Background/Purpose: Cardiovascular diseases (CVD) are the first cause of death in patients with rheumatoid arthritis (RA). The guidelines recommend risk prediction in all RA patients. However, there is currently no precise prediction model for RA patients. Coronary artery calcium (CAC) score by coronary angiography or chest computed tomography (CT) can improve CVD risk prediction in the general population, but no report in RA patients. We aimed to investigate whether adding CAC score to risk algorithms could improve CVD risk prediction in RA patients.

Methods: RA patients from January 1, 2010 to September 30, 2024 were included retrospectively from a RA cohort at the Department of Rheumatology and Immunology, Sun Yat-sen Memorial Hospital, Guangdong, China (SYSEC-2009-06). The inclusion criteria included free of CVD at baseline, and having intact clinical data. CAC scores based on chest CT at baseline were calculated by the Agatston method. The endpoint was CVD, which included coronary heart disease, stroke, heart failure, peripheral arterial disease, and CVD death. LASSO-Cox regression were performed to construct prediction model for CVD. The list of candidate covariates included traditional CVD risk factors (age, sex, BMI, history of hypertension and diabetes, TC, HDL-C, LDL-C), RA-related factors (disease duration, RF, ACPA, ESR, CRP, CDAI, HAQ-DI, and the use of glucocorticoid, csDMARDs, and biologic agents), and CAC score. Harrell’s C-statistic and NRI were used to compare the performance between models with and without CAC, as well as existing RA specific model (ESR-RA).

Results: A total of 822 RA patients were enrolled in this study, with mean age at baseline 54.5 ± 12.0 years, 79.8% females, and median RA disease duration 6.3 years. During a mean follow-up time of 3.9 years, 81 (9.8%) incident CVD events occurred. The LASSO-Cox regression was performed to identify risk factors associated with CVD, and the final clinical model consisted of age, male, BMI, history of hypertension, CRP, CDAI, the use of glucocorticoid, and CAC score (Table 1). Table 2 illustrates the internal discrimination properties of CAC model, with a C-index of 0.888 (95% CI 0.844-0.933), indicating excellent discrimination between events and non-events. The C-index of CAC model was significantly higher than that of the clinical model (0.859) and the ESR-RA model (0.847, both P < 0.05). While, no significant difference was found between the clinical and ESR-RA models, supporting the significant improvement achieved by incorporating CAC. When comparing the CAC model against clinical model, the NRI showed that the accuracy of the model was improved 31.6% (95% CI 15.9-0.62.8) after adding CAC scores to the clinical model, while no significant improvement was observed when comparing the clinical model with the ESR-RA model (Table 2).

Conclusion: Add-on of CAC score significantly improves the prediction of CVD risk in RA patients. Further validation in large-sample, multi-center prospective cohorts are needed in future.

Table 1 The CVD Risk Prediction Models in RA patients

Table 2 The ability of CAC score to predict CVD risk in RA patients

« Back to ACR Convergence 2025

ACR Meeting Abstracts - https://acrabstracts.org/abstract/add-on-of-coronary-artery-calcium-score-significantly-improves-the-risk-prediction-of-cardiovascular-diseases-in-patients-with-rheumatoid-arthritis-from-a-real-world-cohort-study/