Systemic lupus erythematosus in a multiethnic US cohort (LUMINA): XXIII. Baseline predictors of vascular events

Authors


Abstract

Objective

To determine the baseline (time 0) risk factors associated with the subsequent occurrence of vascular events in a multiethnic US cohort (LUMINA [LUpus in MInorities: NAture versus nurture]) of patients with systemic lupus erythematosus (SLE).

Methods

Five hundred forty-six LUMINA patients were assessed at time 0 for traditional and nontraditional (disease-related) risk factors for vascular events. These were defined as 1) cardiovascular (myocardial infarction and/or definite or classic angina and/or the undergoing of a vascular procedure for myocardial infarction [coronary artery bypass graft]), 2) cerebrovascular (stroke), and 3) peripheral vascular (arterial claudication and/or gangrene or significant tissue loss and/or arterial thrombosis in peripheral arteries). The observation time (followup time in the cohort) was the interval between time 0 and the last visit. The unit of analysis was the patient and not each vascular event. Variables at time 0 and vascular events were examined by univariable and multivariable (logistic and Cox proportional hazards regression) analyses. Age, sex, ethnicity, followup time, and all known risk factors for the occurrence of vascular events were included in the model.

Results

Thirty-four patients (6.2%) developed one or more vascular event after time 0. The overall median duration of followup in the cohort was 73.8 months (range 10.8–111.3 months). Vascular events (13 cardiovascular, 18 cerebrovascular, 5 peripheral vascular) occurred in 7 Hispanics from Texas (6.5%), 1 Hispanic from Puerto Rico (1.2%), 15 African Americans (7.5%), and 11 Caucasians (7.1%). The mean total number of traditional risk factors was significantly higher in patients who developed vascular events than in those who did not (7.1 versus 5.6). Independent predictors of vascular events were older age, current smoking status, longer followup time, elevated serum levels of C-reactive protein (CRP), and the presence of any antiphospholipid antibody. The same variables were identified when time-dependent analyses were performed, although azathioprine use was also found to be a contributing factor.

Conclusion

Smoking, previously not reported in SLE, emerged as a predictor of vascular events and should be strongly discouraged. Antiphospholipid antibodies and CRP support the role of inflammation and autoimmunity in the development of accelerated atherosclerosis in SLE. Ethnicity was not associated with vascular events in our patients.

Survival of patients with systemic lupus erythematosus (SLE) has improved dramatically over the last few decades. Early diagnosis and recognition of patients with milder disease (perhaps not recognizable as SLE 50 years ago), the availability and judicious use of antihypertensive, antimicrobial, and immunosuppressive medication, and the increased use of hydroxychloroquine and of renal support probably account for this improved survival (1–6). However, as the lifespan of SLE patients continues to increase, late complications such as atherosclerotic vascular disease may occur. In fact, a bimodal pattern of mortality in SLE was described by Urowitz et al as early as 1976 (7). The first peak was attributed to active disease and infections, and the second peak (late mortality) was attributed to atherosclerotic vascular disease. More recent data have shown a 52-fold increase in the incidence of myocardial infarction in women with SLE ages 35–44 years compared with healthy women of the same age group (8).

In studies assessing the presence of clinical or subclinical atherosclerosis (3, 9, 10), traditional risk factors have failed to fully explain this increased risk (11, 12). Disease-related factors such as glucocorticoid use and renal disease may contribute to this increased risk; it is possible, however, that accelerated atherosclerosis is at least partly mediated by inflammation, a distinct component of the disease itself. In theory, therefore, traditional risk factors interacting with disease-related risk factors may predispose lupus patients to the development of atherosclerotic vascular disease (13–16).

If inflammation indeed contributes to accelerated atherosclerosis in SLE, patients with more active disease should fare worse in terms of vascular complications than those with less active disease, especially if other risk factors are also present. Given that lupus patients of Hispanic and African American ethnicities tend to have more active disease overall (17), it seems logical to study these patients. Thus, we sought to determine the risk factors implicated in the occurrence of vascular events in a multiethnic US lupus cohort that includes a large number of patients from these 2 ethnic groups. We were particularly interested in the possible roles of C-reactive protein (CRP) and antiphospholipid antibodies (aPL) in predisposing these patients to accelerated atherosclerosis and its consequences.

