Systemic lupus erythematosus (SLE) is associated with a dramatically increased risk of atherosclerotic coronary artery disease (CAD), such that women with SLE aged 34 to 44 years are more than 50 times more likely to develop myocardial infarction (MI) than are age-matched peers 1. Traditional risk factors measured at baseline, as defined in the Framingham model, do not fully account for this increased risk 2.

In the general population, it has been shown that recent and remote blood pressure (BP) predict cardiovascular risk incrementally over current BP 3. In an inception cohort of patients with SLE, those with sustained hypercholesterolemia in the first 3 years of their disease were shown to be at greatest risk of cardiovascular events over 12 to 14 years of follow-up, compared with those who had persistently normal cholesterol or "variable" hypercholesterolemia in the first 3 years of disease 4. We previously showed that both TC and BP take a variable course in patients with SLE and that almost half of the total variance over time in both TC and BP is seen within rather than between patients 5. These findings suggest that the risk of future coronary events might be best quantified by using strategies that take into account multiple measurements of risk factors over time.

In this prospective proof-of-concept cohort study, we sought to compare "summary measures" of cumulative exposure to TC, SBP, and DBP, with single-point-in-time measurements of these risk factors (both recent and remote), in terms of ability to quantify CAD risk.

Characteristics of the patients in this study are presented in Table 1. Overall, the BP dataset contained 991 patients, whereas the TC dataset comprised 956 patients. In each dataset, patients were mostly female (88%) and mostly Caucasian (70%). A total of 94 coronary events occurred (75 angina, 25 MI, and two sudden cardiac deaths; eight had both angina and MI) in the BP dataset, whereas the TC dataset contained 86 coronary events (71 angina, 20 MI, and two sudden cardiac deaths; seven had both angina and MI). The mean ± SD age and disease duration at entry into the study were very similar for both datasets (37.1 ± 14.0 and 6.1 ± 7.9 years, respectively, for the BP dataset). Likewise, mean ± SD SLEDAI-2K score and Systemic Lupus Erythematosus International Collaborating Clinics/American College of Rheumatology Damage Index (SLICC/ACR-DI) at study entry were similar in the two datasets (9.2 ± 7.5 and 0.5 ± 1.2, respectively, for the BP dataset), indicating moderate disease activity and minimal disease-related damage 10. For each dataset, at entry into the study, more than 60% of patients were taking corticosteroids, whereas approximately 40% were taking antimalarials, and 25% were taking immunosuppressives. In each dataset, at the start of the study, approximately 22% of patients were hypertensive, 40% had hypercholesterolemia, 3% had diabetes, and 19% were smokers. At study start, in each dataset, 25% were taking antihypertensives, and 5% were taking lipid-lowering medications.

Through the use of proportional hazards regression modeling in a large sample of more than 950 patients, in whom collectively more than 18, 000 serial measurements of TC and BP were taken over a mean duration of 6.3 years, we were able to demonstrate and quantify the association between several important risk factors and CAD events in SLE.

Foremost, this study highlights the important role of traditional risk factors such as elevated TC and BP in SLE-related CAD and demonstrates a continuum of risk associated with these variables across the range of possible values they may assume. Previous studies have shown that traditional risk factors, such as hypercholesterolemia and hypertension, account for only a small proportion of the increased risk of CAD in SLE. However, in these studies, TC and BP were measured at baseline, in keeping with the premise of the Framingham model. Here we have shown that this remote measure of exposure to TC, SBP, or DBP is not predictive of CAD outcome among patients with SLE. Furthermore, recent measurements of TC and BP, which are also taken at a single point in time, are not able to quantify CAD risk to the same degree as "summary measures, " which capture cumulative exposure to these risk factors over the course of disease. We have previously shown that, unlike the general population, wherein TC and BP "track" over time, in patients with SLE, these risk factors take a dynamic course, varying because of changes in disease activity and treatment 5.

In this study, cumulative exposure to TC, SBP, and DBP was measured by using three summary measures, arithmetic mean, time-adjusted mean (AM), and area-under-the-curve (AUC). Time-dependent proportional hazards regression models were then applied to these summary measures. In this way, a sense of cumulative exposure was captured in two ways: first, in the form of a summary measure, and second, by determining the hazard related to this summary measure for an interval just before each and every sequential visit. In these time-dependent models, a sense of cumulative exposure to other covariates, including disease activity score, corticosteroids, and antimalarials, was also captured through the use of serial measurements of these variables, updated from one visit to the next.

