Cardiovascular risk factors in women with and without rheumatoid arthritis

Authors


Abstract

Objective

The risk of cardiovascular disease (CVD) is increased in patients with rheumatoid arthritis (RA). The objective of this study was to examine the distribution of known CVD risk factors and biomarkers of CVD in women with and without RA.

Methods

This study included two components: an examination of clinical CVD risk factors among women participating in the Nurses' Health Study, a prospective longitudinal cohort, and an analysis of CVD biomarkers among a subgroup of women from this cohort who provided a blood specimen in 1989 (biospecimen cohort). Data regarding clinical risk factors for CVD were collected in 1990 by mailed questionnaire. The diagnosis of RA was confirmed through a structured medical record abstraction. We compared clinical risk factors for CVD and biomarkers of CVD between women with and without RA, adjusting for age, body mass index (BMI), smoking status, and menopause status.

Results

Women with RA (n = 287) were significantly more likely than women without RA (n = 87,019) to report no alcohol use (48.2% versus 39.4%) and past cigarette smoking (47.8% versus 38.0%). No significant differences between these groups were observed for current smoker status, BMI, regular aspirin use, diabetes, hypertension, physical activity, and family history of early myocardial infarction. In the biospecimen cohort (69 RA cases and 491 controls), the levels of several inflammatory biomarkers linked to CVD were significantly elevated in women with RA, including CRP, fibrinogen, sICAM-1, sTNFRI, sTNFRII, and osteoprotegerin. Levels of total cholesterol, low-density lipoprotein, triglycerides, apolipoprotein B, and Lp(a) were similar between groups. Levels of homocysteine were similar, but vitamin B12 was significantly higher among women with RA than among the controls.

Conclusion

In women participating in the Nurses' Health Study, most traditional CVD risk factors were similar between those who had RA and those who did not. However, as expected, biomarkers of inflammation associated with CVD were generally elevated in women with RA.

Several epidemiologic studies have found an increased risk of cardiovascular disease (CVD) among patients with rheumatoid arthritis (RA) (1–3). Possible explanations for the increased risk of CVD include systemic inflammation (4), deleterious effects of antirheumatic treatments (5), lack of attention to proven CVD prevention (6), and/or a less favorable distribution of CVD risk factors in patients with RA. While traditional CVD risk factors, such as hypertension, diabetes, and hypercholesterolemia, have been evaluated in previous studies (1, 2, 7), it is not clear how these factors and other known clinical CVD risk factors account for the association between RA and CVD. In the present study, we analyzed data from the Nurses' Health Study to compare clinical risk factors for CVD and biomarkers of CVD among women with and without RA.

SUBJECTS AND METHODS

Study population.

The Nurses' Health Study is a prospective cohort study comprised of 121,700 women who were between the ages of 30–55 when they completed the baseline questionnaire in 1976. Questionnaires have been mailed to these women every 2 years to update information concerning medical, lifestyle, and other health-related information (8). The biennial questionnaires ask participants to report recent physician-diagnosed illnesses, including RA, diabetes, hypertension, and hypercholesterolemia, and the dates of diagnosis, their dietary habits, weight, cigarette smoking, menopause status, physical activity, blood pressure, family history of acute myocardial infarction, and use of selected prescription and over-the-counter medications as well as dietary supplements. Women who reported CVD and cancer (other than nonmelanoma skin cancer) beginning at baseline and continuing through the time of the blood collection (see below) in 1989 were excluded from the analyses. After these exclusions, 87,306 women were eligible for this study.

For the purposes of this study, 2 overlapping cohorts of women from the Nurses' Health Study were included. The first cohort included all eligible women (287 with RA and 87,019 controls). Among this cohort, we examined the distribution of the clinical risk factors for CVD, such as hypertension, diabetes, hypercholesterolemia, and physical activity. The second cohort, the “biospecimen cohort,” was assembled to examine biomarkers of CVD. This cohort came from a subgroup of Nurses' Health Study participants (n = 32,826) who provided a blood sample in 1989. From this group, we included all women diagnosed with RA prior to the blood collection (n = 69) and age-matched non-RA control subjects (n = 491) who were included in the biospecimen cohort. All women included in the biospecimen cohort were free of diagnosed CVD, including no prior myocardial infarction, angina, stroke, or procedures for coronary artery disease, at the time blood was collected.

This study was approved by the Institutional Review Board of the Partners HealthCare System at Harvard.

