Abstract
Background and objectives
Patients with rheumatoid arthritis (RA) face an increased risk of cardiovascular disease (CVD) that traditional risk factors alone cannot fully explain. Chronic inflammation may influence lipid profiles and contribute to this risk. This study evaluates predictors of incident CVD in RA and explores how erythrocyte sedimentation rate (ESR) modifies the relationship between lipid levels and CVD outcomes.
Design, settings, participants, and measurements
This retrospective cohort study included 1,802 RA patients aged 40-79 years, diagnosed between 2015 and 2022, and free of CVD at diagnosis. We evaluated the association between traditional cardiovascular risk factors—including current smoking, diabetes mellitus, systolic blood pressure, body mass index (BMI), HDL cholesterol, and LDL cholesterol—and RA-specific inflammatory markers, including ESR and C-reactive protein (CRP), with the incidence of CVD. Cox proportional hazards models adjusted for age, sex, race/ethnicity, antihypertensive medications, lipid-lowering medications, and antiplatelet medications.
Results
During a median follow-up of 3.5 years, 187 patients (10.4 %) developed CVD. The mean BMI was 32 kg/m² (standard deviation [SD] 10), HDL cholesterol was 53 mg/dL (SD 17), and LDL cholesterol was 104 mg/dL (SD 37). The median ESR was 21 mm/hr (interquartile range [IQR] 11–42) and CRP was 6 mg/L (IQR 3–12). Higher LDL cholesterol was inversely associated with CVD risk (HR 0.77 per SD increase, 95 % CI 0.63–0.94), with this association weakening with increasing ESR levels (interaction term HR 0.84, 95 % CI 0.71–0.99). Elevated HDL cholesterol also showed significantly decreased CVD risk (HR 0.82 per SD increase, 95 % CI 0.68–0.97). Smoking and diabetes were associated with increased risks (HR 1.52, 95 % CI 1.07–2.17 and HR 2.08, 95 % CI 1.39–3.10, respectively).
Conclusion
This study highlights the complex interplay between lipid levels and inflammation in RA, highlighting the nuances of CVD risk assessment in RA.
1
Introduction
Rheumatoid arthritis (RA) significantly increases the risk of cardiovascular disease (CVD), which is a significant contributor to mortality and morbidity in this patient population [ ]. Traditional cardiovascular risk factors, such as dyslipidemia, do not fully account for the heightened cardiovascular risk observed in those with RA [ ]. This discrepancy suggests that RA-related inflammatory processes may have a crucial role in cardiovascular disease pathogenesis beyond what is explained by conventional risk factors alone [ ].
Although previous studies have explored the relationships between lipid levels, inflammatory markers, and cardiovascular outcomes in RA, the findings have been inconsistent [ ]. This variability highlights the need for further investigation into how these factors may interact and lead to the increased CVD observed among those with RA. Further, current cardiovascular risk assessments are predominantly designed for the general population and thus may not adequately reflect the complex interplay of risk factors in RA, potentially leading to inadequate management and prevention strategies in this group [ ]. In light of these inconsistencies and limitations, we conducted a retrospective cohort study in a diverse group of patients newly diagnosed with RA to examine the associations between traditional and RA-specific risk factors and the incidence of CVD. In addition, we explore the role of ESR in altering the association between lipid levels and CVD outcomes.
2
Methods
2.1
Study population
A retrospective cohort study was conducted at the University of Illinois Hospital and Health Sciences System (UI Health), which is defined by the United States Department of Education as a minority-serving institution [ ]. All patients aged 40 to 79 years diagnosed with RA, as indicated by International Classification of Diseases (ICD-10) codes M05-M06, from October 1, 2015, to June 1, 2022, during an outpatient encounter were included. Only patients with at least 1 encounter in an outpatient rheumatology clinic during the study period were included. Patients with documented CVD prior to or within 30 days of the index date of RA diagnosis (initial ICD-10 diagnosis) were excluded to reduce the likelihood of reverse causality. Additionally, patients with a prior diagnosis of RA before or within 30 days of their first encounter within the UI Health system were excluded from the study. The study was approved by the University of Illinois Chicago Institutional Review Board (UIC IRB #2021-0311).
2.2
Exposures and covariables
The electronic medical record (EMR) was utilized to obtain the first recorded measurement of demographic and clinical characteristics following the index date of diagnosis: age, sex, race and/or ethnicity, smoking status, antihypertensive medications, lipid-lowering medications, antiplatelet medications, systolic blood pressure, body mass index, LDL cholesterol, HDL cholesterol, diabetes mellitus, erythrocyte sedimentation rate (ESR), and C-reactive protein. If a variable was not recorded after the index date, the most recent measurement before the index date was obtained. The absence of a smoking history was assumed if a patient’s smoking status was not reported in the EMR. A diagnosis of diabetes mellitus was defined by either an ICD-10 diagnosis of diabetes mellitus (E08) or a hemoglobin A1c laboratory value greater than 6.4 %.
