Risk factors for coronary heart disease in women with systemic lupus erythematosus: The Toronto Risk Factor Study

Authors


Abstract

Objective

Because women with systemic lupus erythematosus (SLE) are 5–8 times more likely to develop coronary heart disease (CHD) than are women in the general population, we assessed the prevalence of classic risk factors for CHD in women with SLE.

Methods

Consecutive female patients with SLE who were without evidence of CHD and were attending a large lupus clinic in Toronto were studied. The control population was recruited from among age-matched subjects attending a family practice unit for an annual physical examination. The prevalence of classic CHD risk factors and the 10-year risk of a CHD-related event were determined using the Framingham risk assessment formula. Lipid subfractions, other metabolic risk factors, lifestyle variables, and demographic characteristics were also compared between the 2 groups.

Results

We studied 250 SLE patients and 250 controls whose mean ± SD age was 44.8 ± 12 years and 44.3 ± 15 years, respectively. Hypertension and diabetes were significantly more common among the SLE patients. Although the SLE patients had a higher mean number of CHD risk factors per patient, the 10-year risk of a CHD-related event, using the Framingham multiple risk factor assessment, was the same in SLE patients and controls (3.2%). Compared with controls, SLE patients had higher levels of very low-density lipoprotein cholesterol and total triglycerides, and had higher levels of homocysteine despite having higher folate levels. Premature menopause, sedentary lifestyle, and an at-risk body habitus were also more prevalent in SLE patients.

Conclusion

Women with SLE have a range of detectable coronary risk factors that are not fully reflected in the Framingham risk factor formula. These factors are likely to contribute to the loss of protection from CHD that has been observed in SLE.

Since the first description of the bimodal mortality pattern in systemic lupus erythematosus (SLE) (1), coronary heart disease (CHD) has been confirmed as a major cause of morbidity and mortality in patients with SLE. In several SLE cohorts, the prevalence of clinical CHD has ranged from 6% to 10% (2– 5). Compared with women in the general population, women with SLE have been found to be 5–8 times more likely to develop CHD, with this risk being particularly marked in women younger than age 55 years (4, 6).

The mechanism of accelerated atherosclerosis in SLE is unclear. It is likely to be due to a combination of factors encountered in this patient population and could be related to the disease itself, to the effects of therapy with corticosteroids, and/or to conventional risk factors. Indeed, these first 2 factors may play a role by having an adverse influence on conventional risk factors such as hyperlipidemia and hypertension. Several studies have noted that dyslipoproteinemia occurs in conjunction with active disease and is characterized by elevated levels of triglycerides (TGs) and very low-density lipoprotein (VLDL) cholesterol, as well as by reduced levels of high-density lipoprotein (HDL) cholesterol and lower frequency of apolipoprotein A-1 (Apo A-1) (7, 8). Changes in lipid levels associated with steroid therapy have also been reported, and there is a correlation between the steroid dose and total cholesterol levels (9, 10). Furthermore, steroid-treated patients also have higher levels of TGs and low-density lipoprotein (LDL) cholesterol, and a greater frequency of Apo B (7, 11). Increases in lipoprotein(a), an independent risk factor for CHD (12), have also been found in patients with SLE, and in one study, this occurred in 56% of patients compared with 30% of control subjects (13). Change in the plasma homocysteine level is also a risk factor for thrombotic events in SLE (14). Thus, inflammatory, metabolic, and other factors may all interact to increase the risk of CHD in this patient population.

