Divergent immune and extracellular matrix transcriptional programs underlie poor prognosis in early-stage lung adenocarcinoma

Background: Early-stage lung adenocarcinoma (LUAD) patients with similar pathological stage exhibit heterogeneous recurrence risk. Prior work suggested that intra-tumoral immune composition, particularly B cells, regulatory T cells, and macrophages stratify recurrence outcomes. These immune programs operate in the context of stromal and tumor-intrinsic transcriptomic programs. Objective: To identify coordinated gene expression programs linking immune, stromal and tumor-intrinsic processes to recurrence in early-stage (IA/IB) LUAD, and to assess their prognostic relevance in an independent cohort using overall survival. Methods: Gene expression data of 292 patients in NCBI GEO datasets was reanalyzed and unsupervised clusters were identified within the differentially expressed genes (DEGs) that associated with recurrence. Protective and risk-associated genes were identified in covariate-adjusted Cox models. GO enrichment was performed using DEGs. Pathway-level principal component scores were tested for association with overall survival in TCGA-LUAD early-stage, treatment-naive patients (n=90). Results: Modules enriched for B-cell and antibody responses, vascular homeostasis, and regulation of fibrotic remodeling were associated with reduced recurrence, whereas those enriched for epithelial organization and extracellular matrix (ECM) remodeling conferred increased risk. The contrast between ECM-driven recurrence risk and protective immune programs may reflect distinct biological states associated with recurrence timing, wherein changes in the ECM precede impaired lymphocyte responses in early-stage disease. In the TCGA cohort, immune-related programs associated with improved survival, whereas extracellular matrix (ECM) and epithelial differentiation/proliferation programs associated with worse survival. Significance: This study integrates immune, tumor, and stromal transcriptional programs to identify a prognostic dichotomy between ECM remodeling and adaptive immune-supportive states in early-stage LUAD. These coordinated tumor states and immune-matrix interactions offer mechanistic insight into recurrence and provide a framework for risk stratification.