PATIENTS AND METHODS

As previously described (17, 18), LUMINA (LUpus in MInorities: NAture versus nurture) is a longitudinal study of outcomes in SLE patients from 3 ethnic groups living in the US: Hispanics from Texas and Puerto Rico, African Americans, and Caucasians. The study is conducted in 3 geographic areas (Alabama, Texas, and Puerto Rico) and at 3 institutions (The University of Alabama at Birmingham, The University of Texas–Houston Health Science Center, and The University of Puerto Rico Medical Sciences Campus).

Patients with SLE according to the American College of Rheumatology (ACR) criteria (19, 20) with disease duration of ≤5 years, defined ethnicity (all 4 grandparents of the same ethnicity as the patient), and living in the geographic catchment areas of the participating institutions are eligible to be enrolled in LUMINA. Every patient has a baseline visit (time 0); followup visits are conducted every 6 months for the first year (time 0.5 and time 1, respectively) and yearly thereafter (time 2, time 3, etc., to time L [the last available visit]). A LUMINA visit consists of an interview, a physical examination, and laboratory testing; a review of all previously available medical records is also performed at every study visit to obtain pertinent clinical information for the interval. Data for missed study visits are obtained, whenever possible, by review of medical records. Vascular events included in these analyses were obtained by physician assessment during LUMINA study visits and/or were documented in the medical records reviewed for the study visits.

Disease duration was defined as the time elapsing from the date the patient met 4 ACR criteria (diagnosis date, or time D) to time 0. Duration of followup in the cohort (hereinafter called followup time) was defined as the interval between time 0 and time L.

Variables.

As previously reported (18, 21), the LUMINA database includes variables from the following domains: socioeconomic–demographic, clinical, immunologic, immunogenetic, behavioral, and psychological. These variables are measured at time 0 and at every subsequent visit. Only the variables included in the present analyses will be described.

Variables from the socioeconomic–demographic domain were ethnicity, age, sex, education, health insurance, poverty (as defined by the US government and adjusted for the number of members in the household) (22), distance to the medical center where health care for SLE is provided, perceived difficulty in obtaining medical care, and unhealthful behaviors (smoking and sedentary lifestyle). Variables from the clinical domain were disease duration, followup time, family history of coronary artery disease and/or stroke (affected first-degree relative under age 60, regardless of sex), morbid obesity (weight >100 kg), disease activity and damage, comorbidities, ancillary laboratory test results, and medications.

Disease activity was assessed at both time 0 and time D using the Systemic Lupus Activity Measure (SLAM) (23). Disease damage was assessed using the Systemic Lupus International Collaborating Clinics/ACR Damage Index (SDI) (24). For patients with disease duration of <6 months at time 0, the first SDI scored (at time 0.5) was used.

Comorbidities included diabetes mellitus (self-reported and/or physician-diagnosed and/or requiring pharmacologic treatment), hypertension (defined as a systolic blood pressure ≥140 mm Hg and/or a diastolic blood pressure ≥90 mm Hg on 2 or more occasions and/or patient self-reported intake of antihypertensive medications, regardless of cause), valvular heart disease (including noninfectious endocarditis), cardiomyopathy, heart failure, peripheral venous vascular disease (thrombosis in visceral and/or peripheral veins or venous thrombosis causing residual swelling, cutaneous ulceration, or venous stasis), and pericardial involvement. Laboratory variables included levels of serum creatinine (>1.4 mg/dl, the cutoff for mild renal insufficiency), nonfasting serum lipoproteins (total cholesterol [high >200 mg/dl], high-density lipoprotein [HDL] cholesterol [low <35 mg/dl], triglycerides [high >205 mg/dl], and low-density lipoprotein [LDL] cholesterol calculated using the Friedewald formula [high >130 mg/dl]), and serum CRP, measured as high-sensitivity CRP (hs-CRP; high >16.5 mg/liter or the highest quintile for the distribution of our patients' values) by immunometric assay (Immulite 2000; Diagnostic Products, Los Angeles, CA).

Tests for autoantibodies included anti–double-stranded DNA by immunofluorescence against Crithidia luciliae (normal titer <1:10), IgG and/or IgM aPL (abnormal >13 IgG phospholipid units/ml and/or >13 IgM phospholipid units/ml; for the purpose of this study, these data are presented as the proportion of patients with abnormal values) by enzyme-linked immunosorbent assay (ELISA) technique (25), lupus anticoagulant by Stat-clot test (Diagnostica Stago, Asnieres-sur-Seine, France) (26, 27), and IgG and/or IgM anti–oxidized LDL antibodies (for the purpose of this study, abnormality was defined as values greater than the mean ± SD of values for 50 unselected healthy individuals) by ELISA (Specialty Laboratories, Santa Monica, CA) (28).

