed racial background (OR 8.3, 95% CI 1.2–59.7) (Table 4).

Table 4. Logistic regression analysis of traditional cardiac and disease-associated risk factors to assess the relationship of SLE with the presence of plaque on carotid ultrasound*
Explanatory variableOR95% CIP
  • *

    Odds ratios (ORs) were determined for each variable based on a yes/no determination in patients with systemic lupus erythematosus (SLE). 95% CI = 95% confidence interval; HDL = high-density lipoprotein; NS = not significant; SDI = Systemic Lupus International Collaborating Clinics/American College of Rheumatology Damage Index; SELENA–SLEDAI = Safety of Estrogens in Lupus Erythematosus: National Assessment version of the SLE Disease Activity Index.

  • The high-sensitivity C-reactive protein (hsCRP) level was logarithmically transformed because of the severely skewed nature of the nontransformed variable. Goodness of fit statistics showed a better fit for the model using the logarithmically transformed hsCRP variable compared with the nontransformed variable. No other variables in the model required transformation.

Proinflammatory HDL16.14.3–59.6<0.001
Older age (in years)1.21.09–1.2<0.001
Presence of dyslipidemia3.41.06–10.70.04
Presence of hypertension3.01.05–8.80.04
Current smoking5.50.8–36.60.07
Increased SDI0.70.5–1.010.08
Presence of diabetes1.070.1–10.4NS
Increased SELENA–SLEDAI0.90.8–1.06NS
Increased hsCRP level (in mg/liter)1.20.8–1.7NS
Higher body mass index (in kg/m2)1.000.97–1.09NS
History of glomerulonephritis2.90.7–12.0NS
Longer disease duration (in years)1.040.97–1.1NS
Use of hydroxychloroquine0.90.3–2.7NS
Cumulative lifetime prednisone dose ≥20 gm2.40.7–8.3NS
Family history of cardiovascular disease1.50.5–4.4NS
African American race3.50.8–14.90.09
Asian race2.20.4–11.6NS
Mixed race8.31.2–59.70.04
Hispanic ethnicity0.80.2–3.8NS

When multivariate analysis was used to determine the factors significantly associated with increased IMT, SLE patients in the highest quartile of mean IMT values (IMT values ≥0.65 mm2) were compared with those in the lowest 3 quartiles of IMT (Table 5). The factors significantly associated with a high IMT were presence of piHDL (OR 2.5, 95% CI 1.1–5.4, P = 0.02), older age (OR 1.1, 95% CI 1.07–1.2, P < 0.001), a higher BMI (OR 1.07, 95% CI 1.01–1.1, P = 0.04), cumulative lifetime prednisone dose ≥20 gm (OR 2.9, 95% CI 1.07–7.6, P = 0.04), and African American racial background (OR 8.3, 95% CI 2.4–28.9, P = 0.001) (Table 5).

Table 5. Logistic regression analysis of traditional cardiac and disease-associated risk factors in the relationship of SLE with the highest quartile of IMT*
Explanatory variableOR95% CIP
  • *

    The highest quartile of intima-media thickness (IMT) was defined as an IMT value ≥0.65 mm2. ORs were determined for each variable based on a yes/no determination in patients with SLE. See Table 4 for other definitions.

  • The hsCRP level was logarithmically transformed because of the severely skewed nature of the nontransformed variable. Goodness of fit statistics showed a better fit for the model using the logarithmically transformed hsCRP variable compared with the nontransformed variable. No other variables in the model required transformation.

