Background/Purpose: Patients with systemic lupus erythematosus (SLE) have increased risk for atherosclerotic cardiovascular disease (CVD) events, often underrepresented in traditional risk models. We previously developed an SLE-specific 10-year CVD risk model, SLECRISK, incorporating SLE-specific variables (Choi et al., 2024). The American Heart Association (AHA) recently introduced the PREVENT (Predicting Risk of cardiovascular disease EVENT) algorithm to assess CVD risk (Khan et al., 2024). We compared the predictive performance of SLECRISK and PREVENT in our SLE cohort.

Methods: Adult SLE patients meeting ACR 1997 and/or EULAR criteria from our academic center were included. SLECRISK and PREVENT estimated 10-year risk of myocardial infarction, stroke, or cardiac death. Outcomes were adjudicated by two cardiologists. Discrimination was assessed using c-statistics. Model performance was compared by sensitivity, specificity, and AIC. Calibration was evaluated by decile-based observed vs. predicted risk plots. Risk categories were defined as ≤7.5% (low), 7.5–20% (moderate), or ≥20% (high). Demographics and clinical features of patients classified as moderate/high risk by SLECRISK vs. PREVENT were analyzed using t-tests and chi-square tests.

Results: In this cohort of 1,243 patients, PREVENT and SLECRISK yielded similar discrimination (AUC: 0.756 vs. 0.744; p = 0.68), but SLECRISK showed better model fit (AIC: 592.2 vs. 611.1). Figure 1 shows proportions of patients categorized into low, moderate, and high CVD risk by each model. Using a ≥7.5% cutoff, 581 patients (46.7%) were classified as moderate/high risk by either model: 491 by SLECRISK only, 11 by PREVENT only, and 79 by both. Patients flagged by SLECRISK alone were significantly younger (mean 42.5 vs. 57.5 years, p < 0.0001), had lower systolic BP (122.0 vs. 145.3 mmHg, p < 0.0001), creatinine (0.93 vs. 2.55 mg/dL, p < 0.0001), and LDL (101.9 vs. 110.4 mg/dL, p = 0.02), but higher prevalence of SLE autoantibodies, including anti-dsDNA (78.6% vs. 36.7%) and anti-RNP (55.7% vs. 27.8%), all significant (Table 1). Calibration analysis showed better concordance between predicted and observed CVD with SLECRISK (Fig. 2).

Conclusion: SLECRISK identified more at-risk SLE patients, effectively capturing younger individuals with greater immunologic activity and lower traditional cardiovascular burden. Despite similar AUCs, SLECRISK showed better calibration and model fit. These findings support integration of SLECRISK into clinical care and guidelines for CVD risk stratification in SLE.

Figure 1. Risk Category Classification of SLECRISK and PREVENT in definite (adjudicated) and probable MACE.

Table 1. Comparison of baseline characteristics between patients classified as moderate/high cardiovascular risk (7.5% and greater) by SLECRISK only versus by PREVENT only or by both models among 1,243 patients

Figure 2. Calibration plot of PREVENT and SLECRISK Models.

« Back to ACR Convergence 2025

ACR Meeting Abstracts - https://acrabstracts.org/abstract/cardiovascular-risk-prediction-in-systemic-lupus-erythematosus-comparing-the-prevent-and-slecrisk-models/