Medications included were current and past use of prednisone, immunosuppressive drugs (azathioprine and cyclophosphamide [either oral or intravenous]), and hydroxychloroquine. Durations of patients' exposures to these medications were not available. Selected class II HLA–DRB1 and HLA–DQA1 alleles and glutathione S-transferase (GST) M1 and T1 genotypes, ascertained by using standard laboratory techniques and appropriate oligonucleotides in previously extracted genomic DNA (21), were included from the immunogenetic domain. Behavioral and psychological variables included were helplessness as ascertained by the Rheumatology Attitudes Index (29), social support as determined by the Interpersonal Support Evaluation List (30), and coping with illness or illness-related behaviors as assessed with the Illness Behavior Questionnaire (31).

Outcome measures.

The observation time for the occurrence of vascular events was the followup time in the cohort (time 0 to time L); thus, only vascular events occurring after time 0 were included. We recognize that it would have been ideal to include all vascular events occurring after time D, but the predictors were from the baseline visit (time 0), and such a model was therefore believed to be less appropriate (and it is only briefly presented).

Vascular events were classified as follows: 1) cardiovascular, defined as the presence of coronary artery disease, including myocardial infarction (ischemic symptoms and electrocardiographic changes and/or biochemical markers of myocardial necrosis) and/or definite or classic angina (substernal chest discomfort provoked by exertion or emotion and relieved by rest or nitroglycerine) and/or the undergoing of a vascular procedure for myocardial infarction (coronary artery bypass graft); 2) cerebrovascular, defined as the presence of irreversible or partially reversible motor and/or sensory deficits of sudden or recent onset on the basis of vascular occlusion or insufficiency, complete, incomplete, or in evolution, persisting >24 hours or lasting <24 hours with an anatomic correlate; and 3) peripheral vascular, defined as the presence of arterial claudication (pain in the muscles of the upper or lower extremities, induced by exercise and relieved by rest, with absent pulses and/or confirmed by Doppler flow studies and/or angiography) lasting >6 months and/or evidence of gangrene or significant tissue loss (loss of a digit or a limb) and/or arterial thrombosis in peripheral arteries (documented by angiography).

Statistical analysis.

We identified 51 patients with vascular events, 17 of whom had an event prior to time 0; 8 events occurred prior to time D, and 9 occurred between time D and time 0. Thus, 34 patients were included in the present analyses; an alternative analysis was performed that included 43 patients whose events occurred after the diagnosis of SLE but prior to time 0. Given the different socioeconomic–demographic and clinical features of the Hispanic patients from Puerto Rico and Texas (32), these 2 Hispanic subgroups were analyzed separately. However, since only 1 Puerto Rican Hispanic patient developed a vascular event, within–ethnic group comparisons could not be performed for this ethnic subgroup.

The association between variables at time 0 and the occurrence of vascular events was first examined by univariable analyses. Non–time 0 variables included were disease activity ascertained at time D and the first available SDI score, if applicable. Categorical and continuous variables were examined by chi-square test and Student's t-test, respectively. Fisher's exact test was used when appropriate. Variables with a P value less than or equal to 0.10 were entered into a multivariable logistic regression analysis in which the dependent variable was the occurrence of a vascular event. The unit of analysis was the patient and not the event. Age, sex, ethnicity, followup time, and previously recognized risk factors for the occurrence of vascular events were entered in the regression regardless of their statistical significance in the univariable analyses.

As an alternative to this multivariable model, time to the occurrence of vascular events was examined by Cox proportional hazards regression. Additionally, traditional (male sex, age >45 years regardless of sex, smoking, sedentary lifestyle, family history of vascular disease, diabetes mellitus, obesity, hypertension, and hyperlipidemia) and nontraditional (followup time, disease activity, damage accrued, mean maximum dose of glucocorticoids, hydroxychloroquine use, aPL, and hs-CRP) risk factors were compared between patients who experienced vascular events and those who did not. All analyses were performed using SAS software version 8.1 (SAS Institute, Cary, NC).