A small number of studies have attempted to identify the risk factors for CHD in lupus by comparing patients who have SLE and documented CHD with individuals who do not develop CHD (2–4). Elevated levels of total cholesterol and older age at diagnosis of SLE are the only 2 risk factors consistently identified by all studies. Such an approach is limited by the small number of events (<55) in each cohort. It also does not help to distinguish which risk factors are overrepresented in a population of SLE patients. In an uncontrolled cross-sectional study, Petri et al studied the prevalence of selected CHD risk factors in the Johns Hopkins lupus cohort, and at least 3 CHD risk factors were identified in 53% of patients (15). The most prevalent factors were sedentary lifestyle (70%), obesity (56%), hypercholesterolemia (56%), and having ever smoked (56%). Since no control group was studied, the contribution of background population risks was not addressed. In addition, only total cholesterol was measured, and more detailed lipid analysis was not performed. Other studies on individual risk factors have included small, highly selected subgroups of patients. Any SLE population, however, varies with regard to organ involvement, disease severity, and treatment. A comparison of the predominant risk factor profiles seen in SLE patients and in the general population is therefore important to begin to determine why these patients have such a markedly increased risk of CHD.

Women with SLE may have an increased number of recognized risk factors per patient. Alternatively, SLE may itself be a risk factor for CHD, independent of conventional risk factors. This may be related to an inherited genetic predisposition for the 2 conditions or to hitherto unidentified factors related to the inflammatory/ immunologic nature of the disease itself. The aim of this study was to test the hypothesis that certain individual risk factors for CHD occur at a greater frequency in women with SLE. We also aimed to calculate the 10-year risk of CHD in SLE patients, to explore whether the risk of CHD, in accordance with the classic Framingham risk factors for CHD, differed in the SLE patients as compared with women in the general population.

PATIENTS AND METHODS

Recruitment.

Ethics approval for this study was obtained from the University of Toronto Health Network Research Ethics Board. Informed written consent was obtained from all participating patients and controls. Patients and controls were recruited from May 1998 to June 2000.

SLE patients.

Patients with SLE have been followed up prospectively at the University of Toronto Lupus Clinic since 1970. Clinical and laboratory information, including the details of therapy, is collected at 2–6-month intervals and stored on a computer database. All patients are assessed by the directors of the clinic (MBU and DDG) or by a clinical fellow trained by them. Two hundred forty-one of the participating patients fulfilled ≥4 of the 1982 American College of Rheumatology classification criteria for SLE (16), and 9 patients in this study had 3 of the criteria and a typical lesion of SLE, such as that identified on renal or skin biopsy (17). Patients in this cohort were similar to those reported from other large centers. Women with SLE followed up consecutively in the Toronto clinic were recruited and invited to attend the clinic in the morning immediately following an overnight fast and avoidance of alcohol for more than 48 hours. We excluded male patients because they constituted only 10% of the entire cohort. Other exclusions were a history of myocardial infarction, angina pectoris, or peripheral vascular disease, current pregnancy or delivery within the past 3 months, current acute or chronic infections, current malignancy, and history of chronic alcohol abuse.

Controls.

A group of age-matched women were recruited as control subjects from the Toronto Western Hospital Family Practice Unit. Subjects were successive patients attending the family practice unit for a routine annual physical examination. In addition to the above exclusions applied to SLE patients, we also excluded women attending the family practice who had a history of SLE or other chronic inflammatory arthropathies. Those women who were receiving treatment with corticosteroids or antimalarial drugs at the time of this study or within the past 6 months, and patients with known renal impairment (>110 μmoles/liter) or significant proteinuria (>1+ on dipstick analysis or >500 mg/day) were also excluded.

Clinical assessments.

Each SLE patient underwent a complete history review and physical examination according to a standard protocol. This assessment included basic demographic data, organ-specific disease-related symptoms, and physical findings. Overall disease activity at presentation to the clinic and at the time of the study was derived by calculation of the SLE Disease Activity Index (SLEDAI) (18). In addition, the average mean SLEDAI score, which reflects the extent of disease activity over time, was calculated for each patient (19). Current therapy was also noted, including the current and mean dose of prednisone since the last visit, as well as the use of antimalarial and immunosuppressive medications. Data were also collected on the following factors: blood pressure and current hypertension, diabetes and specific therapy, smoking history along with current smoking status and number of pack-years, body mass index (BMI) in kg/ m2, waist:hip ratio (a measure of the distribution of body fat), and recent change in body weight (including magnitude). Information on a history of thyroid disease or current thyroid replacement therapy, menstrual status, use of oral contraceptives or hormone replacement therapy, additional medications including vitamin supplements and alternative medications, and average weekly alcohol consumption was also recorded.