For each of TC, SBP, and DBP, mean and AM summary measures were significantly predictive of a CAD event. In addition, the HR and accompanying P value of mean summary measures was the same as for AM summary measures in the case of each of TC, SBP, and DBP. This is likely related to the fact that overall, in this context, in which measurements were taken frequently, mean and AM values were very similar for each patient. However, when applied to a setting in which measurements are more irregular and infrequent, the AM, which is weighted for the interval between measurements, may be expected more accurately to reflect cumulative risk exposure and hence the overall risk of CAD. In the case of TC, a hazard ratio of 2.07 (P = 0.003) means that for every 1 mmol/L increase in mean (or AM) plasma TC level, the hazard of a CAD outcome increases 2.07-fold. In the case of SBP, a hazard ratio of 1.025 means that for every 1-mm Hg increase in mean SBP, the hazard of CAD outcome increases 1.025-fold.

Likewise, in the case of DBP, a hazard ratio of 1.04 means that for every 1-mm Hg increase in mean (or AM) DBP, the hazard of CAD outcome increases 1.04-fold. SBP and DBP are highly correlated, and, based on our analyses, either one or the other may be used to quantify CAD risk in patients with SLE.

AUC is very closely tied to length of follow-up, and for any given variable may only become larger over time. As such, it does not provide a sense of rise and fall in the variable of interest. Furthermore, it is not measured in the original units of the variable from which it is derived. These reasons may underlie the lack of association between AUC measures and CAD outcomes among SLE patients in our study.

In this proof-of-concept study, our chosen lipid marker of CAD risk was TC. In general, low-density lipoprotein cholesterol (LDL-C) is deemed the primary target of lipid-lowering therapy 11. In the Toronto lupus cohort, measurement of nonfasting TC has been routine practice at every visit since 1975. However, measurement of lipid and lipoprotein subfractions is a more-recent addition to the data-collection protocol and performed only once yearly because of the need for a fasting sample. To derive summary measures and test their ability to quantify CAD risk, we required a very large dataset inclusive of a relatively large number of CAD outcomes. The TC and BP datasets fulfilled these methodologic requirements. In future, summary measures derived in this study may be applied to other risk factors, such as LDL-C.

How many measurements are enough to provide a valid summary measure, and how often should these measurements be taken in patients with SLE? In this study, the average of first two TC (or SBP or DBP) measurements was not significantly predictive of CAD outcome. Therefore, ideally three or more serial measurements should be sought. Although in this study, the mean gap between measurements used to calculate summary measures was 4.3 months, patients in whom the gap between one or more serial measurements exceeded 18 months were not included in the analyses. Further studies are required to determine the optimal number and frequency of measurements of TC and BP in evaluating CAD risk in SLE.

This study has provided several important insights into the role of demographic-, disease-, and treatment-related variables in SLE-related CAD. The most noteworthy are the increased risk of CAD with increasing disease activity and corticosteroid use, and the protective effect of antimalarials.

We found that for every unit increase in recent SLEDAI-2K disease activity score, the risk of CAD event by the time of the subsequent visit increased almost 10%. An increase in SLEDAI-2K score of 4 or more is generally deemed clinically significant 12. This means that a minimum clinically significant increase in recent disease activity score is associated with approximately 46% increase in risk of CAD in the interval between sequential visits. Ibanez et al. 13 previously showed that for every unit increase in the time-adjusted mean SLEDAI-2K score (AMS), the hazard of a CAD outcome increases 1.08-fold. Collectively, these associations highlight the underpinning role of inflammation in SLE-related CAD.

Our patients with CAD events had greater disease activity at baseline and during follow-up, manifest in a multitude of organ systems, including musculoskeletal, cutaneous, renal, neural, vascular, and serosal. As the overall prevalence of chronic renal impairment was low and did not differ among groups with and without CAD, the association between disease activity and events seen in this study cannot be attributed to nephritis or renal impairment alone.

In studies of coronary risk factors in SLE, it is often difficult to tease apart the effect of corticosteroids from disease activity and traditional cardiac risk factors. In previous studies, longer duration of steroid use has been shown to be an independent risk factor for CAD in SLE 114. With a retrospective chart review method, Karp et al. 15 showed that a 10-mg increase in the average daily prednisone-equivalent dose in the preceding year is independently associated with a 16% increase in the estimated 2-year CAD risk. Here, we quantified the CAD risk in patients with SLE by using prospectively collected data. Patients with CAD events received a significantly greater cumulative dose of corticosteroids during follow-up than did those who remained CAD free. In the regression models for first-available and average-of-first-two TC levels, exposure to corticosteroids at any time during the course of disease, irrespective of dose and duration of use, was associated with a dramatic 4.17-fold increased risk of a CAD event, independent of other risk factors. This HR was reduced to 1.85, but remained substantial and statistically significant, in the time-dependent models wherein recent exposure to corticosteroids was related to a CAD event.