RESULTS

Five hundred forty-six SLE patients were included in the present analyses. There were 107 Hispanics from Texas, 84 Hispanics from Puerto Rico, 200 African Americans, and 155 Caucasians. Vascular events after time 0 occurred in 34 patients (6.2%). The overall median duration of followup in the cohort was 73.8 months (range 10.8–111.3 months). Vascular events were documented in 7 Hispanics from Texas (6.5%), 1 Hispanic from Puerto Rico (1.2%), 15 African Americans (7.5%), and 11 Caucasians (7.1%). The relatively low frequency of vascular events noted in the Hispanics from Puerto Rico may relate to their having the shortest followup time in the cohort (mean ± SD 9 ± 5 months versus 49 ± 36 months for Hispanics from Texas; 41 ± 34 months for African Americans, and 41 ± 35 months for Caucasians; P < 0.001). Thus, followup time was accounted for in all of the analyses performed, as already noted. Among the vascular events, 13 were cardiovascular, 18 were cerebrovascular, and 5 were peripheral vascular (in 2 patients, 2 different vascular events occurred).

The occurrence of vascular events as a function of the socioeconomic–demographic and clinical features of the LUMINA patients at time 0 are shown in Tables 1 and 2. Patients who developed vascular events were older and were more likely to lack health insurance, to be poor, and to be current smokers than those who did not. Sex, education, and sedentary lifestyle, however, were comparable in the two groups. In addition, distance to the medical center where health care for SLE is provided and perceived difficulties in obtaining medical care were also comparable in both groups (data not shown). Vascular events were associated with longer followup time in the cohort (P < 0.001), but not with disease duration as defined (from time D to time 0). The proportion of patients with morbid obesity was higher in those who developed vascular events than in those who did not (20.6% versus 15.0%), but the difference was not statistically significant. In addition, vascular events were associated with a high SDI score (1.2 versus 0.7; P = 0.005) and with a diagnosis of valvular heart disease (5.9% versus 1.0%), although the difference did not reach statistical significance (P = 0.065). The mean SLAM score and the frequencies of a family history of cardiovascular disease and of comorbidities were comparable in the 2 groups.

Table 1. Occurrence of vascular events in LUMINA patients as a function of baseline socioeconomic and demographic features*
 Vascular eventsAll patients (n = 546)P
Patients with (n = 34)Patients without (n = 512)
  • *

    LUMINA = LUpus in MInorities: NAture versus nurture.

  • Only P values less than or equal to 0.10 are shown.

Ethnicity, %    
 Hispanic from Texas20.719.619.7 
 Hispanic from Puerto Rico2.916.215.4 
 African American44.136.136.6 
 Caucasian32.328.128.4 
Age, mean ± SD years45.7 ± 14.435.9 ± 12.236.5 ± 12.3<0.001
Women, %88.289.789.6 
Education, mean ± SD years12.1 ± 3.612.9 ± 3.112.9 ± 3.1 
Health insurance, %64.780.379.30.030
Poverty, %50.033.634.60.052
Sedentary lifestyle, %55.959.259.0 
Smoking, %32.412.313.60.001
Table 2. Occurrence of vascular events in LUMINA patients as a function of baseline clinical features*
 Vascular eventsAll patients (n = 546)P
Patients with (n = 34)Patients without (n = 512)
  • *

    Recorded at the baseline visit (time 0) unless noted otherwise. LUMINA = LUpus in MInorities: NAture versus nurture; time D = date of diagnosis; time L = date of last visit; CVD = cardiovascular disease; SLAM = Systemic Lupus Activity Measure; SDI = Systemic Lupus International Collaborating Clinics/American College of Rheumatology Damage Index; HDL = high-density lipoprotein; LDL = low-density lipoprotein.

  • Only P values less than or equal to 0.10 are shown.

  • IgG and/or IgM and/or lupus anticoagulant.

  • §

    Either oral or intravenous.

Time, mean ± SD months    
 Disease duration, time D to time 021.1 ± 17.417.0 ± 15.917.3 ± 16.0 
 Followup, time 0 to time L68.8 ± 29.335.5 ± 33.737.6 ± 33.4<0.001
Family history of CVD, %2.93.73.7 
Morbid obesity (weight >100 kg), %20.615.015.3 
SLAM score, mean ± SD (range 0–81)9.4 ± 4.89.6 ± 5.99.6 ± 5.8 
SDI score, mean ± SD (range 0–50)1.2 ± 1.20.7 ± 1.20.7 ± 1.20.005
Comorbidities, %    
 Diabetes mellitus2.92.72.7 
 Hypertension, regardless of cause47.133.834.6 
 Valvular heart disease5.91.01.30.065
 Cardiomyopathy5.92.02.2 
 Heart failure2.91.21.3 
 Thrombosis in peripheral veins8.84.95.1 
Serum creatinine >1.4 mg/dl, %14.76.87.3 
Serum cholesterol >200 mg/dl, %32.323.323.9 
HDL cholesterol <35 mg/dl, %76.581.381.0 
LDL cholesterol >130 mg/dl, %35.321.622.50.065
Triglycerides >205 mg/dl, %8.815.415.0 
Any antiphospholipid antibody, %29.412.113.20.004
IgM anti–oxidized LDL antibodies, mean ± SD units/ml44.1 ± 116.522.8 ± 37.624.1 ± 42.5 
Maximum prednisone dose, mean ± SD mg/day42.6 ± 25.437.2 ± 30.337.5 ± 30.0 
Azathioprine use ever, %20.613.313.8 
Cyclophosphamide use ever, %§11.814.714.5 
Hydroxychloroquine use ever, %29.428.728.7 