In addition, a trained interviewer administered 2 questionnaires. One questionnaire sought to ascertain any family history of premature CHD and its associated risk factors, as well as any family history of SLE. The second questionnaire assessed the level of physical activity using the Physical Activity Index (20).

Laboratory methods.

Routine clinical samples.

As part of the routine laboratory assessment in the clinic, the following parameters were measured (for normal limits of relevance, see Table 3): hemoglobin levels, leukocyte and platelet counts, serum creatinine levels, urine microscopy, fasting plasma glucose levels, antibodies to DNA, complement profile, antibodies to cardiolipin (IgG >23 IU), and the partial thromboplastin time (>32 seconds).

Table 3. Comparison of demographic and clinical characteristics between the patients with systemic lupus erythematosus (SLE) and controls*
CharacteristicSLE (n = 250)Controls (n = 250)RR (95% CI)P
  • *

    RR = relative risk; 95% CI = 95% confidence interval; NS = not significant; NA = not applicable.

Age, mean ± SD years44.5 ± 1244.1 ± 14NS
Race, %    
 White/black77/1088/30.0051
 Chinese/other6/75/4 
Education, % below college39162.45 (1.77–3.39)0.001
Menopause    
 Age at onset, mean ± SD years45.4 ± 6.249.3 ± 4.00.0001
 % postmenopausal38191.98 (1.47–2.67)0.0001
% taking oral contraceptive5130.37 (0.19–0.71)0.0011
% with hyperthyroidism1.60.82.00 (0.37–10.62)NS
% with hypothyroidism11.95.22.29 (1.22–4.29)0.0196
Serum creatinine    
 Mean ± SD gm/ liter87.1 ± 99.670.3 ± 100.0087
 % with >110 gm/liter90NA0.001
Hemoglobin    
 Mean ± SD mg/ dl125.6 ± 14132.8 ± 90.0001
 % with <115 mg/dl1635.06 (2.4–10.6)<0.0001
% with sedentary lifestyle1591.65 (1.02–2.39)0.0404
Waist:hip ratio    
 Mean ± SD0.80 ± 0.060.78 ± 0.06 0.0001
 Abnormal >0.85, no. (%)39 (15.6)23 (9.2)1.70 (1.04–2.75)0.0299
Body mass index    
 Mean ± SD kg/ m225.1 ± 6.325.6 ± 5.8NS
 % with >27 kg/ m228300.92 (0.69–2.05)NS

Lipids.

Blood was obtained for laboratory analyses following a 14- hour fast and 48-hour abstinence from alcohol. Plasma and serum were separated immediately, and ultracentrifugation was started on the day of collection. The plasma and its subfractions were then aliquoted and stored at −70°C until analyzed in batches. (Table 1 defines the target values for the general population, as well as the normal limit for fasting glucose.) Levels of total cholesterol (high total cholesterol defined as >5.2 mmoles/liter) (21) and TGs were measured in plasma using commercially available assays (Boehringer Mannheim kits 236691 and 450032, respectively; Indianapolis, IN). Lipoproteins were separated from plasma into subfractions as follows: Svedberg flotation (Sf) rates Sf60–400 (VLDL), Sf20–60 (VLDL), Sf12–20 (intermediate-density lipoprotein [IDL]), and Sf<12 (LDL and HDL) by ultracentrifuging as previously described (22). HDL and LDL were separated from each other by manganese chloride/heparin precipitation of LDL from the Sf<12 subfraction. The concentration of HDL was calculated by subtraction of the LDL value from the Sf<12 subfraction (see Table 1 for target values for LDL and HDL). For the purpose of comparison with the National Cholesterol Education Program (NCEP) target values (21), the LDL cholesterol level was also calculated using the Friedewald formula (23) in those subjects whose plasma TG level was <4.5 mmoles/liter (data not shown). Lipoprotein(a) was measured by enzyme-linked immunosorbent assay (kit 610221; Biopool, Hamilton, Ontario, Canada) (Table 1). LDL size was determined by nondenaturing polyacrylamide gel electrophoresis of whole plasma using 2% and 16% gradient gels (24) (Table 1). The Apo E phenotype was assessed by isoelectric focusing (25).

