Subclinical atherosclerosis in systemic lupus erythematosus (SLE): the relative contribution of classic risk factors and the lupus phenotype

Transcription

1 Rheumatology 2007;46: Advance Access publication 23 March 2007 doi: /rheumatology/kem002 Subclinical atherosclerosis in systemic lupus erythematosus (SLE): the relative contribution of classic risk factors and the lupus phenotype Y. Ahmad 1,2, J. Shelmerdine 1, H. Bodill 3, M. Lunt 2, M. G. Pattrick 4, L. S. Teh 5, R. M. Bernstein 1, M. G. Walker 3 and I. N. Bruce 1,2 Objectives. We aimed to examine the strength of association between traditional cardiovascular risk factors and carotid plaque development in systemic lupus erythematosus (SLE) patients and controls. We also aimed to determine which lupus-related factors are associated with carotid plaque and whether SLE sensitizes patients to the effects of traditional factors. Methods. We studied 200 women with SLE and 100 controls. Demographic and risk factor data were collected and SLE features, including autoantibody profiles and therapy were noted. All subjects had B- mode ultrasound of their carotid arteries examined for the presence and distribution of plaque. Results. SLE patients <55 years old had more plaque (21% vs 3% P < 0.01) and more SLE patients had plaque in the internal carotid artery (11% vs 4%; P < 0.05). Traditional risk factor models performed less well in SLE compared with controls [area under Receiver Operator Characteristic curves (AUC ROC) ¼ 0.76 vs 0.90; P < 0.01]. A multivariable model using SLE factors only, performed significantly better (AUC ROC ¼ 0.87; P < 0.01). The final model in SLE included age and cigarette pack-years smoking as well as azathioprine exposure ever, antiphospholipid antibodies (APLA) and previous arterial events (AUC ROC ¼ 0.88). Conclusions. SLE patients have a higher prevalence and different distribution of carotid plaque than controls. SLE factors perform significantly better than traditional risk factors in their association with atherosclerosis in SLE and these factors add to the influence of traditional risk factors rather than sensitizing lupus patients to traditional factors. The SLE phenotype helps identify patients at increased risk of atherosclerosis. KEYWORDS: Systemic lupus erythematosus, atherosclerosis, risk factors, smoking. Introduction Premature coronary heart disease (CHD) is a major cause of morbidity and mortality in patients with systemic lupus erythematosus (SLE) and in younger women, the excess risk may be >50-fold [1]. SLE patients also have an increased prevalence of subclinical atherosclerotic disease detected using several modalities [2 4]. Hypertension and diabetes mellitus are also more prevalent in SLE [5] and traditional risk factors, e.g. hypercholesterolaemia, contribute to the development of atherosclerosis in SLE [1, 2, 6]. However, these risk factors alone do not explain the excess CHD risk and after adjusting for traditional risk factors, SLE itself remains independently associated with both clinical and subclinical outcomes [2, 4, 7]. Several lupus-related factors may contribute to the development of accelerated atherosclerosis, for example, there is growing evidence that atherosclerosis itself has a chronic inflammatory component [8]. In addition to chronic inflammation, patients with SLE frequently have lupus anticoagulant (LAC) or associated antiphospholipid antibodies (APLA) and there is in vitro evidence that certain APLA may be pro-atherogenic as well as prothrombotic [9]. Corticosteroid therapy has also been associated with clinical and subclinical disease in several studies [1, 6]. However, more recently Roman et al. [2] suggested that less steroid exposure was associated with carotid plaque. The aim of this current study was to assess the prevalence of carotid plaque in an ethnically homogenous cohort of women with SLE compared with healthy controls. We specifically aimed to compare the strength of association between traditional CHD risk factors and carotid plaque development in patients and controls. We also aimed to determine which lupus-related factors are associated with carotid plaque and assess whether SLE sensitizes patients to the effects of traditional factors or whether any SLE effect is additive to the risk of atherosclerosis development. Patients and methods The study was approved by the North West Multi-Centre Research Ethics Committee and written informed consent was obtained from each participant. SLE patients