Clinical risk factors for CVD and biomarkers of CVD.

We examined the frequency of clinical risk factors for CVD in all eligible women, using data collected through 1990. We chose this point because the blood collection was done in 1989. The clinical variables of interest were race, smoking status, body mass index (BMI; weight in kg/height in meters2), menopause status, current use of aspirin for at least 4 days/week (“regular aspirin use”), diabetes, hypertension, hypercholesterolemia, family history of acute myocardial infarction in a parent or sibling younger than age 50 years, daily alcohol intake (gm/day), and physical activity (metabolic equivalents [METS]/week). These variables were obtained from responses to the 1990 questionnaire. For chronic diseases such as diabetes, any affirmative report on any questionnaire up through 1990 was included as the presence of that disease.

Biomarkers of CVD were measured using standard analytical techniques that have been previously described (9). Briefly, the blood samples were processed within 36 hours of venipuncture. Lipid parameters were assayed on a Hitachi 911 analyzer (Roche Diagnostics, Nutley, NJ). Homocysteine was measured using an IMX homocysteine assay (Abbott Laboratories, Abbott Park, IL). Vitamin B12 was measured by a quantitative sandwich enzyme immunoassay technique on the 2010 Elecsys Immunoanalyzer (Roche Diagnostics). Apolipoprotein B and fibrinogen were measured using immunoturbidimetric assays on the Hitachi 911 analyzer (Roche Diagnostics). C-reactive protein (CRP) and Lp(a) were measured using spectrophotometric techniques on the Hitachi 911 analyzer. Levels of soluble intercellular adhesion molecule 1 (sICAM-1), soluble vascular cell adhesion molecule 1 (sVCAM-1), soluble tumor necrosis factor receptor type I (sTNFRI), and sTNFRII were measured with enzyme-linked immunosorbent assays (ELISAs) from R&D Systems (Minneapolis, MN). Osteoprotegerin (OPG) and anticardiolipin antibodies were measured using ELISAs (Alpco Diagnostics, Windham, NH).

Confirmation of RA.

We conducted followup on the 6,730 women who reported having RA on any of the biennial questionnaires from 1978 to 1990. Respondents were asked to complete a connective tissue disease screening questionnaire (10). Based on their responses, medical records were obtained on 1,515 participants who had symptoms suggestive of RA and who granted permission for us to review their medical charts. Two rheumatologists trained in chart abstraction independently reviewed the medical records for data indicative of RA, using the American College of Rheumatology (ACR; formerly, the American Rheumatism Association) 1987 criteria for RA (11). These criteria were assessed in a cumulative manner, examining all available medical records. Patients with 4 of the 7 ACR criteria were considered to have RA. We excluded from analysis women in whom the diagnosis of RA was not confirmed by the medical record review.

Statistical analysis.

We conducted 2 sets of comparisons between women with RA and women without RA who served as controls. First, we examined the frequency of clinical CVD risk factors among all eligible women. The frequencies were compared using chi-square tests for categorical variables or Student's t-tests for continuous variables. Because of statistically significant differences in several potential confounders, we examined age-adjusted and multivariate-adjusted logistic models to assess differences between the women with RA and the non-RA controls. The multivariate models included age and BMI as continuous variables, as well as smoking and menopause status. Continuous variables that were not normally distributed were log-transformed. The second set of comparisons assessed the biomarkers of CVD among the women in the biospecimen cohort. The means for women with RA were compared with those for the non-RA controls using similar age-adjusted and multivariate-adjusted linear models. All analyses were performed with SAS software (version 6.12; SAS Institute, Cary, NC). P values less than 0.05 were considered statistically significant.

RESULTS

Table 1 shows the clinical risk factors for CVD as of 1990 for all eligible women in the Nurses' Health Study. The age of the study population was between 43 and 68 years. On average, women with RA were 2 years younger than women without RA. After adjusting for age, BMI, and menopause status, women with RA were more likely to report having been smokers in the past, but current smoking was reported with similar frequency. Mean BMI was similar among women with RA and controls. In the age-adjusted analysis, postmenopause status was more common among women with RA, but after adjusting for age, BMI, and smoking status, the difference was no longer statistically significant. Women with RA and those without RA did not differ in the frequency of regular aspirin use, diabetes, hypertension, and a family history of early myocardial infarction. Self-reported hypercholesterolemia was less common among women with RA. Women with RA were more likely to report no alcohol use.

