Patients with rheumatoid arthritis (RA) have a higher cardiovascular risk (CVR) compared to the general population contributing to a decreased life expectancy. Systematic and periodic assessment of CVR is recommended in all patients with RA. Several CVR algorithms developed using data from the general population do not perform well in patients with RA, and only the QRISK3 (American Heart Association, Dallas, TX, USA) and Expanded Cardiovascular Risk Prediction Score for Rheumatoid Arthritis (ERS-RA) includes RA as a variable [1]. The newly developed Predicting Risk of Cardiovascular Disease EVENTs (PREVENT; American Heart Association, Dallas, TX, USA) calculator was designed to estimate the absolute risk of cardiovascular disease (CVD) and provides an estimate of atherosclerotic CVD and heart failure [2]. This study aimed to evaluate the sensitivity and specificity of the new PREVENT algorithm to identify carotid plaque (CP) in RA patients without prior CVD and to compare these results with other CVR algorithms.

A cross-sectional study was carried out that included patients recruited in a rheumatology outpatient clinic at the “Dr. José Eleuterio González” University Hospital. Patients were 40 to 79 years old and fulfilled the 2010 American College of Rheumatology (ACR)/European Alliance of Associations for Rheumatology (EULAR) Classification Criteria for RA. Patients with previous CVD (myocardial infarction, stroke, or peripheral artery disease), pregnancy, or overlap syndrome were excluded. A B-mode carotid ultrasound was performed on all patients. The presence of CP was defined as diffuse carotid intima-media thickness ≥1.2 mm or focal thickness ≥0.5 mm [3,4]. The CVR was evaluated using Globorisk (Harvard School of Public Health, Boston, MA, USA), HEARTS (World Health Organization, Geneva, Switzerland), QRISK3, ERS-RA, SCORE2 (European Society of Cardiology, Sophia Antipolis, France), and PREVENT algorithms [1,2,5,6]. The result was multiplied by 1.5 in algorithms where RA is not included as a variable, according to EULAR 2015/2016 recommendations [7]. Receiver operating characteristic (ROC) curve analysis was performed to evaluate the different CVR algorithms’ performance to identify CP. Youden index (Youden’s J statistic) was calculated to select the optimum cutoff point and posteriorly calculate sensitivity, specificity, and predictive values. A p-value of ≤0.05 was considered statistically significant. Statistical analysis was performed with IBM SPSS version 29.0 (IBM Co., Armonk, NY, USA). This study was approved by the Ethics and Research Committee of the School of Medicine and University Hospital of the Autonomus University of Nuevo León (MI14-006), and it was conducted following the ethical standards outlined in the Declaration of Helsinki and its subsequent amendments.

A total of 261 patients with RA were included, mostly females (n=244, 93.5%), with a mean age of 56.0±9.3 years. The median disease activity measured by Disease Activity Score 28 (DAS28)- C-reactive protein (CRP) was 3.3 (2.1~4.4) and the median disease duration was 7.7 (3.0~14.7) years. Dyslipidemia was the most prevalent CVR factor (n=104, 39.8%). The prevalence of CP was 38.0%. According to the ROC curves, the Globorisk, HEARTS, and QRISK3 algorithms showed a higher area under the curve than the other algorithms. The HEARTS and QRISK3 algorithms showed the highest positive likelihood ratios, showing rates of 1.87 and 1.84, respectively, to identify RA patients with CP (Figure 1).

Our study showed that the Globorisk, HEARTS, and QRISK3 algorithms demonstrated better diagnostic accuracy for CP detection in patients with RA. This observation might be attributed to the HEARTS and Globorisk algorithms being designed specifically for the Hispanic population, and the QRISK3 algorithm including RA as a variable. However, it’s noteworthy that the cutoff points established to identify patients with CP initially categorize individuals into the low or intermediate-risk category across all algorithms employed in this study. Previous studies have described the underestimation of CVR in RA patients with conventional calculators [8]. The new PREVENT algorithm does not seem to be superior to the other calculators in our cohort to detect subclinical atherosclerosis. Imaging studies evaluating coronary artery calcium and CP may offer better alternatives in high-risk populations like patients with RA [7].

In conclusion, our study showed that Globorisk, HEARTS, and QRISK3 calculators presented the best diagnostic accuracy to detect CP in patients with RA. The novel CVR calculator PREVENT did not show better performance than older calculators in our population.