The aim of this study is to evaluate the relationship between novel composite metabolic indices (e.g., AIP, NHHR, CMI, LAP, CVAI, TyG) and incident CVD in RA patients, as well as whether those novel composite metabolic indices outperform traditional single lipid measures (e.g., TC, HDL-C) in predicting CVD risk in RA patients.

RA patients free of baseline CVD were included and followed up in a prospective RA cohort from January 1, 2001 to June 30, 2025 at the Department of Rheumatology and Immunology, Sun Yat-sen Memorial Hospital, Guangdong, China. Demographic characteristics and clinical data were collected, and 7 composite metabolic indicators at baseline were calculated. The endpoint was CVD, which included coronary heart disease, stroke, heart failure, peripheral arterial disease, and CVD death. Cox proportional hazards models were employed to estimate hazard ratios (HR) and 95% confidence intervals (CI) for the associations between novel composite metabolic indicators (analyzed as quartile categorical) and incident CVD. The discriminative ability of 13 indicators including 7 composite metabolic indicators and 6 individual traditional risk factors (TC, TG, HDL-C, LDL-C, FPG, WC) for predicting 10-year CVD risk was evaluated using receiver operating characteristic (ROC) curve analysis. Area under the curve (AUC) values with 95% CI were calculated using non-parametric methods and visualized in a horizontal bar plot with reference lines at AUC=0.5 (random prediction) and AUC=0.7 (good prediction threshold). 

A total of 1,438 RA patients were enrolled, with a mean age of 52.3 ± 12.7 years at baseline, 82.7% females, and median RA disease duration 3.6 years. During a median of 5.9 (3.3-9.3) years, 147 (10.2%) CVD events occurred. After adjusting for potential confounding, Cox regression analysis showed that all composite metabolic indicators except NHHR and TC/HDL-C ratio were significantly associated with CVD in RA patients (Figure 1A, all P < 0.05). Notably, CVAI index demonstrated a nonlinear relationship with CVD risk (P for nonlinear = 0.01, Figure 1B). Its association peaked in the second quartile (HR=2.44, 95% CI: 1.06-5.62, P = 0.04), with declining and non-significant estimates in the third (HR=2.29, 95% CI: 0.98-5.35, P = 0.06) and fourth quartiles (HR=1.83, 95% CI: 0.68-4.89, P = 0.23, Figure 1A). In contrast, the other composite metabolic indicators, including TyG, CMI, AIP, and LAP exhibited a clear dose-response relationship. Compared with the first quartile, TyG was significantly associated with CVD in Q3 (HR = 2.15, 95% CI: 1.21-3.83, P = <0.01) and Q4 (HR = 2.40, 95% CI: 1.38-4.17, P = <0.01). LAP also exhibited significant associations in Q3 (HR = 2.41, 95% CI: 1.40-4.14, P = <0.01) and Q4 (HR = 2.08, 95% CI: 1.20-3.61, P = 0.01). Both CMI and AIP reached significance only in the highest quartile (CMI: HR = 1.91, 95% CI: 1.12-3.26; AIP: HR = 1.83, 95% CI: 1.09-3.06; all P < 0.05; Figure 1A). Overall, composite metabolic indicators showed higher discriminatory accuracy than individual traditional risk factors (all P < 0.05, Table 1). CVAI exhibited the strongest discriminative ability (AUC = 0.74, 95% CI: 0.69-0.79), with an optimal cutoff of 77.11 corresponding to 79.0% sensitivity and 60.7% specificity. The TyG index ranked second (AUC=0.69, 95% CI: 0.63-0.74). CVAI also outperformed the other composite metabolic indicators in AUC comparisons (all P < 0.05), with the exception of TyG.

Composite metabolic indicators, especially CVAI or TyG, deliver a more complete cardiovascular risk assessment than single traditional biomarkers. Therefore, both of them could be considered for integration into routine cardiovascular risk management protocols for RA patients. This integration would enhance risk stratification and facilitate early, targeted interventions.