Spatial decoupling of plaque and circulating immune programs defines a robust gene signature in atherosclerosis

Atherosclerosis (AS) is driven by chronic vascular inflammation, yet the relationship between immune activity within atherosclerotic lesions and systemic immune signals in peripheral blood remains poorly defined. In this study, we performed an integrative multi-cohort analysis of gene expression profiles from atherosclerotic tissues, peripheral blood, and single-cell transcriptomic datasets to characterize spatial immune heterogeneity in AS. Differential expression and enrichment analyses revealed divergent immune-related transcriptional patterns between plaques and blood, indicating a dissociation between local and systemic immune responses. Using weighted gene co-expression network analysis and multiple machine learning approaches, we identified a 5-gene signature (with XAF1 [XIAP-associated factor 1], RNF213 [ring finger protein 213], OAS2 [2'-5'-oligoadenylate synthetase 2], and OAS3 [2'-5'-oligoadenylate synthetase 3] upregulated, and BEND5 [BEN domain containing 5]) downregulated in plaques) that robustly distinguished atherosclerotic from control samples across multiple vascular beds. This gene signature demonstrated strong diagnostic performance in tissue datasets. While its diagnostic accuracy was attenuated in peripheral blood, it retained moderate prognostic value for ischemic stroke risk, suggesting its utility as a hypothesis-generating tool for noninvasive risk stratification. Together, these findings highlight spatially distinct immune programs in AS and support the use of integrated tissue- and blood-based gene signatures for improved disease diagnosis and risk stratification.