Integrated clinical and multi-omics analysis links composite inflammatory indices to macrophage-associated molecular...

Aortic dissection (AD) is a life-threatening vascular disease characterized by high mortality and complex pathophysiology involving inflammation and vascular wall degeneration. Composite inflammatory indices derived from routine blood tests have shown prognostic value in AD; however, their role in disease assessment and their integration with molecular mechanisms remain unclear.

A total of 288 participants (144 AD patients and 144 controls) were enrolled for clinical analysis. Six composite inflammatory indices (NLR, MLR, SIRI, SII, PHR, and AISI) were calculated and evaluated using logistic regression and restricted cubic spline (RCS) models. Generalized propensity score (GPS) weighting was applied to reduce confounding. Multi-omics analyses were conducted by integrating single-cell RNA sequencing (GSE213740), bulk transcriptomic data (GSE153434), and inflammation-related gene sets. Weighted gene co-expression network analysis (WGCNA) and differential expression analysis were performed to identify candidate genes. Machine learning algorithms (LASSO, random forest, and SVM-RFE) were used for feature selection. External validation (GSE52093) and RT-qPCR experiments were conducted to verify feature genes.

All six inflammatory indices were significantly elevated in AD patients (all P < 0.001) and independently associated with AD. RCS analysis revealed distinct nonlinear patterns, with AISI and SII showing continuous dose-response relationships, whereas NLR, MLR, and SIRI exhibited threshold effects. Associations remained robust across subgroups and after GPS weighting, with improved covariate balance (correlation coefficients <0.1). Single-cell analysis identified macrophages as the most prominently altered cell population, with 1,063 differentially expressed genes indicating extensive transcriptional reprogramming. Bulk RNA-seq analysis identified 2,766 DEGs (1,331 upregulated and 1,435 downregulated), and WGCNA revealed a key module (2,551 genes) strongly associated with AD (r = 0.98). Integrative analysis yielded 25 candidate genes, from which four genes (HIF1A, ITGA5, PLAUR, and TLR2) were consistently selected by machine learning. External validation and RT-qPCR confirmed significant upregulation of HIF1A, ITGA5, and PLAUR in AD tissues.

Composite inflammatory indices are strongly associated with AD risk, and inflammatory-associated genes, particularly HIF1A, ITGA5, and PLAUR, may serve as potential diagnostic biomarkers and mechanistic targets.