SLE patients were recruited from out-patient clinics in Manchester Royal Infirmary, North Manchester General Hospital, Blackburn Royal Infirmary, South Manchester University Hospital and Stepping Hill Hospital, Stockport. We included SLE patients >18 yrs old who fulfilled updated ACR criteria for SLE [10] and also patients with three criteria for SLE in the absence of any alternative diagnosis. All cases were white Caucasian females of British Isles descent; we limited recruitment to this ethnic group as we were also collecting DNA for future genetic studies. Patients were on stable therapy for at least 2 months and we excluded women who were pregnant or lactating mothers within 6 months. Patients underwent a clinical interview and examination according to a standard protocol that included demographic information, family history and lifestyle factors. We also assessed the presence of traditional coronary risk factors, anthropomorphic measures and history of previous 1 The University of Manchester, Rheumatism Research Centre, 2 ARC Epidemiology Unit, and 3 Department of Vascular Surgery, Manchester Royal Infirmary, Central Manchester and Manchester Children s University Hospitals NHS Trust, Oxford Road, Manchester, 4 North Manchester General Hospital, Crumpsall Road, Manchester and 5 Blackburn Royal Infirmary, Blackburn, Lancashire, UK. Submitted 21 June 2006; revised version accepted 3 January Correspondence to: I. N. Bruce, MD, FRCP, Reader in Rheumatology, University of Manchester, Rheumatism Research Centre, Central Manchester and Manchester Children s University Hospitals NHS Trust, Oxford Road, Manchester M13 9WL, UK. ian.bruce@manchester.ac.uk 983 ß The Author Published by Oxford University Press on behalf of the British Society for Rheumatology. All rights reserved. For Permissions, please journals.permissions@oxfordjournals.org

2 984 Y. Ahmad et al. arterial events, i.e. myocardial infarction, angina, stroke, transient ischaemic attacks or peripheral vascular disease. SLE disease activity and cumulative damage was measured on the day of the assessment, using the Systemic Lupus Erythematosus Disease Activity Index (2000 version) (SLEDAI-2K) [11] and the Systemic Lupus International Collaborating Clinics/ACR damage index (SDI) [12]. Controls We recruited healthy female controls from the same ethnic background and geographical region using a best friend system. We asked patients in the current study to invite a friend (non-relative) to take part in the study. Since the prevalence of carotid plaque is very low in young women we targeted older SLE cases to bring along a friend thereby recruiting women from a slightly older age profile than the whole SLE group to enable us to model factors associated with plaque in controls. Controls were excluded if they had any history of systemic autoimmune disease, e.g. rheumatoid arthritis or SLE. Blood sampling and laboratory testing On the morning of study following an overnight (12 h) fast and avoidance of alcohol for 48 h a blood sample was drawn and used to measure routine blood counts, a lipid profile, blood glucose and serum creatinine. Serology tested (SLE only) included antibodies to double stranded-dna (ds-dna) (normal <25 U), cardiolipin (acl) (co-factor dependent assay) (normal <16U) (Aesku Diagnostics Wendelsheim, Germany) and C3 and C4 complement. We also recorded if patients were ever positive for these serological tests and in the case of acl antibodies ever positive was defined as having two positive tests >6 weeks apart [13]. LAC was measured using the dilute Russell Viper Venom test and again, ever positive was defined as two positive tests >6 weeks apart. Carotid ultrasound This was performed at the Manchester Royal Infirmary Vascular Laboratory by a single operator (HB) blinded to the diagnosis of the subject. An ATL HDI 5000 scanner equipped with 7 4 MHz linear array transducer was used. The right and left common carotid artery (CCA), carotid bulb and the first 1.5 cm of the internal and external carotid arteries were examined in longitudinal and cross-sectional planes. Intima-medial thickness (IMT) measurements were made in a longitudinal plane at a point of maximum thickness on the far wall of the CCA along a 1 cm section of the artery proximal to the carotid bulb. Measurements were repeated three times on each side, unfreezing the image on each occasion and relocating the maximal IMT, the average of six measurements (3 left and 3 right) were then used to