Table 1. Categorical classification of the risk factors for coronary heart disease (CHD) assessed in patients and controls*
Risk factorDefinition
  • *

    The first 7 risk factors were used in the comparison of the numbers of risk factors per person per group. See refs. 21, 45 and 46. MI = myocardial infarction; LDL = low-density lipoprotein; HDL = high- density lipoprotein; BMI = body mass index; TGs = triglycerides; Lp(a) = lipoprotein (a); Apo E = apolipoprotein E.

  • Age was not counted as a risk factor because this factor was matched with the control population.

  • High HDL is a negative factor which, if present, leads to the subtraction of one risk factor.

  • §

    Cole DEC: personal communication.

Positive family history of CHDDefinite MI or sudden death in a first-degree relative: male, age <55 years or female, age <65 years
Diabetes mellitusFasting plasma glucose >7.0 mmoles/liter, or current diabetic therapy
Current smokingCurrent smoking of ≥1 cigarettes per day
HypertensionBlood pressure ≥140 mm Hg systolic or 90 mm Hg diastolic on repeated measurements, or taking antihypertensive medication
Raised LDL cholesterolFasting plasma LDL cholesterol >3.4 mmoles/liter
Low HDL cholesterolPlasma HDL cholesterol <0.9 mmoles/liter
High HDL cholesterolPlasma HDL cholesterol >1.6 mmoles/liter
ObesityBMI >27.0 kg/ m2
Increased waist:hip ratio>0.85 (female)
Sedentary lifestylePhysical Activity Index < 28.0
Raised TGsFasting plasma TGs >2.3 mmoles/liter
Raised Lp(a)Plasma Lp(a) >30 mg/dl
Raised homocysteine§Plasma homocysteine >15.0 μ moles/liter
Small LDL phenotypeType B; diameter <25.8 nm
Apo E phenotypePresence of one or two ε4 alleles

Homocysteine.

Plasma was separated immediately and frozen at −70°C until analyzed in batches. Total plasma homocysteine was measured by high-performance liquid chromatography with pulsed integrated amperometry as previously described (26). Since homocysteine levels are inversely correlated with folate levels, plasma folate and red blood cell folate concentrations were also measured.

Statistical analysis.

The prevalence of each individual risk factor in SLE patients was compared with that in controls. Categorical classification of the Framingham risk factors was determined using currently accepted definitions (Table 1). For those factors in which no defined level of risk is available, e.g., for VLDL or IDL, the mean levels were compared between groups. The prevalence of each factor was then compared between groups using t-tests or chi-square tests. We also compared the mean number of risk factors per patient with that in controls, by counting each standard risk factor as defined by the NCEP 1993 guidelines (21) (Table 1); the maximum number of risk factors was 6 per person.

Using a reverse Bonferroni correction for 10 independent outcomes, we calculated that a sample size of 250 per group would provide 80% power to detect a relative risk of 1.75 for patients versus controls (α = 0.005). This sample size is conservative given the nature of the correction. This sample size also provided >99% power to detect a difference of 2 in the total number of risk factors between groups.

RESULTS

Lifestyle and demographic factors.

We studied 250 patients and 250 controls. The patients with SLE had a mean ± SD disease duration of 13.7 ± 9.7 years. The mean (±SD) SLEDAI at the time of the study was 4.5 (±4.5), while at presentation, the disease activity score was 9.0 (±7.5). The adjusted mean SLEDAI score to the time of the study was 5.1 (±3.6). Renal disease had occurred in 44.8% of the SLE patients, and nephrotic syndrome had occurred in 5 patients at the time of the study. Fifty-four percent of patients were taking steroids at the time of the study, with a mean dose of 12.1 mg/day and a cumulative steroid dose of 12.4 grams (Table 2).

