To determine if systemic lupus erythematosus (SLE) is associated with a higher prevalence of coronary artery disease (CAD) in select patients undergoing coronary angiography. We compared the extent of angiographic abnormalities, CAD risk factors, and all-cause mortality in SLE patients with non-SLE controls.
We identified SLE patients (n = 86) and controls matched by sex and year of cardiac catheterization (n = 258) undergoing cardiac catheterization for the evaluation of CAD (median followup duration of 4.3 years). Multivariable logistic regression was used to determine if SLE was associated with obstructive CAD, defined as ≥70% stenosis in a major epicardial coronary artery. Risk-adjusted survival differences between the 2 groups were assessed using Cox proportional hazards modeling.
The SLE patients (85% women) were younger than the non-SLE patients (median age 49 years versus 70 years; P < 0.001) and were less likely to have diabetes mellitus and hyperlipidemia, but had similar rates of hypertension (70% versus 71%; P = 0.892). In unadjusted analyses, SLE and non-SLE patients had similar rates of obstructive CAD by angiography (52% versus 62%; overall P = 0.11). After adjustment for known CAD risk factors, SLE was associated with a significantly increased likelihood of CAD (odds ratio 2.24 [95% confidence interval (95% CI) 1.08–4.67]). SLE was also associated with a nonsignificant increase in all-cause mortality (hazard ratio 1.683 [95% CI 0.98–2.89], P = 0.060).
In this selected population, SLE was significantly associated with the presence of CAD as defined by coronary angiography, the gold standard for assessing flow-limiting lesions in this disease. The patients with SLE showed a similar severity of CAD as the controls despite having less than half the rate of diabetes mellitus and being 20 years younger.
Coronary artery disease (CAD) is a significant cause of morbidity and mortality for patients with systemic lupus erythematosus (SLE) (1). Epidemiologic data suggest that SLE is an independent risk factor for CAD (2–5), and in a prospective study, after controlling for traditional risk factors, the relative risk for developing a myocardial infarction (MI) was shown to be 10.1 (6). Also, women with SLE between the ages of 35 and 44 years are more likely to have an MI than controls (2). It is believed that these adverse cardiovascular outcomes are a result of accelerated atherosclerosis, leading to the development of occlusive CAD.
To date, however, most of our understanding about atherosclerosis in patients with SLE is derived from autopsy studies or investigations of surrogates of atherosclerosis, such as increased carotid intima-media thickening (IMT), reduced brachial artery flow-mediated dilation (BAFMD), and coronary artery calcification as detected by electron beam computed tomography (7–10). Recent studies have suggested that these modalities may not be as reliable in patients with SLE as they are in the general population. In one such study of SLE patients, the results of BAFMD correlated poorly with results from myocardial perfusion scanning (11); another study in this patient population suggested that carotid IMT measurements do not correlate with the presence of carotid plaque (8).
Little is known about the severity and distribution of coronary atherosclerosis in patients with SLE. The gold standard for measuring flow-limiting CAD is coronary angiography, which is an invasive procedure with its own inherent risks. Most descriptions of the coronary angiograms of SLE patients are from case reports. There are a few studies evaluating coronary angiograms in SLE (12, 13), but to our knowledge, no studies have explored in a large cohort the severity of CAD defined by coronary angiography in this population. Therefore, we took advantage of a large cardiovascular data bank at our institution to investigate the severity of CAD by coronary angiography in a selected population of patients with SLE. We investigated the role of SLE in predicting coronary angiographic abnormalities and adverse clinical outcomes after accounting for traditional CAD- and SLE-related risk factors.
Significance & Innovations
This is a large study evaluating coronary angiograms in a select group of patients with systemic lupus erythematosus (SLE).
This study suggests that patients with SLE develop accelerated atherosclerosis, as confirmed by coronary angiogram.