calculate the mean IMT [14]. In a preliminary study of 15 healthy volunteers the intra-observer reliability for this technique was very high (intraclass correlation coefficient 0.92, 95% CI ). Carotid plaque was defined if two of the following three conditions were met: (i) a distinct area of protrusion >50% compared with the surrounding area into the vessel lumen, (ii) increased echogenicity than the adjacent boundaries and (iii) IMT >0.15 cm [15]. Statistical analysis Comparison of continuous data was carried out using the Mann Whitney U-test. For categorical data the chi-squared was employed. A logistic regression model was used to assess the relationship between predictor variables and carotid plaque (present or not). Univariable analysis using the individual factors was carried out for the association with carotid plaque. For each multivariable model, all baseline variables that were found on univariable analysis to be significant at P 0.20 were entered. This model was simplified using backward stepwise regression retaining variables that were significant at P < Previously excluded variables (univariable; P > 0.20) were added back to the model to determine whether they contributed significantly and any that became significant at P < 0.10 were retained. Logistic multivariable model assumption analyses were checked and only commented if the data set did not satisfy the assumptions. The predicted probability of each model was then used to derive the area under the Receiver Operator Characteristic curve (AUC ROC) [16]. The ROC curve is used to determine the accuracy of diagnostic tests or for judging the discrimination ability of statistical models. The ROC curve is constructed by measuring the sensitivity (true positive) and 1 specificity (false positive) across all possible threshold values that define the positivity of a condition. The accuracy of a test depends on the area and this is classified as: ¼ excellent, ¼ good, ¼ fair, ¼ poor, ¼ fail. Comparisons of the areas under the curves were made using the methods of DeLong et al. [17]. STATA TM 8.0 (STATA corporation, 1993) was used to perform all statistical procedures and significance was set at a P-value < Results We studied 200 SLE patients, mean (S.D.) age was 48.2 (10.4) yrs. The basic characteristics of the patients with SLE are summarized in Table 1. As can be seen, 96% satisfied 4 of the 1997 revised ACR criteria. The remaining eight patients (4%) satisfied three ACR criteria but they had other additional SLE-related clinical and laboratory features and no evidence for any other medical diagnosis. All eight patients had a positive ANA, other clinical and laboratory features included arthritis (7), lymphopaenia (4), Raynaud s phenomenon (2), alopecia (2), hypocomplementaemia (2), photosensitivity (1) malar rash (1), anti-dsdna antibodies (1) and serositis (1). Overall patients had low disease activity at the time of study. In addition, using the modified Cockcroft Gault formula, the median (IQR) egfr was 77.5 ( ) ml/min, 66 (33%) patients had an egfr <60 ml/min. Ninety-four (45%) were postmenopausal, including 18 (15%) of 120 <50 yrs old. In addition to the drugs outlined in Table 1, 53 (26.5%) patients were also were taking antiplatelet therapy, 28 (14%) were taking oral anticoagulants and 20 (10%) were receiving statins. A total of 100 controls were also studied, their mean (S.D.) age was 51.9 (9.2) yrs and 59 were postmenopausal including only 3% of those under 50 yrs. The distribution of classic risk factors in patients and controls are shown in Table 2. SLE patients were by TABLE 1. Baseline disease-related features of the SLE patients recruited Variable n (%) a 4 ACR criteria 192 (96%) Median (IQR) age at diagnosis 36.4 (11.8) yrs Median (IQR) disease duration 11.7 (9.4) yrs Previous cerebrovascular event 16 (8%) Previous coronary event 7 (3.5%) Peripheral vascular disease 1 (0.5%) Ever ANA positive 188 (94%) Ever dsdna positive 115 (57.5%) Ever acl or LAC positive 73 (37%) Antiphospholipid syndrome 21 (10.5%) Current steroid therapy 106 (53%) Median (IQR) SLEDAI-2K 1 (0 4) Median (IQR) SLICC damage index 0 (0 4) Median (IQR) current steroid dose taken 5 (5 10) mg/day prednisolone Median (IQR) average daily steroid 6.13 (5 10) mg/day prednisolone dose in past 6 months Current antimalarial therapy 105 (52.5%) Current immunosuppressive therapy 74 (37%) Azathioprine 42 (21%) Methotrexate 23 (11.5%) a Variables are presented as n(%) unless indicated otherwise.