In patients who developed vascular events, there was a trend toward increased total serum cholesterol levels; likewise, the proportion of patients with raised LDL cholesterol was higher in those who developed vascular events than in those who did not (35.3% versus 21.6%). There were no differences related to the other serum lipoproteins examined (either low HDL cholesterol or triglycerides). Of the autoantibodies, only aPL were found to be positive more frequently in patients who developed vascular events than in those who did not (29.4% versus 12.1%; P = 0.004).

Past or current use of hydroxychloroquine, cyclophosphamide, and azathioprine was comparable in patients with and those without vascular events. The mean maximum dose of prednisone was higher in patients who developed a vascular event than in those who did not, although the difference did not reach statistical significance.

Neither the class II HLA–DRB1 and HLA–DQA1 alleles nor the GST M1 and T1 genes examined were associated with the occurrence of vascular events. Likewise, none of the psychosocial and behavioral features included in these analyses were associated with the occurrence of vascular events (data not shown).

The CRP data are shown in Table 3. The proportion of patients with hs-CRP values in the highest quintile was significantly higher among those who developed vascular events than among those who did not (38% versus 16%; P < 0.001); however, this difference only reached statistical significance in the Caucasians. Likewise, hs-CRP median values (mg/liter) were found to be significantly higher in patients who developed vascular events than in those who did not (12.6 mg/liter versus 4.8 mg/liter; P = 0.030). These differences were statistically significant in the Hispanic patients from Texas and in the Caucasians, but not in the African Americans.

Table 3. Vascular events in LUMINA patients as a function of ethnic group and CRP level*
 Percentage of patients with CRP level in highest quintileCRP level, median mg/liter
  • *

    Measured as high-sensitivity C-reactive protein (CRP). LUMINA = LUpus in MInorities: NAture versus nurture.

  • Of the distribution of values in patients studied.

  • Only P values less than or equal to 0.10 are shown.

  • §

    Within–ethnic group comparisons could not be performed for Hispanic patients from Puerto Rico, since only 1 patient in this group developed a vascular event.

Hispanic from Texas  
 With vascular events (n = 7)4314.5
 Without vascular events (n = 100)195.4
 P 0.025
African American  
 With vascular events (n = 15)273.4
 Without vascular events (n = 185)185.3
 P  
Caucasian  
 With vascular events (n = 11)5522.2
 Without vascular events (n = 144)134.3
 P<0.0010.022
All patients§  
 With vascular events (n = 34)3812.6
 Without vascular events (n = 512)164.8
 P<0.0010.030

The mean ± SD total number of cardiovascular risk factors in the patients who developed vascular events was 7.1 ± 2.2, compared with 5.6 ± 2.2 in those who did not (P < 0.001). In addition, the mean ± SD number of traditional and disease-related risk factors was also higher in patients who developed vascular events than in those who did not. These data, along with the breakdown by ethnic group, are shown in Table 4.

Table 4. Vascular events in LUMINA patients as a function of ethnic group and traditional and disease-related risk factors*
 Risk factors
TraditionalDisease-relatedTotal
Mean ± SDMedianMean ± SDMedianMean ± SDMedian
  • *

    See Table 3 for definitions.

  • Male sex, age >45 years, smoking, sedentary lifestyle, family history of vascular disease, diabetes mellitus, obesity, hypertension, and hyperlipidemia.

  • Followup time, disease damage, mean maximum dose of glucocorticoids, hydroxychloroquine use, antiphospholipid antibodies, and high-sensitivity CRP.

  • §

    Only P values less than or equal to 0.10 are shown.

  • Within–ethnic group comparisons could not be performed for Hispanic patients from Puerto Rico, since only 1 patient in this group developed a vascular event.