PATIENTS AND METHODS
Patients undergoing cardiac catheterization (between January 1986 and October 2008) with International Classification of Diseases, Ninth Revision (ICD-9) diagnoses of SLE (code 710.0), SLE and rheumatoid arthritis (codes 710.0 and 714.0, respectively), and/or undifferentiated connective tissue disease (code 710.9) up to 5 years precatheterization or 1 year postcatheterization were identified. The inclusion criteria were 1) a diagnosis of SLE, as confirmed by a chart review or communication with the primary physician (see details below), and 2) a cardiac catheterization performed to evaluate suspected CAD, i.e., diagnosis including acute coronary syndrome, angina, shortness of breath, or findings of cardiac ischemia by noninvasive stress testing. Patients who had a cardiac catheterization for clinical heart failure or valvular heart disease believed to be related to CAD were also included. This selection included patients in whom the cardiologist at the time of the intervention had a reasonable clinical suspicion of CAD. Patients with ICD-9 codes for scleroderma (710.1), dermatomyositis (710.3), or polymyositis (710.4), as well as a history of congenital heart disease or heart transplant, were excluded.
The diagnosis of SLE was confirmed by a chart review and communication with other physicians involved in the patient's care using a previously validated physician SLE criteria checklist (14) that includes components of the American College of Rheumatology (ACR) classification criteria for SLE (15). If patients fulfilled the ACR criteria for SLE, they were referred to as having definite SLE. The patients whose medical information was insufficient to fulfill the ACR criteria but had documented use of immunomodulatory or immunosuppressive medications (prednisone, hydroxychloroquine, azathioprine, mycophenolate mofetil, methotrexate, or cyclophosphamide) for the treatment of active SLE were categorized as having clinical SLE. For the purpose of this study, both groups were analyzed as a single group.
The controls were randomly selected and matched by sex and year of catheterization. According to our study design, 3 controls were matched to every 1 SLE patient. Similar to the SLE patients, the controls were individuals who had undergone cardiac catheterization for the evaluation of CAD. Possible controls were excluded if they had any of the following ICD-9 codes: 710.x (diffuse diseases of the connective tissues), 714.x (rheumatoid arthritis and other polyarthropathies), 720.x (ankylosing spondylitis and other spondylarthropathies), 725 (polymyalgia rheumatica), 446.x (polyarteritis nodosa and allied conditions), 447.x (arteritis, unspecified), and 696.x (psoriasis). Patients with congenital heart disease and a history of a cardiac transplant were also excluded from the control group.
All subjects were identified from the Duke Databank for Cardiovascular Diseases (DDCD), a prospective clinical data set containing detailed clinical, angiographic, and therapeutic data for over 180,000 patients that have undergone cardiac catheterization at Duke University Medical Center. Established in 1969, the DDCD is one of the oldest and largest continuously maintained observational databases in the world. Patients in the DDCD represent a broad population of individuals with CAD and a varied clinical presentation.
In the DDCD, the results of the coronary angiograms are collected to generate a clinical report and are entered into the database by a cardiology fellow in a standardized way shortly after the procedure. The data are reviewed by an attending cardiologist and included in the medical chart and are distributed to referring physicians. Other data collected in a standardized way on these patients at the time of cardiac catheterization consist of baseline characteristics (age, sex, race, and body mass index) and information about traditional CAD risk factors, including hypertension, diabetes mellitus, smoking, hyperlipidemia, and family history, as well as history of peripheral arterial disease, cerebrovascular disease, and other comorbidities as specified in the Charlson comorbidity index (16). Determination of hyperlipidemia as a risk factor is based on a previous diagnosis or a history of statin use; however, serum cholesterol levels are not routinely collected for the database.
Outcome data are collected for all patients with obstructive CAD by angiogram. These patients are routinely contacted at 6 months, 1 year, and annually thereafter using mailed questionnaires and telephone interviews. Patient medications and vital status, including death, cardiovascular events, hospitalizations, and revascularizations, are determined during followup. A search of the National Death Index and Social Security Death Index databases is performed for those patients whose vital status or cause of death is unknown. Clinical events are verified using source documentation. Further details regarding the organization of the Duke computerized cardiovascular database and followup methods have been previously described (17, 18).