2.3
Outcomes
The primary endpoint included incident CVD. CVD was defined by ICD-10 codes for ischemic heart disease, myocardial infarction, stroke, heart failure, angina, peripheral artery disease, transient ischemic attack, abdominal aortic aneurysm, or history of carotid intervention (Supplemental Table 1). The EMR was also manually reviewed to verify death certificates. In a sensitivity analysis, we excluded heart failure from the composite outcome to evaluate for atherosclerotic CVD (ASCVD).
2.4
Statistical analyses
Characteristics of patients were summarized using the ‘gtsummary’ package in R with mean ± standard deviation (SD) or median (interquartile range) for continuous variables and percentages for categorical variables. Cox proportional hazard models were used to estimate the hazard ratios of the studied predictor variables, which included systolic blood pressure, smoking status, diabetes mellitus, body mass index, LDL cholesterol, HDL cholesterol, ESR, and C-reactive protein. We fit hierarchically adjusted models based on the biological and clinical plausibility of covariables as possible confounders: model 1 adjusted for age, sex, and race/ethnicity; model 2 further adjusted for antihypertensive medications, lipid-lowering medications, and antiplatelet medications). The ‘mice’ package (version 3.15.0) was employed to perform multiple imputations using predictive mean matching to handle missing covariate data. This process was carried out across ten imputed datasets and involved 100 iterations. The results from the imputed datasets were aggregated using Rubin’s rules [ ]. All analyses were performed using R (version 4.1.2), “Funny-Looking Kid.”
To evaluate the presence of multicollinearity among the predictor variables, we utilized two approaches: correlation matrices and variance inflation factor (VIF) analysis. The correlation matrix revealed no pairwise correlations exceeding 0.7. Furthermore, the VIF values for all variables were less than 5, indicating minimal multicollinearity within our dataset.
3
Results
3.1
Baseline characteristics of the study population
During the study period, 3,382 patients were diagnosed with RA. Among these patients, 1,802 individuals aged 40 to 79 years were free of CVD within 30 days following their RA diagnosis and included in the final study cohort. As shown in Table 1 , the mean age of patients at the time of RA diagnosis was 57 years, with a majority (79 %) being female. The racial/ethnic composition was predominantly non-Hispanic Black (41 %), followed by Hispanic (29 %), and non-Hispanic White (23 %). Notable clinical characteristics included 18 % smokers, a mean systolic blood pressure of 134 mmHg, average HDL cholesterol of 53 mg/dL, and LDL cholesterol of 104 mg/dL. Few patients were on common risk-modifying medications, including antihypertensive (8.5 %), lipid-lowering (3.4 %), and antiplatelet (5.3 %) agents.
Characteristics | N = 1,802 |
---|---|
Age, years | 57 ± 9 |
Male | 375 (21 %) |
Race/Ethnicity | |
Hispanic | 523 (29 %) |
Non-Hispanic White | 423 (23 %) |
Non-Hispanic Black | 745 (41 %) |
Other | 111 (6.2 %) |
Current smoking | 332 (18 %) |
Diabetes mellitus | 149 (8.3 %) |
Antihypertensive medication | 153 (8.5 %) |
Lipid-lowering medication | 61 (3.4 %) |
Antiplatelet medication | 96 (5.3 %) |
Systolic blood pressure, mmHg | 134 ± 21 |
Body mass index, kg/m 2 | 32 ± 10 |
HDL cholesterol, mg/dL | 53 ± 17 |
LDL cholesterol, mg/dL | 104 ±37 |
Erythrocyte sedimentation rate, mm/hr | 21 (11 – 42) |
C-reactive protein, mg/L | 6 (3 – 12) |
3.2
Risks of incident CVD
Over an average follow-up of 3.5 years, 187 patients (10.4 %) developed incident CVD, with ischemic heart disease (34 %) and myocardial infarction (30 %) being the most common causes (Supplement 2). There was a total of 4 deaths (0.22 %) in the final cohort, two of which were attributed to cardiac causes. In fully adjusted Cox proportional hazards models ( Table 2 ), current smoking (HR 1.52, 95 % CI 1.07–2.17) and diabetes mellitus (HR 2.08, 95 % CI 1.39–3.10) demonstrated significant associations with an increased risk of future CVD. Interestingly, increased levels of LDL cholesterol were associated with a reduced risk of cardiovascular events (HR 0.77 per SD, 95 % CI 0.63–0.94). Similarly, each SD increase in HDL cholesterol conferred a significant reduction in risk (HR 0.82, 95 % CI 0.68–0.97). Among RA-specific risk markers, an elevated ESR consistently showed an increased risk of the composite outcome (HR 1.27 per SD, 95 % CI 1.06–1.51). Other factors, including systolic blood pressure, body mass index, and C-reactive protein, did not exhibit significant relationships with future CVD. In a further analysis of ASCVD risk that excluded heart failure in the composite outcome, the magnitudes of association were qualitatively unchanged (Supplemental Table 3).