Table 2. Demographic and clinical characteristics of the 250 patients with systemic lupus erythematosus*
VariableValue
  • *

    Except where indicated otherwise, values are the mean ± SD. SLEDAI = Systemic Lupus Erythematosus Disease Activity Index.

Age at diagnosis, years30.9 ± 11.3
Disease duration, years13.7 ± 9.7
SLEDAI 
 At presentation9.0 ± 7.5
 At study4.5 ± 4.5
 Average mean score5.1 ± 3.6
Renal disease, no. (%) 
 At study111 (44.8)
 Ever176 (71.0)
Nephrotic syndrome, no. (%) 
 At study5 (2)
 Ever25 (10)
Steroids 
 No. (%) taking at study136 (54.8)
 No. (%) ever taking198 (79.8)
 Dosage at study (prednisone equivalent), mg/day12.1 ± 9.2
Cumulative lifetime dose (prednisone equivalent), gm12.4 ± 7.4
No. (%) currently taking antimalarial therapy132 (53)
No. (%) currently taking immunosuppressive therapy136 (55)

Patients and controls were matched for age (Table 3). There was a higher proportion of blacks in the patient group (10% versus 3% of controls) and a lower proportion of whites (77% versus 88% of controls) (P < 0.01). The patient group also had a lower educational attainment, with 39% of SLE patients compared with 16% of controls having reached a below- college level of education (P < 0.01). As can also be seen in Table 3, 9% of SLE patients had a raised serum creatinine level, and 5% of SLE patients were receiving lipid-lowering therapy (data not shown) at the time of the study. More control subjects were taking the combined oral contraceptive pill (13% versus 5% of patients; P < 0.01). Despite being matched for age, SLE patients were more likely to be postmenopausal (38% versus 19%; P < 0.01) and the mean ± SD age at menopause was significantly younger in the women with SLE (45.4 ± 6.2 years versus 49.3 ± 4 years; P < 0.01).

We also found that 15% of women with SLE compared with 9% of controls had a sedentary lifestyle, as determined by the Physical Activity Index (P = 0.040). The BMI was comparable in both groups, with a similar proportion having a BMI >27.0 kg/m2. In contrast, the distribution of body fat assessed by the waist:hip ratio was different. SLE patients had a mean ± SD waist:hip ratio of 0.80 ± 0.06 compared with 0.78 ± 0.06 in controls (P < 0.01).

Framingham risk factor profile.

We compared the prevalence of classic Framingham risk factors in the SLE and control groups (Table 4). Women with SLE were significantly more likely to have hypertension (33% versus 13%; P < 0.01) and diabetes mellitus (5% versus 1%; P < 0.01). Of the 12 SLE patients with diabetes, 8 were insulin dependent; neither of the 2 controls with diabetes were taking insulin. There was also a trend toward lower HDL cholesterol levels in the SLE group, which did not reach statistical significance (P = 0.332). We dichotomized each risk factor according to the first 7 definitions in Table 1. The mean (±SD) number of risk factors per person was 1.01 (±1.0) in SLE patients compared with 0.72 (±1.0) in controls (P = 0.0014).

Table 4. Framingham risk factors in patients with systemic lupus erythematosus (SLE) and controls*
Risk factorSLE (n = 250)Controls (n = 250)RR (95% CI)P
  • *

    Except where indicated otherwise, values are the no. (%) of subjects.

  • RR = relative risk; 95% CI = 95% confidence interval; NS = not significant; HDL = high-density lipoprotein; CHD = coronary heart disease; NA = not applicable.