For patients without obstructive coronary atherosclerosis determined by angiogram (i.e., atherosclerotic lesion <70%), detailed baseline information is collected in the same manner and is included in the database. However, outcome data are not routinely collected, and therefore information on patient vital status was ascertained from a medical chart review and search of the National Death Index and Social Security Death Index databases. The Duke University Institutional Review Board (IRB) approved the study. The IRB granted a waiver for patient consent since it was a retrospective review.
The baseline characteristics were compared using Wilcoxon's rank sum test for continuous variables and Pearson's chi-square test for categorical variables. A multivariable logistic regression model was developed to evaluate variables that were associated with CAD, defined as obstructive stenosis (≥70%) in at least 1 major coronary artery. The variables examined for possible inclusion in the model were selected by 1 of 3 criteria: statistical strength of the association with the outcome (unadjusted association P < 0.1), known risk factors associated with CAD, and clinical judgment of the investigators regarding other factors that may contribute to accelerated atherosclerosis, such as taking prednisone or other immunosuppressives. The final model was developed using forward stepwise selection. Additionally, we examined the following interactions: SLE and age, SLE and diabetes mellitus, diabetes mellitus and hyperlipidemia, diabetes mellitus and race, diabetes mellitus and hypertension, diabetes mellitus and smoking, and diabetes mellitus and prior MI. The interactions were tested in the multivariable setting while including all significant main effects.
The unadjusted survival results were examined using Kaplan-Meier methods, and comparisons between groups were made using the log rank test. Variables were examined using Cox proportional hazards regression modeling in both the unadjusted and adjusted settings. Candidate variables were selected using the same criteria as described above, and variables were entered into the model in a stepwise approach. Continuous and ordinal categorical variables were tested for linearity over the log hazard or logit (depending on the model type) and were transformed as necessary to meet this modeling assumption. All tests were 2-sided and were performed using SAS, version 8.2. Results were declared significant at a P value of less than 0.05.
Characteristics of the study cohort.
In total, 86 patients were confirmed to have SLE (73 with definite SLE and 13 with clinical SLE) (Figure 1), and 258 controls that were matched for sex and year of catheterization were identified. The median followup duration for all subjects in this study was 4.3 years (interquartile range [IQR] 1.9–8.0 years). The majority of the subjects were women, with significantly more African Americans in the SLE group compared with the non-SLE controls. Both groups had similar rates of cardiovascular risk factors, except for diabetes mellitus and hyperlipidemia, which occurred less commonly in patients with SLE. There were also significantly more patients with SLE who were receiving dialysis (Table 1).
Table 1. Baseline demographics and cardiovascular risk factors*
SLE (n = 86)
Control (n = 258)
OR (95% CI)
Values are the percentage unless otherwise indicated. SLE = systemic lupus erythematosus; OR = odds ratio; 95% CI = 95% confidence interval; IQR = interquartile range; CAD = coronary artery disease; MI = myocardial infarction.
Age, median (IQR) years
History of hypertension
History of diabetes mellitus
History of hyperlipidemia
History of smoking
Family history of CAD
History of MI
Seventy-three of the 86 patients fulfilled the ACR criteria for the diagnosis of SLE (Table 2). The SLE patients had a mean ± SD of 5 ± 1 of the 11 ACR SLE criteria met. Although we were unable to assess SLE severity or calculate disease activity scores (i.e., the Systemic Lupus International Collaborating Clinics [SLICC] score) at the time of catheterization, information regarding medications used prior to catheterization was collected by a chart review. Among the patients with SLE, 64 (88%) of the 73 patients were taking prednisone, 33 (45%) were taking hydroxychloroquine, 18 (25%) were taking cyclophosphamide, 7 (10%) were taking azathioprine, 7 (10%) were taking methotrexate, and 4 (5%) were taking mycophenolate mofetil. Twenty-five percent of the SLE patients had a history of kidney disease as defined by the ACR classification criteria for SLE, and 10% were receiving hemodialysis. Twenty-six of the SLE patients tested positive for serum antiphospholipid antibodies, and 9 of these patients were confirmed by a chart review to have a diagnosis of antiphospholipid syndrome. Of the 13 patients categorized as having clinical SLE, 6 had a history of nephritis; the others were confirmed by their referring physician to have SLE or were receiving immunosuppressive therapy for nonrenal indications related to SLE.