3 Subclinical atherosclerosis in SLE 985 TABLE 2. Comparison of baseline demographics and classic risk factors in SLE patients and healthy controls Variable Controls (n ¼ 100) median (IQR) a SLE (n ¼ 200) median (IQR) a P-value Age, yrs 53 (46 59) 48 ( ) <0.01 Postmenopausal n (%) (47) 0.05 b Postmenopausal aged 50 yrs, 1/39 (3) 18/117 (15) 0.03 b n (%) Non-smoker, n (%) 55 (55) 103 (51.5) NS b Current smoker, n (%) 17 (17) 40 (20) Ex-smoker, n (%) 28 (28) 57 (28.5) Family history of premature 25 (25) 53 (26.5) NS b coronary heart disease, n (%) Total cholesterol, mmol/l 5.7 ( ) 5.1 ( ) <0.01 Triglycerides, mmol/l 1.1 ( ) 1.0 ( ) NS HDL, mmol/l 1.7 ( ) 1.6 ( ) 0.06 Calculated LDL, mmol/l 3.42 ( ) 2.88 ( ) <0.01 Fasting plasma glucose, mmol/l 4.7 ( ) 4.6 ( ) <0.01 Systolic blood pressure, mmhg ( ) 126 ( ) 0.1 Diastolic blood pressure, mmhg 75 (68 81) 76.5 ( ) NS Previous or current 24 (24) 83 (41.5) <0.01 b hypertension, n (%) BMI, kg/m ( ) 26.0 ( ) NS Waist:hip ratio 0.81 ( ) 0.82 ( ) NS 10-yr predicted coronary risk (%) 3.55 ( ) 2.7 ( ) 0.06 a Variables are presented as median and interquartile range unless indicated otherwise. b All comparisons are Mann Whitney U-test unless where a chi-squared test is indicated. design younger. Interestingly, however, they had more hypertension and had an earlier menopause. Hypercholesterolaemia was more common in controls that also tended to have higher fasting blood glucose. There was a trend towards lower HDL cholesterol in patients. Subclinical atherosclerosis in SLE and controls There was a trend towards higher median IMT in controls [median (IQR) 0.051cm ( ) vs cm ( ), P ¼ 0.09]. The prevalence of plaque in the control and SLE groups was 22% and 29%, respectively. Plaque prevalence increased with age and was significantly higher in the <45 yrs and yrs SLE patient groups (P < 0.05 for both age bands) (Fig. 1). The overall prevalence of plaque in SLE patients <55 yrs old was 21% compared with 3% in controls (P < 0.01). When stratified by tertiles of IMT, the prevalence of plaque in the highest tertile (IMT >0.055 cm) was similar in patients and controls (45% vs 47% P ¼ NS). In the middle tertile (IMT cm), there was more plaque in the SLE group (38% vs 8% P < 0.01). In the IMT <0.048 cm tertile, 11% of patients and 6% of controls had plaque (P ¼ NS). The distribution of plaque also showed differences and while the commonest site for plaque was at the carotid bulb in both groups (SLE group 23% vs controls 21%) significantly more SLE patients had plaque in the internal carotid artery (11% vs 4% P < 0.05). The influence of classic risk factors on carotid plaque development Controls with plaque had a higher prevalence of several classic risk factors including older age, smoking, higher blood pressure, total and LDL cholesterol. They also had a higher carotid IMT (data on file). Similarly, SLE patients with carotid plaque also showed a similar pattern of risk factors but also had higher triglyceride concentrations (data on file). The median (IQR) IMT was also higher in SLE patients with plaque compared with those without [0.053 ( ) vs ( ) P < 0.01]. In the backward stepwise multivariable model variables that were significant at P 0.2 were included (Table 3), we excluded LDL-cholesterol, diastolic blood pressure and 10-yr estimated risk as these are either derived from other values and/or closely Prevalence of carotid plaque (%) * * <45 yrs yrs >55 yrs associated with other variables in the model. In controls, our final model included age [OR (95% CI) 1.25 ( )], pack years smoking [1.04 ( )] and systolic blood pressure [1.04 ( )]. This model was excellent for its association with plaque (AUC ROC ¼ 0.90) (Fig. 2A). In SLE patients this model derived from controls performed significantly less well in its association with plaque (AUC ROC ¼ 0.75, P < 0.01). We also constructed the best model derived only from classic risk factors in SLE patients. This included age [OR (95% CI) 1.08 ( )], pack years smoking [1.04 ( )] and triglycerides [(1.59 ( )]. Again, this model performed significantly less well compared with our classic risk factor model in controls (AUC ROC ¼ 0.76, P < 0.01) (Fig. 2B). SLE factors and carotid plaque *P<0.05 Controls SLE FIG. 1. The prevalence of carotid plaque in women with SLE compared with healthy controls by age groups. The prevalence of plaque was significantly higher in SLE patients in the <45 yrs and yrs age groups (*P < 0.05). Number of cases and controls per age band Age group Controls Cases <45 yrs yrs >55 yrs TABLE 3. Univariable association of traditional risk factors and the presence of carotid plaque in SLE patients and healthy controls SLE patients Controls Risk factor OR 95% CI OR 95% CI Age, yrs , , 1.44 Family history of CHD , , 4.40 Cigarette pack/yrs , , 1.09 Glucose, mmol/l , , 2.83 Total cholesterol, mmol/l , , 2.64 Triglyceride, mmol/l , , 1.98 HDL-c, mmol/l , , 1.44 LDL-c, mmol/l , , 3.62 Systolic BP, mmhg , , 1.08 Diastolic BP, mmhg , , 1.08 % 10 year risk of CHD , , 1.5 * Factors included in the multivariable analysis (see text for details) are indicated in bold. In univariable analysis we assessed the strength of association between SLE-related variables and the presence of plaque (Table 4). Patients with plaque were older at diagnosis of SLE and had longer disease duration. They also had more organ damage including a history of vascular events, antiphospholipid syndrome and lower estimated GFR. There was also a significant association with a longer duration of steroid use and ever being treated with azathioprine; an opposite trend was seen with methotrexate and there was no association with