Proinflammatory HDL2.51.1–5.40.02
Older age (in years)1.11.07–1.2<0.001
Higher body mass index (in kg/m2)1.071.01–1.10.04
Cumulative lifetime prednisone dose ≥20 gm2.91.07–7.60.04
Use of hydroxychloroquine0.50.2–1.10.09
Presence of dyslipidemia0.90.3–2.5NS
Presence of diabetes1.50.3–9.2NS
Presence of hypertension2.20.9–5.2NS
Current smoking0.90.2–5.0NS
Increased hsCRP level (in mg/liter)0.80.6–1.09NS
Longer disease duration (in years)1.030.98–1.09NS
Increased SDI0.90.7–1.2NS
Increased SELENA–SLEDAI1.070.97–1.2NS
Family history of cardiovascular disease1.10.5–2.6NS
History of glomerulonephritis0.40.1–1.3NS
African American race8.32.4–28.90.001
Asian race1.30.3–5.1NS
Mixed race3.40.7–16.9NS
Hispanic ethnicity0.90.3–2.9NS

DISCUSSION

SLE is associated with an increased risk of subclinical and clinical atherosclerosis (1–3), although the biologic mechanisms underlying this risk are not well understood. The results presented herein describe piHDL as a risk factor for subclinical atherosclerosis as determined on carotid ultrasound, with results indicating both an increase in the frequency of plaque and high IMT. There was no association between quantitative HDL levels and either the presence of plaque or high IMT in our cohort. Furthermore, quantitative HDL levels were not associated with HDL function in either this cohort (results not shown) or in our previously reported cohort of SLE patients (21), highlighting that biologic HDL function, not simply the quantity of HDLs, confers the risk in SLE. In agreement with this concept, higher levels of dysfunctional HDL have been demonstrated in a cohort of patients with known cardiovascular disease (9), as well as in patients with both inactive and active Crohn's disease, as compared with healthy controls (22). In addition, dysfunctional HDLs were associated with increased IMT in a small cohort of South Asian immigrants to the US, even after adjusting for quantitative HDL level, age, family history of cardiac disease, and hypertension (23).

Normal-functioning HDL cholesterol has several antiatherogenic properties. HDL transports excess cholesterol from cells in the artery walls to the liver for disposal (10, 24), removes reactive oxygen species from oxidized LDL (ox-LDL), prevents ox-LDL–mediated recruitment of inflammation mediators and monocytes into the vessel wall (25), and inhibits endothelial cell expression by adhesion molecules (26) and release of chemokines/cytokines (27). Several components of HDL contribute to these protective effects, including the plasma levels of Apo A-I and the activity of the enzyme paraoxonase 1 (28).

Conversely, piHDL cannot prevent oxidation of LDL cholesterol and actually increases it, leading to impairment of reverse transport of cholesterol, increased recruitment of monocytes, and probably an enhanced inflammatory response (11, 19). Multiple mechanisms confer proinflammatory characteristics on HDL molecules (29). In acute inflammation, hepatic synthesis of the protective lipoproteins in HDL cholesterol, including Apo A-I and antioxidant enzymes such as paraoxonase 1, is decreased (28). In addition, protective components in the HDL particles, including Apo A-I, are partly replaced with pro-oxidant acute-phase reactants such as serum amyloid A and ceruloplasmin (29, 30). Furthermore, HDL cholesterol and Apo A-I can be readily oxidized during periods of inflammation by myeloperoxidase, a product of white blood cell activation (30); oxidation of HDL probably contributes to its dysfunction. Oxidized HDL has proinflammatory characteristics (29, 30) and up-regulates the expression of proinflammatory genes such as cyclooxygenase 2 (31) and plasminogen activator inhibitor 1 (32) in endothelial cells.

Previous studies in SLE patients have described alterations in some protective components of HDL, including decreased paraoxonase 1 enzymatic activity (33, 34) and decreased Apo A-I levels (35). Our results show that piHDL is a better predictor of subclinical atherosclerosis than is either Apo A-I or paraoxonase 1 activity. Decreased paraoxonase 1 activity correlated with higher IMT, but not with the presence of plaque, in univariate, but not multivariate, analysis (results not shown). This is similar to the findings described in a group of subjects with metabolic syndrome who had higher numbers of piHDL than did control subjects with dyslipidemia, despite both groups having similar levels of HDL cholesterol and paraoxonase 1 activity (36). Interestingly, there was no association in either this cohort or our previously published cohort (21) between piHDL function and traditional markers of disease activity and inflammation, such as the hsCRP level or SELENA–SLEDAI score (results not shown).