Overall, at the time of cardiac catheterization, patients with SLE were significantly younger compared to controls (median age 49 years versus 70 years; P < 0.001); this difference held true when examining the age distribution of only the patients with significant CAD (Table 3). The mean (IQR) age for the subgroup of SLE patients with CAD was 50 years (IQR 41–59 years) as compared with 73 years (IQR 66–79 years) for the subgroup of controls with CAD. Age was a risk factor for CAD in both the SLE and control groups.
Table 3. Distribution of age in patients with CAD*
The primary indication for catheterization was specified as the evaluation for possible CAD in 88% of the patients with SLE and 92% of the non-SLE controls. For the remaining few subjects, the primary indication was listed as shortness of breath, congestive heart failure, or valvular disease, while the secondary indication was listed as the evaluation for possible CAD. Approximately one-third of both the SLE patients and controls had a history of congestive heart failure or MI. One patient with SLE and 4 of the controls had a history of percutaneous coronary artery intervention, and 2 patients with SLE and 18 controls had a history of coronary artery bypass grafting (CABG). At the time of catheterization, equivalent numbers in the SLE group and the control group were taking cardiovascular medications, including aspirin, statins, beta-blockers, angiotensin-converting enzyme inhibitors, or angiotensin II receptor blockers.
Association between SLE and the presence of CAD on coronary angiogram.
Forty-five (52%) of the SLE patients and 160 (62%) of the controls had obstructive CAD shown on the angiogram (P = 0.11). The distribution of diseased vessels was similar between the 2 groups. The most common site of coronary stenosis for both the patients with SLE and the controls was the left anterior descending artery, and the least common site was the left main artery. There was a higher proportion of patients with SLE and without significant CAD than controls; however, a higher proportion of controls than SLE patients had 3 diseased vessels, although these differences were not statistically significant (Table 4).
Table 4. Severity of CAD defined by the number of diseased vessels*
No. of diseased vessels
SLE (n = 86)
Control (n = 258)
Values are the percentage (number). Overall P = 0.081. CAD = coronary artery disease; SLE = systemic lupus erythematosus.
The unadjusted relationships between baseline variables and obstructive CAD are shown in Table 1. Without covariate adjustment in the logistic regression model, SLE was not associated with an increased risk. However, after adjustment for traditional risk factors, SLE was significantly associated with the presence of CAD on coronary angiogram (Table 5). In the multivariable logistic regression model, other risk factors that were associated with the presence of CAD included a history of MI, age, hyperlipidemia, and male sex. The association of these other risk factors with CAD was observed in both the full model with all candidate variables as well as the reduced model of significant covariates (P = 0.03 in both settings). The only significant interaction was between diabetes mellitus and SLE (P = 0.04), such that diabetes mellitus was associated with the presence of CAD in controls (odds ratio [OR] 1.97 [IQR 1.04–13.71]), but not in patients with SLE (OR 0.40 [IQR 0.10–1.59]).