Hypertension83 (33)32 (13)2.59 (1.79–3.75)0.001
Hypercholesterolemia84 (34)91 (36)0.92 (0.73–1.17)NS
Low HDL cholesterol level of < 0.9 mmoles/liter33 (13)26 (10)1.27 (0.78–2.06)NS
Current smoker42 (17)49 (20)0.86 (0.59–1.24)NS
Diabetes mellitus12 (5)2 (1)6.00 (1.36–26.53)0.0066
Family history of premature CHD49 (20)42 (17)1.16 (0.80–1.69)NS
Mean ± SD no. of risk factors per person1.01 ± 1.00.72 ± 1.0NA0.0014
Mean 10-year risk for CHD-related event, %3.23.2NANS

Using the 10-year risk calculation based on the number of Framingham risk factors present (27), there was no significant difference in the mean 10-year risk of a cardiac event in SLE patients compared with controls (Table 4).

Levels of lipid subfractions, lipoprotein(a), and homocysteine.

Thirteen of the SLE patients (5%) compared with only 1 of the control patients (0.4%) were taking lipid-lowering therapy at the time of the study (P = 0.0011). For the measurement of lipid subfractions, we excluded the patients receiving lipid-lowering therapy. The mean total cholesterol and LDL cholesterol levels were similar in both groups (Table 5). Women with SLE did, however, have significantly higher VLDL cholesterol and total TG levels compared with those in the controls. The mean LDL size and the percentage of patients with small, dense LDL (<25.8 nm in diameter) did not differ between the patients and controls. Similarly, the mean lipoprotein(a) levels and percentages of white patients with the Apo E-4 or E-2 phenotype were not different. The mean plasma homocysteine levels were significantly higher in SLE patients, and the proportion with a homocysteine level >15 μmoles/liter was 11.6% of patients compared with 0.8% of controls (P < 0.01). These differences in homocysteine levels occurred despite SLE patients having a higher mean red blood cell folate level.

Table 5. Levels and frequency of lipid subfractions, Lp(a), the Apo E phenotype, and homocysteine in patients with systemic lupus erythematosus (SLE) and controls*
Risk factorSLEControlsRR (95% CI)P
ValueNo. of subjectsValueNo. of subjects
  • *

    Except where indicated otherwise, values are the mean ± SD. RR = relative risk; 95% CI = 95% confidence interval; NS = not significant; LDL = low-density lipoprotein; VLDL = very low-density lipoprotein; HDL = high-density lipoprotein; Lp(a) = lipoprotein(a); Apo = apolipoprotein; RBC = red blood cell.

  • White subjects only.

Total cholesterol      
 Level, mmoles/ liter4.69 ± 1.122374.80 ± 0.95249NS
 % with >6.7 mmoles/liter4.62.02.30 (0.81–6.53)NS
LDL cholesterol      
 Level, mmoles/ liter2.78 ± 0.972372.92 ± 0.84249NS
 % with >4.6 mmoles/liter3.82.0 1.88 (0.64 –5.54)NS
VLDL cholesterol, mmoles/ liter0.45 ± 0.342370.38 ± 0.272490.0073
Total triglycerides      
 Level, mmoles/ liter1.41 ± 0.882371.19 ± 0.672490.0021
 % with >2.53 mmoles/liter6.34.01.57 (0.72–3.42)NS
HDL triglycerides, mmoles/ liter0.28 ± 0.102370.27 ± 0.10249NS
Lp(a)      
 Level, mg/dl15.3 ± 16.522914.4 ± 16.5235NS
 % with >30 mg/ dl20141.43 (0.95–2.15)NS
LDL size      
 Size, nm25.3 ± 1.023725.3 ± 1.0249NS
 % with <25.8 nm in diameter72721.00 (0.89–1.12)NS
Apo E-4, %30.723124.72231.25 (0.92–1.68)NS
Apo E-2, %11.315.70.72 (0.45–1.15)NS
Homocysteine      
 Level, μmoles/ liter9.76 ± 4.42506.53 ± 3.12490.0001
 % with >15 μ moles/liter11.60.814.4 (3.5–59.9)0.001
RBC folate, nmoles/ liter1,124 ± 5492501,020 ± 4172503.50 (1.77–6.91)0.0175