4 986 Y. Ahmad et al. (A) True positive rate (B) True positive rate (C) True positive rate Area under ROC curve = 0.90 False positive rate Area under ROC curve = 0.76 False positive rate Area under ROC curve = False positive rate FIG. 2. ROC curves for the final models using traditional risk factors in healthy controls (A), traditional risk factors in SLE (B) and using SLE-related factors in patients (C). Comparison of AUC A vs B, P < 0.01; comparison of AUC B vs C, P < cyclophosphamide exposure. Using backward stepwise multivariable analysis, the final model using only SLE factors included age at diagnosis, disease duration, previous arterial events, higher neutrophil count, azathioprine use ever and acl/lac positivity ever (Table 5). The model was unchanged when restricted to patients free of previous arterial disease. Using different acl and LAC subtypes individually, IgG acl ever appeared to be the most significant association (data on file). This model had a very good association with plaque (AUC ROC ¼ 0.87). Indeed when we compared this lupus factors only model to our classic risk factor model in SLE patients the lupus factors model had a significantly better association with carotid plaque (P < 0.01; Fig. 2C). We then combined in a final model classic and SLE-related factors which had shown a univariable association with carotid plaque in SLE patients with P 0.2. This final model included age and pack-years of smoking as well as azathioprine exposure ever, acl and/or LAC ever and previous arterial events (Table 5) (AUC ROC ¼ 0.88). TABLE 4. Univariable association of lupus-related factors and the presence of carotid plaque Variable OR 95% CI Demographics Age at diagnosis, years Disease duration, years SLE clinical features Photosensitivity Malar rash Discoid rash Mucosal ulcers Serositis Arthritis Renal disease CNS involvement SLE lung involvement Coronary and/or cerebral event Venous thromboembolism Raynaud s phenomenon Antiphospholipid syndrome Laboratory results Antinuclear antibody IgG anti-ds-dna antibodies IgG anticardiolipin antibodies (acl) IgM anticardiolipin antibodies (acl) Lupus anticoagulant (LAC) Anticardiolipin antibody and/or LAC Anti-Ro antibodies Anti-La antibodies Anti-Sm antibodies Anti-RNP antibodies Low C4 complement Total white cell count Total lymphocyte count Total neutrophil count Estimated glomerular filtration rate SLE indices SLEDAI-2K SLICC damage index Prednisolone Therapy Total duration, months Average dose, past 6 months, mg/day Current dose, mg/day Other therapy Anti-malarial ever Azathioprine ever Cyclophosphamide ever Methotrexate ever Anti-platelet therapy Anti-coagulation *Autoantibodies were recorded if they were ever present in a particular patient during the course of their SLE. Factors included in the multivariable analysis (P < 0.2) are indicated in bold. Discussion The focus of this study was to examine factors associated with the presence of subclinical atherosclerosis in patients with SLE. We found that in the general population classic risk factors show an excellent association with plaque. However, the same multivariable model performed significantly less well in SLE patients. In contrast, a multivariable model including only SLE factors performed significantly better. Like others we also found that subclinical atherosclerosis develops at an earlier age in SLE patients [2, 19]. In addition we have also shown that plaque begins to develop at a lower IMT in SLE and that the distribution of plaque is different in SLE patients. There was a higher prevalence of internal carotid artery plaque in SLE. Since our primary aim was to examine factors associated with subclinical atherosclerosis, control patients were deliberately chosen to be older so that there would be an adequate number of subjects with plaque. As expected, plaque development in controls was strongly associated with increasing age as well as other classic risk factors. Since all subjects were drawn from the same ethnic background and geographical region we were

5 Subclinical atherosclerosis in SLE 987 TABLE 5. Final models in SLE patients using either SLE factors alone (model 1) and when combining traditional and SLE-related factors (model 2). In both models independent variables were retained at P 0.20 Model 1 SLE patients using SLE factors only (n ¼ 183) Model 2 SLE patients using SLE and traditional risk factors (n ¼ 182) Variable OR (95% CI) P Variable OR (95% CI) P Age at diagnosis, yrs 1.12 ( ) <0.01 Age, yrs 1.11 ( ) <0.01 Disease duration, yrs 1.12 ( ) <0.01 Cigarette packs/yr 1.05 ( ) <0.01 Current/previous azathioprine 3.32 ( ) <0.01 Current/previous azathioprine 3.78 ( ) <0.01 acl/lac a 4.22 ( ) <0.01 acl/lac a 5.18 ( ) <0.01 Neutrophil count 1.24 ( ) 0.04 Neutrophil count 1.22 ( ) 0.07 Coronary and/or cerebral event 7.62 ( ) <0.01 Coronary and/or cerebral event 7.17 ( ) <0.01 AUC ROC ¼ 0.87 AUC ROC ¼ 0.88 The final model for each is presented as odd ratios (OR) and 95% CI. The area under the ROC for each model is also presented. a acl/lac: anticardiolipin antibodies or lupus anticoagulant ever positive. able to test the classic risk factor model from controls in SLE patients. This model performed significantly less well in lupus patients as did the best traditional risk factor model developed in SLE patients. Therefore, in lupus, classic risk factors do not have the same magnitude of impact on subclinical atherosclerosis as they have in the background population. If SLE patients were sensitized to the effects of classic risk factors then our model would have performed at least as well in SLE. A unique finding in this study is that when developing a model using SLE factors only, we can significantly improve the ROC for plaque. Therefore, knowledge of