Olink proteomic inflammatory biomarkers enhance precision risk prediction for long-term outcomes in coronary artery disease
K Zhang, K Yan, M Liu, F Xu, Y Yang, J Yuan, X ZhaoAbstract
Background
Chronic inflammation plays a critical role in the pathogenesis of coronary artery disease (CAD), yet its representation in current clinical risk models remains limited. With the advancement of proteomics technologies, novel inflammatory biomarkers may be identified to enhance risk prediction in CAD patients.
Purpose
To identify novel inflammatory biomarkers associated with long-term outcomes in CAD using Olink proteomics and to evaluate their incremental prognostic value beyond traditional clinical risk factors.
Methods
This study included 250 patients with CAD derived from a prospective multicenter cohort, including 125 who experienced major adverse cardiac and cerebrovascular events (MACCE) and 125 without events. MACCE was defined as a composite of all-cause mortality, myocardial infarction, stroke, and in-stent thrombosis. The cohort was randomly divided into a training set (80%) and an internal validation set (20%). Proteomic profiling was performed using the Olink Target 96 Inflammation panel, which measures 92 inflammatory-related proteins. Feature selection was performed using LASSO, and the optimal risk prediction model was determined through recursive feature elimination.
Results
Among the 250 patients, the mean age was 63.5±11.0 years and 182 (72.8%) were male. Differential expression analysis identified 28 proteins associated with MACCE, among which 25 were upregulated and 3 were downregulated. Seven inflammatory proteins were ultimately selected into the final model:IL6, FGF_21, CDCP1, TNF, FGF_23, TNFB, and TRAIL. The proteomics-based risk model achieved a C-index of 0.68 in the training set and 0.73 in the validation set, with AUC values of 0.71 and 0.75, respectively. Incorporation of the proteomic signature into a clinical model (including sex, AMI status, hyperlipidemia, and left ventricular ejection fraction) significantly improved model discrimination and reclassification. In the validation cohort, the AUC increased from 0.7143 to 0.7537 (P < 0.001), with a net reclassification improvement (NRI) of 41.05% and an integrated discrimination improvement (IDI) of 8.14%.
Conclusion
Inflammatory proteins identified by Olink proteomics alone effectively predicted long-term risk in CAD patients and provided incremental predictive value when added to traditional clinical risk factors. These findings support the utility of protein-based biomarkers for precise assessment of residual inflammatory risk and optimization of patient management.Figure 1.Differentially expressed proteinsFigure 2.ROC curves in validation cohort