DISCUSSION

As the prognosis of SLE has improved, accelerated atherosclerosis and premature CHD have become increasingly recognized as significant late complications. Several studies have confirmed that patients with SLE have a 5–8-fold increased risk of CHD-related events compared with the general population (4, 5, 28). To date, there has been no detailed study of the CHD risk factor profile in a large lupus cohort as compared with a local control population. We set out to determine if patients with SLE had a greater number of risk factors per patient, and which particular risk factors were significantly overrepresented in an SLE cohort. Our lupus clinic has received primary, secondary, and tertiary referrals of patients with a variety of disease manifestations and levels of activity. Our cohort is comparable with those of several other large clinic series, both in terms of clinical features and in the reported rates of CHD (29). Our findings are therefore generalizable to these settings.

We found differences between the SLE patients and controls in the 3 areas addressed, namely, demographic and lifestyle factors, traditional Framingham risk factors, and several other metabolic factors. Our cohort–control design primarily matched the patients and controls by age. There was a small, but significant, difference in the racial distribution of the cohort, with fewer white subjects in the SLE group. In addition, women with SLE were more likely to have finished their education below the college level. This may suggest a potential bias in the population of women selected as controls, since better-educated women may be more likely to attend an annual physical examination, and this may have an influence on several of the risk factors. It is interesting, however, that several factors, such as the BMI, family history of CHD, and smoking rates, all of which may be of relevance in influencing screening behavior, did not differ between the groups. Nevertheless, the level of education may have influenced other lifestyle factors such as physical activity, but they are unlikely to have had a major influence on the other metabolic measures.

The frequency of several risk factors in our controls was comparable with that reported in an epidemiologic survey of cardiovascular disease risk factors in Canadian adults (30). We also found that SLE patients had a more sedentary lifestyle, a higher prevalence of hypothyroidism, and an earlier menopause. SLE patients have an increased prevalence of other autoimmune diseases, especially hypothyroidism, and as such, this represents a population characteristic (31). All of our patients had stable disease and were receiving adequate doses of thyroid replacement therapy in the period prior to the study. The earlier onset of menopause in SLE patients is also of particular interest. SLE patients had their menopause an average of 4 years earlier than did the controls (at a mean age of 45.4 years versus 49.3 years). This means that the CHD risk associated with menopausal status will occur earlier in SLE patients. If hormone replacement therapy is ineffective as a primary prevention strategy, as has been recently suggested (32), then patients with SLE may have a longer period “at risk” of CHD in their lifetime compared with their age-matched peers. Several studies have suggested that the mean age at the first occurrence of a CHD-related event in women with SLE is 48–50 years (5). The peak incidence of CHD-related events in lupus is therefore in the perimenopausal and immediate postmenopausal period. Increased vigilance and screening for risk factors may therefore be of particular importance in this age group of SLE patients.

A more sedentary lifestyle has been noted previously in a cohort that was ∼50% African American (3). Our study confirms this finding in a predominantly white cohort. Physical inactivity is likely to be the result of multiple factors and can be related to disease variables such as fatigue, arthritis, and reduced aerobic capacity (3, 33). There is clearly the potential to modify this aspect of a patient's lifestyle, although sustaining prolonged benefits from aerobic training programs in SLE patients has been proven to be difficult (34).

With regard to the Framingham risk factors currently utilized in assessing the risk of CHD for primary prevention purposes, we found that our SLE patients were significantly more likely to have hypertension and diabetes mellitus. Our definition of diabetes was based on the medical history and fasting blood glucose estimation. The addition of a standard glucose tolerance test would probably have increased the overall prevalence of diabetes in both the patients and the controls. The relative difference would likely be maintained.

These 2 factors (hypertension and diabetes) contributed to the overall increased mean number of Framingham factors present in the SLE patients as compared with the controls. Despite this, the 10-year risk estimate was similar in both groups (3.2%). This is partly explained by the inclusion of current blood pressure readings in such equations, which will be modified by therapy. Also, the low absolute prevalence of diabetes mellitus would have little impact across the population as a whole. What is clear, however, is that the standard estimate of risk used for primary prevention in the general population fails to reflect the 5–8-fold overall increased risk of CHD observed in SLE patients. We have previously shown that patients with SLE and CHD had one less risk factor than did a control cohort of individuals with accelerated CHD but without SLE (35). This finding confirms the results of a prior study by Esdaile et al (28), who noted that even after adjusting for the presence of Framingham risk factors at baseline, the risk of CHD in lupus patients was still increased by a factor of 8 compared with that in the population controls. As such, the Framingham formula should not be used alone for primary prevention purposes in patients with SLE.