the SLE phenotype alone gives better information with regard to likelihood of having a carotid plaque than knowledge of the traditional risk factor status alone. In addition to older age at diagnosis and longer disease duration, the key factors associated with the plaque in SLE included azathioprine use ever, acl and/or LAC positivity ever, a higher neutrophil count and previous arterial events. When we combined lupus and traditional risk factors our final model performed only marginally better than this lupus only model with the neutrophil count no longer being significant having been replaced by cigarette pack years of smoking. This final model was robust in that it was not affected by removing patients with prevalent cardiovascular disease and was also not affected by holding traditional risk factors within the model (data on file). With regard to the lupus factors identified, the association with previous arterial events demonstrates the validity of carotid plaque as a useful surrogate marker for study in this context. The fixed factors of older age at diagnosis and longer disease duration have also been found to be important in other studies of clinical and subclinical atherosclerosis [2, 6, 20]. Exposure to azathioprine was also strongly associated with plaque in our study and other groups have recently noted this [21, 22]. Azathioprine is our maintenance drug of choice in patients with lupus nephritis or neuropsychiatric involvement; it is also employed to control inflammatory disease not responsive to steroids and antimalarial drugs. Azathioprine may therefore be a surrogate for a more severe lupus phenotype or, of more previous steroid exposure. Indeed our azathioprine treated patients had longer disease duration and longer duration of steroid use as well as more renal, cutaneous and serosal disease in their past history (data on file). Also, in our univariable analysis, longer duration of steroid use was associated with plaque, which also supports this explanation. This, of course, is in contrast to Roman s findings; however in that study average steroid dose was assessed over the previous 5 years which may not reflect the overall steroid exposure in a lupus cohort [2]. It is, however, interesting to note that transplantation studies suggest that immunosuppressive drugs such as mycophenolate mofetil may be superior to azathioprine in reducing the rates of cardiovascular events in these populations [22]. Therefore there remains the possibility that azathioprine may not control vascular inflammation as effectively as other immunosuppressive agents. Further prospective studies examining the differential effects on the vasculature of different immunosuppressive agents in SLE seems warranted. We also found an association with acl/lac either currently or in the past. On subgroup analysis there was a particularly strong association with IgG acl. APLAs are recognized precipitants of thrombosis and it is therefore not surprising that they are associated with clinical CHD [20]. Since our study examined subclinical atherosclerosis, this is an important observation to consider. The association of APLA with atherogenesis is controversial but several mechanisms may be of relevance. Beta-2 glycoprotein-1 antibodies can promote uptake of oxidized LDL by macrophages [9]. There is also cross reactivity between certain APLA subgroups and antibodies to oxidized LDL, HDL or APO-A1 that may reflect oxidant stress and/or an adverse lipid profile [24, 25]. One other recent study suggested that such antibodies may be associated with increased carotid IMT [26]. Others however have found no association [2, 19, 21] and indeed certain antibody subgroups may have an atheroprotective role [27]. The direction of the association we have observed needs further evaluation and is the subject of an ongoing study in our unit. With regard to the association with a higher neutrophil count, there was only a weak association between neutrophil count and steroid dose that would not explain the association found (data on file). Of relevance, however, smokers have a higher white cell count and cigarette pack-years replaced this in our final model [28]. The neutrophil count may also reflect persisting inflammatory disease activity and in the general population is of value in predicting future events [28, 29]. As well as risk factors identified, we also noted that plaque developed at a lower IMT in SLE patients. This supports the hypothesis that the accelerated process of plaque development may be driven by different factors than those that chiefly determine IMT progression [30]. In addition, as far as we are aware, this is also the first study to report a different distribution of plaque in SLE with more patients than controls having plaque in the internal carotid artery. There is evidence that the risk factors associated with plaque at different sites may show subtle but significant differences [31, 32]; however, this study was under-powered to address this issue. Further studies will aim to determine whether any SLE-related factors contribute to this difference and also whether these plaques contribute to the increased risk of stroke in this population [7]. Several limitations of this study need to be considered. Firstly our study only included white Caucasian patients from the North West of England, this was originally to enable future examination of genetic biomarkers. However, it also provided a unique opportunity to compare our classic risk factor model derived from controls in a lupus cohort of an identical ethnic and regional background. Importantly our findings are similar to a recent study of clinical CHD in a multiethnic cohort from the USA [22].

