RETN and BMP2 as potential biomarkers for prognosis and immune landscape in lung squamous cell carcinoma: from bioinformatics to clinical validation

Although immunotherapy has altered the treatment landscape for lung squamous cell carcinoma (LUSC), reliable biomarkers for prognosis and therapeutic response remain elusive. This study aimed to construct a robust immune-related gene signature to stratify LUSC patients. Immune-related differentially expressed genes were screened from TCGA and ImmPort databases. A prognostic model comprising Resistin (RETN), Bone Morphogenetic Protein 2 (BMP2), and Neuropeptide Y (NPY) was established via univariate and multivariate Cox regression analyses. The model was validated using the GEO dataset (GSE73403). Functional enrichment, immune infiltration, and drug sensitivity analyses were performed. Clinical validation was conducted using immunohistochemistry (IHC) on an independent cohort of 30 LUSC patients. The risk signature demonstrated stable prognostic accuracy in both training and validation sets. High risk scores were significantly associated with an immunosuppressive tumor microenvironment, characterized by neutrophil accumulation and upregulated immune checkpoints (e.g., CTLA-4, TIM-3), as well as resistance to standard chemotherapeutics like cisplatin. Single-gene analysis highlighted the crucial roles of RETN and BMP2 in immune modulation. Notably, clinical IHC validation revealed that BMP2 protein positivity was associated with favorable overall survival, presenting a distinct prognostic pattern compared to transcriptomic predictions, highlighting the biological complexity of LUSC. We constructed a novel immune-related risk signature that effectively predicts patient survival and therapeutic sensitivity. Despite the discrepancy between mRNA and protein levels for specific genes, the composite model serves as a valuable tool for optimizing personalized treatment strategies in LUSC.