In time-constant models for TC and BP, antimalarial use during the course of SLE was associated with a remarkable 50% to 59% reduction in hazard of a CAD event. The lack of a significant association between recent antimalarial use and CAD in the time-dependent models may point to a beneficial effect with long-term rather than short-term use. In previous studies, use of antimalarials has been associated with a reduction in TC level 5141617. However, ours is one of only a few studies in which the use of antimalarials has been linked with a reduction in the risk of actual CAD events 1819. Furthermore, the halving of coronary risk makes a strong case for the use of antimalarials in patients with SLE, not only to control disease activity but also for cardioprotection.

As in previous studies, we have shown that male sex and older age are associated with increased hazard of a CAD event. In previous studies, anti-phospholipid antibodies, diabetes, and smoking were shown to be independent risk factors for CAD events in SLE 20. However, in this study, in multiple regression analysis, such an association was not found, possibly because of the limited number of patients who smoke or have diabetes or anti-phospholipid antibodies. This study is not an exhaustive evaluation of traditional risk factors, and the role of other variables, such as family history of ischemic heart disease, body mass index (BMI), and waist/hip ratio, merit further investigation in future studies.

In summary, we showed that elevated TC and BP are both potentially treatable risk factors for CAD in SLE. Our study highlighted the importance of frequent measurements of BP and TC in management of patients with SLE. Although clinicians might consider the need for change in treatment based on single TC and BP measurements, their decision actually to do so should be made on the basis of the mean of several measurements.

Overall, this study has both conceptual and practical significance. From a conceptual point of view, our findings illustrate that in a systemic inflammatory disease, for a dynamic risk factor such as TC or BP, summary measures such as mean and AM better reflect cumulative exposure and hence better quantify CAD risk. This is an important consideration in future studies of dynamic risk factors for CAD in a chronic relapsing-remitting disease such as SLE and may be used to derive risk-prediction models specifically for SLE. From a practical point of view, this study has shown that assessment of coronary risk related to TC and BP in patients with SLE relies on serial measurement of these risk factors throughout the relapsing-remitting course of SLE. Additionally, this study has shown that disease activity and corticosteroid use are CAD risk factors in SLE, indicating that disease activity should be recognized and optimally controlled by using the minimal effective dose of corticosteroids. The demonstration of a cardioprotective association of antimalarials highlights the staple role of this class of drugs in the management of patients with SLE. Finally, this study has quantified the risk associated with exposure to TC, SBP, and DBP over time, emphasizing the importance of these potentially treatable traditional risk factors in patients with SLE. Future efforts must be directed toward determining TC and BP cut-points for CAD risk stratification specifically among patients with SLE and the role of treatment of risk factors in reducing the incidence of CAD events.

ACR: American College of Rheumatology; AM: time-adjusted mean; AMS: adjusted mean SLE disease activity index 2000; AUC: area under the curve; BP: blood pressure; CAD: coronary artery disease; CI: confidence interval; DBP: diastolic blood pressure; ECG: electrocardiographic; HR: hazard ratio; LDL-C: low-density lipoprotein cholesterol; Max: maximum; Min: minimum; mm Hg: millimeters of mercury: mmol/L; millimoles per liter; MI: myocardial infarction; SBP: systolic blood pressure; SD: standard deviation; SLE: systemic lupus erythematosus; SLEDAI-2K: Systemic Lupus Erythematosus Disease Activity Index 2000; SLICC/ACR DI: Systemic Lupus Erythematosus International Collaborating Clinics/American College of Rheumatology Damage Index; TC: total cholesterol.

The authors declare that they have no competing interests.

MN participated in the study design, collection and analysis of data, interpretation of results, and preparation of the manuscript. DDG and MBU participated in the study design, collection of data, interpretation of results, and preparation of the manuscript. DI participated in the study design, analysis of data, interpretation of results, and preparation of the manuscript. PJH participated in the study design, interpretation of results, and preparation of the manuscript. All authors read and approved the manuscript for publication.