Hispanic from Texas      
 With vascular events (n = 7)4.4 ± 1.45.04.0 ± 1.44.08.4 ± 1.69.0
 Without vascular events (n = 100)2.7 ± 1.52.03.2 ± 1.03.05.9 ± 1.96.0
 P§0.0040.0050.043 0.0010.002
African American      
 With vascular events (n = 15)3.3 ± 2.03.03.3 ± 1.03.06.6 ± 2.36.0
 Without vascular events (n = 185)3.0 ± 1.73.03.1 ± 1.33.06.1 ± 2.46.0
 P§      
Caucasian      
 With vascular events (n = 11)4.0 ± 1.93.03.2 ± 0.93.07.2 ± 2.27.0
 Without vascular events (n = 144)3.0 ± 1.63.02.5 ± 1.13.05.5 ± 2.16.0
 P§0.0450.0740.0680.0580.0130.025
All patients      
 With vascular events (n = 34)3.7 ± 1.93.03.4 ± 1.13.07.1 ± 2.27.0
 Without vascular events (n = 512)2.9 ± 1.63.02.7 ± 1.23.05.6 ± 2.26.0
 P§0.0050.0110.0030.003<0.001<0.001

Multivariable analyses.

The results of the multivariable analyses are shown in Table 5. Variables significantly and independently associated with the occurrence of vascular events in this multiethnic US SLE cohort were older age, current smoking status, longer followup time in the cohort, an elevated serum level of hs-CRP, and the presence of any antiphospholipid antibody.

Table 5. Predictors of the occurrence of vascular events in LUMINA patients by multivariable logistic regression analyses*
 OR (95% CI)P
  • *

    OR = odds ratio; 95% CI = 95% confidence interval; CRP = C-reactive protein (see Table 2 for other definitions).

  • Only P values less than or equal to 0.10 are shown.

  • Highest quintile of the distribution of values in patients studied.

  • §

    IgG and/or IgM and/or lupus anticoagulant.

Age1.075 (1.037–1.114)<0.001
Smoking3.731 (1.391–10.000)0.009
Followup time (time 0 to time L)1.452 (1.223–1.725)<0.001
High-sensitivity CRP, mg/liter3.356 (1.264–8.929)0.015
Any antiphospholipid antibody§4.717 (1.675–13.158)0.003

When the data were examined using a Cox proportional hazards regression model, the variables found to be significant were the same as those in the model presented in Table 5; however, using this approach, azathioprine use was also found to be a risk factor for the occurrence of vascular events. These data are shown in Table 6.

Table 6. Predictors of the occurrence of vascular events in LUMINA patients by multivariable Cox proportional hazards regression analyses*
 HR (95% CI)P
  • *

    LUMINA = LUpus in MInorities: NAture versus nurture; HR = hazard ratio (see Table 5 for other definitions).

  • Only P values less than or equal to 0.10 are shown.

  • Highest quintile of the distribution of values in patients studied.

  • §

    IgG and/or IgM and/or lupus anticoagulant.

Age1.066 (1.032–1.100)<0.001
Smoking2.596 (1.043–6.463)0.0404
High-sensitivity CRP, mg/liter3.904 (1.507–10.115)0.0050
Any antiphospholipid antibody§3.463 (1.413–8.487)0.0066
Azathioprine use1.452 (1.215–10.378)0.0205

Alternative analyses.

Univariable analyses that included all patients who developed vascular events after time D (n = 43) yielded results that were entirely consistent with the data presented (for 34 patients in the present analyses) and, in some cases, even more convincing. Such was the case, for example, for family history of cardiovascular disease, diabetes mellitus, maximum glucocorticoid dose, and SLAM scores, which were either more frequent or higher among patients who developed a vascular event than among those who did not (data not shown). In terms of the multivariable analyses, the variables identified were exactly the same as when only those patients who developed vascular events after time 0 were included (older age, smoking, longer total disease duration, elevated serum level of hs-CRP, and the presence of aPL) (data not shown).

DISCUSSION

We assessed the baseline risk factors associated with the occurrence of vascular events in a multiethnic US lupus cohort that includes patients from 2 Hispanic subgroups, one predominantly of Mexican ancestry and the other from the island of Puerto Rico. Although Hispanic patients have been included in other studies examining atherosclerosis and/or vascular complications in SLE (33–35), their total number has been small, precluding within–ethnic group comparisons.

The overall frequency of vascular events in our cohort was ∼7.0% across all ethnic groups, with the exception of the Hispanics from Puerto Rico (1.2%); however, the mean followup time for this Hispanic subgroup was shorter than that for all other groups. As these patients are followed up longer in the cohort, they may accrue more vascular damage. We should add, however, that followup time was included as a variable in the analyses performed. Moreover, the variables identified as predictors of vascular events were exactly the same when time-dependent analyses were performed.