Although high quantities of HDL cholesterol have been regarded as a negative risk factor for atherosclerosis, there is increasing evidence that HDL function may be as important as quantity for atheroprotection (28). This was highlighted recently by the failure of the experimental drug torcetrapib, a cholesterol ester transfer protein (CETP) inhibitor that increases quantitative levels of HDL cholesterol to protect from coronary artery disease (32–34). It has been suggested that CETP inhibition results in the formation of dysfunctional piHDL cholesterol (37). Evidence from studies by other groups also suggests that abnormal HDL function can contribute to excess mortality. In patients undergoing hemodialysis, the presence of piHDL was associated with a 2.5-fold increased risk of mortality over a 30-month period (14). It is not clear whether the abnormalities in HDL function described in patients undergoing hemodialysis are similar to those in patients with piHDL in SLE; however, the results presented herein highlight the importance of HDL function, in addition to HDL quantity, in the prevention of atherosclerosis in the general population. In patients with SLE, abnormal function of HDL seems more important than quantities of HDL in influencing the excess risk of atherosclerosis.

Our study has some limitations. Our study population differs from previously published SLE cohorts (3, 38) in that the prevalence of plaque in our SLE study group was lower than in previously published cohorts. Possible explanations include the exclusion from our study of individuals taking statins (which creates a bias toward patients without known hyperlipidemia and/or without clinical atherosclerosis [9]) and inclusion of a higher proportion of Asians, in whom subclinical atherosclerosis may have a lower prevalence than in other racial groups (39, 40). It is also possible that the SLE population in Los Angeles differs significantly from patients in other geographic areas, not only in ethnic composition but also in habits such as physical exercise (41), years of protection from public smoking (42), and hours of exposure to sunlight, which could influence vitamin D levels (43). The fact that cohorts from different centers thousands of miles apart differ in prevalence of plaque is not surprising given the differences in geography, climate, ethnic mix, behaviors, diet, exercise, and the many other factors that influence health.

Another limitation of our study is that we did not measure all potential atherosclerosis biomarkers. Most notably, since the inception of our study, several groups have demonstrated an association between high homocysteine levels and subclinical atherosclerosis in SLE (6, 44, 45). Future studies in our cohort will determine whether homocysteine levels and piHDL are independent predictors of atherosclerosis, or whether synergy exists between them.

For practical reasons, our study focused on the association between piHDL and subclinical atherosclerosis. There are currently no published studies that specifically demonstrate that lupus patients with carotid plaque or increased IMT have an elevated risk of cardiovascular events, and it is possible that these are not valid measures in women with lupus. Multiple studies in large cohorts, however, have demonstrated the predictive power of these measures in the general population (46). Further longitudinal studies are needed to establish the predictive power of carotid plaque and IMT in women with SLE, and also to determine whether the presence of piHDL can be predictive of future cardiovascular events in these patients.

In summary, the presence of piHDL contributes to an up to 17-fold increase in the likelihood of developing atherosclerosis in female patients with SLE. With a negative predictive value of ∼96%, piHDL may be one effective biomarker to determine which SLE patients are at low risk of developing subclinical atherosclerosis. These data also suggest that further studies are needed to determine whether interventions that restore the protective, antiinflammatory functions of HDL cholesterol, including treatment with statins (9), exercise and diet (47), and/or treatment with Apo AI–mimetic peptides (48), will be useful to prevent atherosclerosis in patients with SLE.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. McMahon had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. McMahon, Grossman, Skaggs, FitzGerald, Weisman, Wallace, Hahn.

Acquisition of data. McMahon, Grossman, Skaggs, FitzGerald, Sahakian, Ragavendra, Charles-Schoeman, Volkmann, Chen, Gorn, Karpouzas, Weisman, Wallace, Hahn.

Analysis and interpretation of data. McMahon, FitzGerald, Wong, Weisman, Wallace.

Acknowledgements

We thank Alan Fogelman, MD, and Mohamad Navab, PhD, for their advice regarding the analysis and interpretation of the data. We also thank Khoan Vu for assistance with data acquisition and management.

Ancillary