Table 5. Multivariable logistic regression model for obstructive CAD*
There were a total of 137 deaths over the entire followup period. Unadjusted survival was not significantly different in the 2 groups, with a 1-year survival rate of 83% in the SLE group compared with 92% in the control group; the 5-year survival rates were 72% in the SLE group compared with 69% in the control group (Figure 2A). After adjusting for age, ejection fraction, diabetes mellitus, hyperlipidemia, glomerular filtration rate, noncardiac Charlson comorbidity index, and extent of CAD, SLE was not found to be significantly associated with all-cause mortality (hazard ratio [HR] 1.68 [95% confidence interval (95% CI) 0.98–2.89], P = 0.060) (Figure 2B). However, when the adjusted survival rate was examined at 1 year, it was significantly lower in the SLE group than the control group (P = 0.0004). When adding medications to the adjustment, prednisone taken prior to catheterization was significantly associated with increased all-cause mortality (HR 2.08 [95% CI 1.26–3.45], P = 0.005), while azathioprine appeared to be protective (HR 0.17 [95% CI 0.04–0.73], P = 0.017). Following cardiac catheterization, the 2 groups with obstructive CAD did not differ in the use of either medical management or interventional therapy such as percutaneous angioplasty or CABG (overall P = 0.327).
In SLE, occlusive CAD is presumed to be a major cause of increased cardiovascular morbidity and mortality. However, to date, the evidence for this association has been confined to epidemiologic observations and surrogate measures of CAD. In this select group of patients from a cardiovascular disease database, SLE was shown to be associated with the presence of CAD on coronary angiography, the gold standard for assessing flow-limiting lesions.
The results of this study also suggest there is a relationship between SLE and early development of CAD; however, this association does not necessarily prove causality. It has been hypothesized that the systemic inflammation occurring in SLE triggers atherosclerosis in the coronary vessels (19). Alternatively, the cause may not be SLE per se, but the use of corticosteroids and other immunosuppressive agents that influence CAD risk factors, such as diabetes mellitus or hypertension (20). The extent to which either traditional cardiovascular risk factors (e.g., diabetes mellitus, smoking, hyperlipidemia, family history, and hypertension) or disease-related risk factors (taking medication, systemic inflammation) contribute to the development of CAD in patients with SLE remains unclear. Consistent with previous findings (8), in our cohort, we found that age and hyperlipidemia were associated with an increasing risk of CAD. The results from the SLICC registry for atherosclerosis have shown that there is a high prevalence of several risk factors for CAD in patients with SLE at the time of the initial diagnosis (21). In our study, the rates of traditional risk factors were equivalent in the patients with SLE and controls at the time of catheterization, except for the rates of hypercholesterolemia and diabetes mellitus, which were less in the SLE group. Previous studies, albeit limited, have suggested that diabetes mellitus and SLE do not commonly occur in the same individual (21, 22). In our study, diabetes mellitus was diagnosed in only 14% of the selected patients with SLE.
Despite being 20 years younger and having approximately half the rate of diabetes mellitus compared with the control group, the selected patients with SLE in our study were similar in their severity of CAD and showed comparable mortality rates. As expected, an increased risk of CAD was found to be associated with age. The interaction of age with SLE was not significant, suggesting that the effect of age was the same for patients with SLE and controls. Our matching strategy had 2 advantages. First, we were able to study the extent of CAD in a selected group of patients with SLE that could be compared with the extent of CAD in a control group with typical CAD. This approach avoided the possibility of comparing our SLE group with a younger, atypical control group that had an unusual susceptibility to CAD. Second, our matching strategy also enabled us to confirm that patients with SLE do in fact develop CAD at a much younger age than is usually the case. Our findings therefore support the hypothesis that patients with SLE prematurely develop coronary atherosclerosis (8).
SLE was not significantly associated with an increased risk of all-cause mortality, although a trend toward this result was detected in our selected group. Since the Cox proportional hazards model assumption was violated for SLE, the power to detect a significant difference in all-cause mortality may have been diminished in this case; however, the adjusted survival curves were generated using a stratified model. The 1-year survival rate after cardiac catheterization was significantly lower in the SLE group. This finding could be explained by the SLE patients having more severe illness than the controls at the time of catheterization. However, if the patients with SLE survived past this point, they had lower mortality rates than the controls due to their younger age. The older control group may have had a higher, competing non-CAD mortality risk than the SLE group, resulting in a higher mortality rate than would be expected in a comparable group of patients that were similar in age to the SLE patients.