6 988 Y. Ahmad et al. Therefore our key finding of a differential influence of traditional and lupus factors is likely to be generalizable to other SLE populations. The other key limitation is the cross-sectional nature of the study that limits our ability to identify the direction of association of some of these factors with carotid plaque. This is particularly true for azathioprine exposure and antiphospholipid antibodies; prospective follow-up of this population is now underway. In this study of subclinical atherosclerosis in SLE, we have therefore found an increased prevalence of plaque occurring at a younger age and starting at a lower carotid IMT in patients compared with ethnically matched controls. There was also a higher prevalence of plaque in the internal carotid artery in SLE patients. The strength of association between traditional risk factors and plaque was significantly stronger in controls than patients. A statistical model employing only lupus-related factors showed a much stronger association with plaque in SLE and knowledge of these factors may help identify patients at increased risk of atherosclerosis in SLE. Our study also demonstrates that lupus adds to the influence of traditional risk factors rather than simply sensitizing lupus patients to classic risk factors. Addressing classic risk factors and SLE-related factors will therefore both be required to reduce the burden of atherosclerosis in SLE. Prospective studies are now needed to examine the mechanism by which these factors mediate plaque development in lupus and these may suggest novel preventative strategies for this population. Acknowledgements The authors would like to acknowledge the help and support of Drs Neil Snowden, Beverley Harrison, Patricia Smith, Jeff Marks, Paul Sanders, Faieza Qasim, Colin Short and Mike Venning for helping provide patients for this study. We also acknowledge the support of the Wellcome Trust Clinical Research Facility, Central Manchester Health Care Trust for hosting the study and we acknowledge the funding support of the Arthritis Research Campaign, The Wellcome Trust and Lupus, UK. The authors have declared no conflicts of interest. References 1 Manzi S, Meilahn EN, Rairie JE et al. Age-specific incidence rates of myocardial infarction and angina in women with systemic lupus erythematosus: comparison with the Framingham Study. Am J Epidemiol 1997;145: Roman MJ, Shanker BA, Davis A et al. Prevalence and correlates of accelerated atherosclerosis in systemic lupus erythematosus. N Eng J Med 2003;349: Bruce IN, Burns RJ, Gladman DD, Urowitz MB. Single photon emission computed tomography dual isotope myocardial perfusion imaging in women with systemic lupus erythematosus. I. Prevalence and distribution of abnormalities. J Rheumatol 2000;27: El Magadmi M, Bodill H, Ahmad Y et al. Systemic lupus erythematosus: an independent risk factor for endothelial dysfunction in women. Circulation 2004;110: Bruce IN, Urowitz MB, Gladman DD, Ibanez D, Steiner G. Risk factors for coronary heart disease in women with systemic lupus erythematosus: the Toronto Risk Factor Study. Arthritis Rheum 2003;48: Petri M, Perez-Gutthann S, Spence D, Hochberg MC. Risk factors for coronary artery disease in patients with systemic lupus erythematosus. Am J Med 1992; 93: Esdaile JM, Abrahamowicz M, Grodzicky T et al. Traditional Framingham risk factors fail to fully account for accelerated atherosclerosis in systemic lupus erythematosus. Arthritis Rheum 2001;44: Hansson GK. Inflammation, atherosclerosis and coronary artery disease. N Eng J Med 2005;352: Hasunuma Y, Matsuura E, Makita Z, Katahira T, Nishi S, Koike T. Involvement of beta 2-glycoprotein I and anticardiolipin antibodies in oxidatively modified low-density lipoprotein uptake by macrophages. Clin Exp Immunol 1997;107: Hochberg MC. Updating the American College of Rheumatology revised criteria for the classification of systemic lupus erythematosus. Arthritis