Table 1. Clinical risk factors for cardiovascular disease among women enrolled in the Nurses' Health Study, grouped according to history of rheumatoid arthritis*
Clinical variableRheumatoid arthritisP
Present (n = 287)Absent (n = 87,019)Age-adjusted modelMultivariate-adjusted model
  • *

    Except where indicated otherwise, values are the age-adjusted percentage. Percentages may not add up to 100 because of rounding. Variables adjusted for in the multiple variable adjustment include age, body mass index (BMI), menopause status, and smoking status. Data were taken from the 1990 questionnaire in order to correspond to the blood draw for the biospecimen cohort. MI = myocardial infarction; METS = metabolic equivalents.

  • Women who did not report race were excluded (n = 7,807).

  • Women who did not report alcohol intake were excluded (n = 25,586).

Age, mean years56.158.20.002
Race    
 Caucasian93.794.80.130.7
 African American0.71.50.30.3
 Other5.73.70.020.01
Cigarette smoking    
 Never smoked32.743.8<0.001<0.001
 Past smoker47.838.0<0.001<0.001
 Current smoker19.218.00.70.8
 Missing data0.20.20.60.6
BMI, mean kg/m225.825.70.90.9
Postmenopausal81.977.60.040.08
Regular aspirin use23.825.60.60.5
Diabetes4.84.40.80.9
Hypertension31.429.00.60.7
Hypercholesterolemia29.235.70.030.02
Family history of early MI30.438.00.70.7
Alcohol use, gm/day    
 None48.239.40.001<0.001
 <532.232.70.90.9
 5–98.110.10.60.4
 10–144.98.40.10.07
 ≥156.69.30.40.2
Physical activity, mean METS/week13.415.60.20.2

When we restricted the analyses of the clinical risk factors for CVD to the 69 women with RA and the 491 non-RA controls in the biospecimen cohort, the findings were similar to those in the total Nurses' Health Study population, except for previous tobacco use, alcohol use, and physical activity. In the biospecimen cohort, the proportion of women reporting prior tobacco use and current alcohol use as well as the physical activity levels were similar between women with and women without RA (data not shown) (P > 0.2 for each comparison).

The mean levels of each biomarker examined in the biospecimen cohort are shown in Table 2, according to the presence and absence of RA. In multivariate-adjusted models, except for a slightly lower level of high-density lipoprotein (HDL) in women with RA, the lipid parameters were similar between women with and without RA. No difference in homocysteine levels was found, but vitamin B12 levels were higher in women with RA. Fibrinogen and CRP levels were significantly higher in the women with RA compared with the controls. The levels of sICAM-1 were also higher in women with RA, but the levels of sVCAM-1 were not. The levels of both sTNFRI and sTNFRII were higher in women with RA, as were levels of OPG. However, anticardiolipin antibody levels were similar in both groups.

Table 2. Biomarkers of cardiovascular disease risk among women in the biospecimen cohort, grouped according to history of rheumatoid arthritis*
Laboratory markerRheumatoid arthritisP
Present (n = 69)Absent (n = 491)Age-adjusted modelMultivariate-adjusted model
  • *

    Values are the age-adjusted mean. Variables adjusted for in the multiple variable adjustment include age, body mass index, menopause status, and smoking status. HDL = high-density lipoprotein; LDL = low-density lipoprotein; CRP = C-reactive protein; sICAM-1 = soluble intracellular adhesion molecule 1; sVCAM-1 = soluble vascular cell adhesion molecule 1; sTNFR = soluble tumor necrosis factor receptor (types I and II).

Cholesterol, mg/dl2182240.50.7
HDL, mg/dl58.060.10.120.02
LDL, mg/dl1251320.80.8
Triglycerides, mg/dl1181260.40.8
Apolipoprotein B, mg/dl941000.20.6
Lp(a), mg/dl18.720.90.50.5
Homocysteine, nmoles/ml10.810.80.70.9
Vitamin B12, pg/ml4924470.0020.005
Fibrinogen, mg/dl3683430.020.002
CRP, mg/dl0.980.34<0.001<0.001
sICAM-1, ng/ml357269<0.001<0.001
sVCAM-1, ng/ml6486550.70.4
sTNFRI, pg/ml1,4591,232<0.001<0.001
sTNFRII, pg/ml2,9812,432<0.001<0.001
Osteoprotegerin, pmoles/liter14.212.70.020.01
Anticardiolipin antibody, units/ml4.64.00.90.8

Several sets of secondary analyses were conducted. Blood was drawn while some of the women were in a fasting state (>8 hours since last eating): 57% of those with RA and 63% of those without RA. When we restricted the comparison of triglyceride levels to only the women who were fasting, the mean levels between the 2 groups were still not statistically significantly different (P = 0.9).