The incidence of vascular events in our cohort is comparable with data from 3 large North American lupus cohorts (6.7–10.0%) of different ethnic composition (predominantly Caucasian for the Toronto and Pittsburgh cohorts [8, 14] and African American and Caucasian for the Johns Hopkins cohort [13]). As in our study, ethnicity per se was not found to predispose to the occurrence of vascular events in these studies (8, 13, 14). Given the fact that African Americans tend to have more active disease, we were somewhat surprised when we found that ethnicity was not an independent contributor to the occurrence of vascular events in our patients (even if we included all cases occurring after SLE diagnosis). However, disease activity was not found to be an independent contributor to the occurrence of vascular events in our patients. Moreover, African Americans with or without vascular events had the same number of traditional and nontraditional risk factors, which was not the case for the Hispanics from Texas and the Caucasians. It is entirely possible that other factors which we have not measured are responsible for the occurrence of vascular events in our African American SLE patients; one such factor that could be operative is an enhanced endogenous fibrinolytic activity with increased fibrinogen, factor VIII, von Willebrand factor, and antithrombin III levels, which have been observed in African Americans to a larger extent than in Caucasians (36, 37).

We, like other investigators (8, 10, 13, 16, 38–40), have found that older age is a risk factor for the occurrence of vascular events in SLE. It should be emphasized, however, that we are referring to older age within a cohort of (predominantly) young women, which contrasts in a very substantial manner with what occurs in the general population. LUMINA patients who developed vascular events had a higher total number of risk factors (including traditional and disease related) than those who did not (an overall excess of 1.5 risk factors); however, these differences did not reach statistical significance for the African Americans. In addition, modifiable risk factors such as smoking and obesity were also found to be higher in patients who developed vascular events; again, the differences did not reach statistical significance for the African Americans (data not shown).

Of note, this is the first study to demonstrate that smoking contributes significantly and independently to vascular events in SLE. Such an association has been suggested by Wajed et al (41); however, smoking and subclinical atherosclerosis (carotid plaque and coronary artery calcifications) have been shown to be associated (15, 42). Smoking is known to increase CRP levels; however, we have shown that smoking and CRP contributed independently to vascular events. These data, as well as the literature cited, suggest that smoking cessation programs should be part of the overall treatment strategy for lupus patients who smoke.

Dyslipidemia in SLE, as manifested by high levels of total and LDL cholesterol, high levels of triglycerides, and low levels of HDL cholesterol, is proatherogenic (3, 8–13, 16, 38, 43, 44). In our study, we found no differences in triglyceride and HDL cholesterol levels in these 2 patient groups, but the proportion of patients with elevated LDL cholesterol was significantly higher in those who developed events than in those who did not; this variable, however, was not retained in the multivariable model. It is possible that the lipid abnormalities described in SLE by other investigators reflect active disease and enhanced activity of the tumor necrosis factor α (TNFα)/TNF receptor system (45); furthermore, only sustained lipid profile abnormalities (e.g., hypercholesterolemia) may constitute a vascular risk factor (46). Although hypertension was not retained in the multivariable model, it was found to be present more frequently in our patients with vascular events. Hypertension has also been found to be associated with vascular events by other investigators (13, 33, 47); therefore, tight control of blood pressure may significantly reduce the rate of atherogenesis and its associated events in SLE.

Of all the risk factors that we have identified, followup time has also been shown to be important by Manzi et al (8) and Petri et al (13), although those investigators measured total disease duration (the time elapsed from diagnosis to the time of the analyses) rather than followup time as we did in the present study. Since age was controlled in our multivariable model, longer “exposure” to the SLE inflammatory process and to medications (e.g., glucocorticoids) is probably the basis for followup time contributing to this finding (8, 13, 14). We have found CRP levels in a clearly inflammatory range and the presence of any antiphospholipid antibody to be of importance; they can be considered vascular biomarkers and can be used to identify at-risk patients who may benefit from prophylaxis with antithrombotic agents and/or statins. However, the exact meaning of the antiphospholipid antibody and CRP findings in SLE deserves to be further explored.