Rheum 1997;40: Gladman DD, Ibanez D, Urowitz MB. Systemic lupus erythematosus disease activity index J Rheumatol 2002;29: Gladman DD, Urowitz MB, Goldsmith CH et al. The reliability of the Systemic Lupus International Collaborating Clinics/American College of Rheumatology Damage Index in patients with systemic lupus erythematosus. Arthritis Rheum 1997;40: Wilson WA, Gharavi AE, Koike T et al. International consensus statement on preliminary classification criteria for definite antiphospholipid syndrome: report of an international workshop. Arthritis Rheum 1999;42: Sidhu PS, Desai SR. A simple and reproducible method for assessing intimal-medial thickness of the common carotid artery. Br J Radiol 1997;70: Li R, Duncan BB, Metcalf PA et al. B-mode-detected carotid artery plaque in a general population. Atherosclerosis Risk in Communities (ARIC) Study Investigators. Stroke 1994;25: Hanley JA, McNeill BJ. The meaning and use of the area under a receiver operating characteristic ROC curve. Radiology 1982;143: DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988;44: Durrington PN. Risk in cardiovascular disease. Joint British Societies recommend their computer program for risk calculations. Br Med J 2000;321: Asanuma Y, Oeser A, Shintani AK et al. Premature coronary-artery atherosclerosis in systemic lupus erythematosus. N Eng J Med 2003;349: Svenungsson E, Jensen-Urstad K, Heimburger M et al. Risk factors for cardiovascular disease in systemic lupus erythematosus. Circulation 2001;104: Doria A, Shoenfeld Y, Wu R et al. Risk factors for subclinical atherosclerosis in a prospective cohort of patients with systemic lupus erythematosus. Ann Rheum Dis 2003;62: Toloza SM, Uribe AG, McGwin G, Jr et al. Systemic lupus erythematosus in a multiethnic US cohort (LUMINA). XXIII. Baseline predictors of vascular events. Arthritis Rheum 2004;50: Kobashigawa J, Miller L, Renlund D et al. A randomized active-controlled trial of mycophenolate mofetil in heart transplant recipents. Transplantation 1998; 66: Lahita RG, Rivkin E, Cavanagh I, Romano P. Low levels of total cholesterol, highdensity lipoprotein, and apolipoprotein A1 in association with anticardiolipin antibodies in patients with systemic lupus erythematosus. Arthritis Rheum 1993;36: Delgado Alves J, Kumar S, Isenberg DA. Cross-reactivity between anti-cardiolipin, anti-high-density lipoprotein and anti-apolipoprotein A-I IgG antibodies in patients with systemic lupus erythematosus and primary antiphospholipid syndrome. Rheumatology 2003;42: Ames PR, Margarita A, Delgado Alves J, Tommasino C, Iannaccone L, Brancaccio V. Anticardiolipin antibody titre and plasma homocysteine level independently predict intima media thickness of carotid arteries in subjects with idiopathic antiphospholipid antibodies. Lupus 2002;11: Nicolo D, Goldman BI, Monestier M. Reduction of atherosclerosis in low-density lipoprotein receptor-deficient mice by passive administration of antiphospholipid antibody. Arthritis Rheum 2003;48: Van Oostrum AJHHM, van Wijk JPH, Cabezas MC. Lipaemia, inflammation and atherosclerosis: novel opportunities in the understanding and treatment of atherosclerosis. Drugs 2004;64: Huang ZS, Jeng JS, Wang CH, Yip PK, Wu TH, Lee TK. Correlations between peripheral differential leukocyte counts and carotid atherosclerosis in non-smokers. Atherosclerosis 2001;158: Selzer F, Sutton-Tyrrell K, Fitzgerald SG, Pratt JE, Tracy RP, Kuller LH, Manzi S. Comparison of risk factors for vascular disease in the carotid artery and aorta in women with systemic lupus erythematosus. Arthritis Rheum 2004;50: Tell GS, Howard G, McKinney WM. Risk factors for site specific extracranial carotid artery plaque distribution as measured by B-mode ultrasound. J Clin Epidemiol 1989;42: Crouse JR, Goldbourt U, Evans G et al. Risk factors and segment-specific carotid arterial enlargement in the Atherosclerosis Risk in Communities (ARIC) cohort. Stroke 1996;27:69 75.