DISCUSSION

We examined clinical risk factors for CVD in a cross-sectional study of women with and without RA who were enrolled in the Nurses' Health Study. Biomarkers of CVD risk, including both traditional lipid parameters and more novel markers of inflammation, were also assessed in a subgroup of these women. No significant differences between these groups were observed for current smoking status, BMI, menopause status, aspirin use, diabetes, hypertension, and family history of early myocardial infarction. However, women with RA were more likely to have been smokers in the past. In the biospecimen cohort, lipid profiles, except for HDL, were similar in both study groups. Several biomarkers of inflammation that have been linked to CVD were significantly higher in women with RA, including CRP, fibrinogen, sICAM-1, sTNFRI, sTNFRII, and OPG. Possibly because of the frequent use of folate supplementation among patients who take methotrexate, the levels of homocysteine were similar despite small differences in vitamin B12 levels.

We conducted this study to examine whether traditional clinical risk factors for CVD as well as newer biomarkers, such as sTNFRI and sICAM-1, could potentially explain the elevated risk of CVD in women with RA. These data suggest that the frequencies of most of the traditional clinical risk factors for CVD are similarly distributed in RA patients and controls. However, in this study, several biomarkers of inflammation that have been linked to CVD outcomes were higher in women with RA, suggesting that the excess risk of CVD could be related to inflammation.

Several epidemiologic studies have found an association between CRP and both CVD and peripheral arterial disease in the general population (12, 13). Other biomarkers of inflammation, such as fibrinogen, sICAM-1, sVCAM-1, sTNFRI, and sTNFRII, have also been linked to an increased risk of CVD (14–17). Anticardiolipin antibodies are associated with CVD end points in systemic lupus erythematosus (18, 19). Elevations in OPG levels, a soluble decoy receptor associated with osteoclastogenesis and rheumatoid bone erosions, have been linked to CVD (20, 21). The evidence for a link between inflammation and CVD in the general population supports our hypothesis that systemic inflammation in RA may underlie the increased risk of CVD in this population. Other data supporting this hypothesis include the positive association between levels of acute-phase reactants (erythrocyte sedimentation rates, CRP levels) and carotid intima-medial thickness in RA (22), as well as the observation that several disease-modifying antirheumatic drugs (DMARDs) used in RA have been associated with improved endothelial function (23, 24) and lower cardiovascular mortality rates (25). It may be important in future studies to determine the source of increased inflammatory biomarkers (synovium versus endovasculature). We were not able to make this distinction using our present data.

There are potential limitations of our methods. First, this study was cross-sectional and did not relate the clinical CVD risk factors or biomarkers to end points, such as acute myocardial infarction. There were too few clinical events in the biospecimen cohort to conduct this type of analysis. Thus, these data do not establish a causal link between any of the factors studied and CVD, but suggest possible relationships. Second, only female nurses between the ages of 43 and 68 years were included in the study population, and this may limit the generalizability of our findings. Third, at the time blood was drawn, information about the use of several classes of medications commonly prescribed for patients with RA, including glucocorticoids and DMARDs was largely unavailable in the database. These drugs may influence the levels and distribution of the markers examined in this study. Finally, biomarker data were available on a relatively small number of women with RA, who were chosen from the entire cohort based on a priori criteria. This study should be repeated in a larger number of “unselected” women and men with RA.

In this analysis of women enrolled in the Nurses' Health Study, clinical risk factors for CVD were similar in women with and in women without RA. However, many biomarkers associated with inflammation and CVD were significantly elevated in women with RA. These data support the hypothesis that the increased risk of CVD observed in patients with RA may be due to underlying inflammation. Future studies that include clinical risk factors, biomarkers, and medication use in predictive models of CVD will help to determine the relative contribution of each of these factors. However, based on these data and those from previous studies (1, 2), we anticipate that the differences seen in biomarker levels between women with and without RA will underlie much of the difference in the risk of CVD in RA.

Acknowledgements

Elaine Coughlan provided excellent statistical programming support. Staff in Dr. Nader Rifai's laboratory ran all of the biomarker assays. Dr. Lisa Mandl provided valuable help validating the RA cases.

Ancillary