Prognostic model construction and mechanistic analysis of pancreatic adenocarcinoma based on RCD-related genes and immune microenvironment features

Abstract Background Regulated cell death (RCD) has been extensively implicated in the pathogenesis, progression, and immunosuppressive tumor microenvironment (TME) of pancreatic adenocarcinoma (PAAD). However, systematic investigations into the prognostic relevance of RCD-related genes in PAAD remain limited. This study aimed to construct an RCD-based prognostic model and delineate the mechanistic role of integrin β4 ( ITGB4 ) in immunosuppressive TME remodeling. Methods RNA sequencing (RNA-seq) data from PAAD specimens were obtained from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) databases. We identified differentially expressed RCD-related genes (DE-RCDs) and classified molecular subtypes via consensus clustering. Survival disparities and immune microenvironment profiles were comparatively analyzed. A prognostic model (RCD score) was developed via Least Absolute Shrinkage and Selection Operator (LASSO) penalized Cox regression, integrating ITGB4, MUC1, and ATG16L2 , and validated in independent GEO cohorts. Tumor-infiltrating immune cells were quantified using the CIBERSORT algorithm, while immunotherapy responsiveness was predicted via the Tumor Immune Dysfunction and Exclusion (TIDE) framework. ITGB4 expression and tumor-associated macrophages (TAMs) polarization states were assessed via multiplex immunofluorescence co-staining. Results Thirteen genes were significantly upregulated (adj. P < 0.05; |log 2 FC|> 1.5). The C2 subtype exhibited superior overall survival (hazard ratio [HR] = 0.62, 95% confidence interval [CI]: 0.41–0.93; P = 0.0062), accompanied by heightened infiltration of CD8 + cytotoxic T cells and M1-polarized TAMs ( P < 0.001). The RCD score independently predicted PAAD prognosis and improved 3-year survival prediction accuracy when combined with TNM staging (misclassification rate < 5%). Low-risk patients demonstrated enhanced responsiveness to programmed death-ligand 1 (PD-L1) inhibitors (odds ratio [OR] = 2.17, 95% CI 1.32–3.56; P = 0.014). Immunofluorescence confirmed ITGB4 overexpression in PAAD tissues, demonstrating a robust correlation with M2-polarized TAMs (r = 0.9168, P < 0.0001). Conclusion The RCD score-based signature demonstrates robust predictive utility for PAAD outcomes. Clinically, the RCD score could serve as a complementary biomarker to CA19-9 and TNM staging, potentially enhancing early risk stratification and informing personalized immunotherapy decisions for PAAD patients. RCD-related genes orchestrate immunosuppressive TME remodeling, with ITGB4 functioning as a central regulator of tumor progression via M2-polarized TAM modulation.