In our study, the management of cardiovascular disease after cardiac catheterization appeared to be comparable in the SLE and control groups, including therapy with aspirin, statins, and beta-blockers. Another study also found that treatment with cardiovascular medications was similar in SLE patients and controls following cardiac catheterization and percutaneous coronary intervention (21). The SLE patients in this study were more likely to develop an MI and undergo repeat percutaneous intervention during followup when compared with controls.
Our study examined a select group of SLE patients that underwent cardiac catheterization for suspicion of CAD at a single academic center. The results from this study are not meant to be a generalization for all patients with SLE or imply that there is a prevalence of CAD across the entire spectrum of disease. Our study was also not designed to include asymptomatic patients with SLE, although there is evidence that these patients may also have cardiac abnormalities (12). Our filtering strategy was chosen in order to enrich our SLE sample with CAD. It is therefore notable that almost 50% of the patients with SLE who were being evaluated for CAD had no significant CAD (Table 4), suggesting other causes for cardiac and cardiac-like symptoms are frequent in this disease group (vasospasm, microvascular disease, etc.); however, 40% of the control group also had no significantly diseased vessels, and the difference in the proportion of patients with no diseased vessels between the groups was not statistically significant.
Our study does have limitations. Since patients with SLE have a chronic disease and are more closely monitored by physicians than most other patients, our results were subject to surveillance bias leading to earlier referral for cardiac catheterization than the controls. This possibility, along with their younger age, may have allowed less time for patients with SLE to develop more advanced CAD, and may explain why the SLE patients had less multivessel CAD than controls. We emphasize again the selection bias in our study, given that only SLE patients who were referred for a cardiac catheterization due to suspicion of CAD were included in this analysis. Additionally, since the outcomes in our database are only obtained systematically for patients with obstructive CAD, there may be some shortcomings in the documentation of survival among those patients in the database with nonobstructive disease. We attempted to overcome this limitation by ascertaining vital statistics from the National Death Index and Social Security Death Index databases. We were unable to determine if nontraditional risk factors such as hemodialysis were potential confounders that may have been associated with an increased risk of obstructive CAD. It is unclear, therefore, if renal failure contributed to the development of CAD in the patients with SLE. However, we did select a large control group from the same cohort of patients using the same filtering strategy to minimize confounders. Although some data regarding immunosuppressive therapies were obtained, we were unable to analyze this issue further by examining the possible effects of dose or duration of use. Similarly, we did not take into account the presence of antiphospholipid antibodies and their possible contribution to the development of atherosclerotic plaques, since these results were not available for all patients. Finally, the definition of hypercholesterolemia was based on patient history of this diagnosis or statin use. Since cholesterol values were not routinely obtained for these patients, we could not calculate a Framingham Risk Score.
In this study, we demonstrated that SLE was associated with the presence of CAD on coronary angiogram, which is the gold standard for measuring obstructive atherosclerosis, in a select group of patients referred for a cardiac catheterization. We also showed that coronary atherosclerosis developed prematurely in SLE, as demonstrated by the significantly younger age of SLE patients with angiographically defined CAD when compared with controls. Further studies will be required to explore the contribution of genetic risk factors to the premature development of CAD in SLE, as well as to improve our understanding about the outcomes in this population of predominately young women with a chronic inflammatory disease.
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. Kaul 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. Kaul, Rao, Shaw, Ardoin, St.Clair.
Acquisition of data. Kaul, Shaw, Honeycutt.
Analysis and interpretation of data. Kaul, Rao, Shaw, Honeycutt, St.Clair.