CRP levels are now considered a powerful predictor of cardiovascular events, both in healthy individuals and in those with disease (48–51). We found higher median CRP levels in patients with vascular events than in those without them; the differences within ethnic groups were significant in the Hispanics from Texas and in the Caucasians. When the proportions of patients with elevated CRP levels in the highest quintile were examined instead of median CRP levels, the differences between those who developed events and those who did not were significant for all patients as well as for the Caucasians (the proportions of patients with and those without events were still higher in the other ethnic groups). It is possible that genetic variations in the CRP gene(s) may account for different CRP levels prior to the occurrence of SLE, and as the disease ensues, patients genetically predisposed to high CRP levels will be at higher risk than other patients of developing accelerated atherosclerosis, especially if other risk factors are also present. Of note, in our patients, CRP levels were modestly correlated with disease activity and disease damage (data not shown).

Our findings further support a role for CRP as an independent predictor of vascular events in a disease that usually fails to elicit a major CRP response (52). Our results differ from those reported by Roman et al (3), who found no statistical association between the levels of CRP and the presence of carotid plaque, but they are consistent with those reported by Selzer et al (9) and Svenungsson et al (16), who showed an association between CRP levels and carotid intima-media thickness.

Our findings also support the notion of an atherogenic role for aPL, as has been reported by other investigators (53–55). These antibodies may enhance the development of atherosclerosis by oxidative modification of lipoproteins (LDL or lipoprotein[a]) and by promoting the uptake of lipoproteins by macrophage scavenger receptors at the endothelial level (56). More recently, aPL have been shown to reduce the activity of plasma paraoxonase (an antioxidant enzyme circulating in plasma attached to HDL) in patients with SLE and primary antiphospholipid syndrome (57).

Like Asanuma et al (38), we found no association between long-term treatment with glucocorticoids and atherosclerosis. In contrast, Roman et al (3) found a negative association between carotid plaque and a high mean daily dose of prednisone; it is possible that glucocorticoids, by modulating the immune response and diminishing CRP levels, may protect against atherosclerosis, but this has yet to be proven (58). However, treatment duration and cumulative dose, variables that we did not examine, have been found to predispose to atherosclerosis (8, 13, 39).

Patients who developed vascular events were more likely to have low socioeconomic status (SES) compared with those who did not; however, none of the SES variables examined were retained in the multivariable model. Despite these negative data, we cannot rule out the possibility that low SES may have contributed to some extent to the development of atherosclerosis, as has been suggested by other investigators (59–61).

The present study is not without limitations. First, we have ascertained risk factors not at disease onset, but at the time patients entered the cohort (time 0). Moreover, because the risk factors were ascertained at time 0 and not at diagnosis, we excluded those patients who developed vascular events prior to time 0. Of interest, these patients had higher frequencies of diabetes mellitus and a family history of cardiovascular disease, had received higher doses of glucocorticoids, and had higher SLAM scores than the patients included in the analyses presented, suggesting once more that the exposure to traditional and disease-related factors explains the occurrence of vascular events in SLE (for consistency, however, patients who developed vascular events between time D and time 0 were excluded from the analyses altogether, as cases and as controls).

Second, we recognize that not all vascular events result from the same pathophysiologic mechanism; some may be primarily thrombotic, others may be the result of atherosclerosis, and still others may result primarily from an ongoing inflammatory process. However, in many instances, atherosclerosis and thrombosis occur together; lacking histopathologic evidence, it is impossible in many cases to state with any degree of certainty whether or not there is a component of active vascular inflammation. Thus, we chose to examine all vascular events together regardless of their possible different underlying pathophysiologic mechanisms or involved organs. Furthermore, even if we had been able to separate the events according to their type, we would not have had sufficient statistical power to reach adequate conclusions.

Third, lipoproteins were measured in stored sera obtained from nonfasting patients, which is less than ideal according to the National Cholesterol Education Program, which recommends testing lipoproteins in sera obtained from fasting patients (62). Fourth, aPL were ascertained only at time 0, which may significantly limit their predictive value. Finally, we have not measured homocysteine levels, and therefore, this variable was not included in the model.

In conclusion, despite their young age, SLE patients are at increased risk of developing vascular events (8). Several factors may account for the development of endothelial damage and early vascular events, including genetic factors, older age, smoking, hypertension, glucocorticoids, dyslipidemia, immune complexes, autoantibodies, CRP, and complement activation, among others. Of these, we found smoking, aPL, CRP, and followup time in the cohort to be of importance in this multiethnic SLE cohort. The approach to the prevention of vascular events in SLE should therefore include the control of traditional (e.g., smoking) and disease-related risk factors to substantially reduce or delay the occurrence of these potentially fatal events.

Ancillary