We also observed differences in the levels of lipid subfractions between patients and controls. Whereas the total cholesterol and HDL cholesterol levels were comparable in the 2 cohorts, SLE patients had higher levels of VLDL cholesterol and TGs. Several previous studies have described a dyslipoproteinemia associated with active SLE that is characterized by increased TGs, increased VLDL cholesterol, and reduced HDL cholesterol (7, 8). Steroid therapy is also known to increase the levels of LDL cholesterol and TGs (7, 9). Others have noted a lipid-lowering effect of antimalarial drugs that may also be of relevance in this setting (36, 37).

The study of selected populations is important in understanding the mechanisms of lipid metabolism in inflammatory disorders such as lupus. It is, however, only in a cohort study such as ours that the overall effect of the disease, its therapy, and other factors can be studied to identify which lipid subfractions are most significantly affected in SLE. Our study suggests that although the levels of total cholesterol and HDL cholesterol are comparable between the patient and control populations, other lipid fractions are more likely to show abnormal patterns and may warrant investigation and intervention. The changes in VLDL cholesterol and TGs may both be related to the increased waist:hip ratio observed in our patients. They may also reflect a combination of ongoing inflammation and the influence of steroid therapy. It is suggested that TGs and VLDL cholesterol may be a more important risk factor in women than in men (38, 39). Since SLE is predominant in women, these factors may assume a particularly important role in this context.

Increased homocysteine levels were also found in 11.6% of SLE patients compared with only 0.8% of controls. This occurred despite a higher overall red blood cell folate level in the SLE group. Homocysteine has been found to be an important risk factor for thrombosis in lupus (14), although in the general population, the primary role of homocysteine as an independent risk factor for CHD remains somewhat controversial (40, 41). In SLE, in which accelerated atherosclerosis is a recognized problem and in which other factors, such as antiphospholipid antibodies, also increase the thrombotic tendency, homocysteine may, similar to VLDL cholesterol and TGs, have a more significant impact on the CHD risk.

This is the first large-scale study of its kind. We have noted significant differences in the risk factor profile of patients with SLE compared with healthy controls in the areas of demographics, lifestyle, traditional risk factors, and more novel CHD risk factors. Overall, patients with SLE have more of these risk factors than do individuals in the general population. We and others have previously demonstrated that sustained hypercholesterolemia and hypertension are risk factors for CHD-related events and mortality in patients with SLE (42, 43). We believe that many of these risk factors contribute to the increased risk of CHD observed in SLE. It will, however, require a prospective study to determine the relative contribution of each of these factors to the incidence of CHD in SLE. Such a study is currently underway through the Systemic Lupus International Collaborating Clinics.

Many of the factors identified are potentially amenable to screening and intervention. They may also be amenable to better long-term control of the underlying inflammatory process. As such, careful control of disease activity and a proactive policy of screening patients with SLE for potentially modifiable risk factors seem appropriate (44). We also conclude that reliance on the standard Framingham risk equations in this population is not appropriate, because the Framingham assessments do not accurately reflect the CHD risk in SLE. Further studies to understand the additional contribution of the inflammatory process to CHD risk in lupus are needed, as are studies to accurately predict the future risk of CHD in this high-risk population. In addition, our results suggest a firm rationale for initiating intense therapy to prevent CHD-related outcomes in patients with SLE.

Acknowledgements

The authors acknowledge the helpful contributions of Joanna Cichon and Anne MacKinnon for patient recruitment and study coordination, and Dr. David Cole for determining the homocysteine levels in the patients and control subjects.

Ancillary