Abstract
BACKGROUND: Endoplasmic reticulum stress (ERS) can affect the efficacy of anti-tumor therapy (chemotherapy, targeted therapy and immunotherapy) by regulating tumor immune microenvironment. This study aims to develop a prognostic prediction model based on the expression profiles of ERS-related genes to guide the treatment of lung adenocarcinoma (LUAD). METHODS: The 16-gene prognostic signature was established to predict the prognosis of LUAD patients using The Cancer Genome Atlas (TCGA) database. We applied consensus clustering and found that LUAD could be divided into two groups based on the expression of PIK3CG and DMD. We performed gene set enrichment analysis (GSEA) to identify functional differences, and used ESTIMATE, CIBERSORT, and single-sample GSEA (ssGSEA) to assess immune infiltration. In addition, we compared the expression of immunomodulatory targets between the two clusters. RESULTS: We successfully constructed a 16-gene prognostic signature and a nomogram to help individualize outcome prediction in LUAD. The ERS risk signature is an independent prognostic factor for LUAD patients, and a higher score indicates a poorer prognosis. Through consensus clustering based on the expression of PIK3CG and DMD, LUAD patients can be divided into two groups. Cluster 1, with high PIK3CG and low DMD expression, shows stronger immune infiltration and higher expression of immunomodulatory targets, suggesting a better response to immunotherapy compared to cluster 2. These findings were further validated in an independent cohort (GSE68465), confirming the reproducibility of the immune landscape distinction between the clusters. CONCLUSIONS: The ERS-associated gene signature can effectively predict the prognosis of LUAD patients. In addition, we found that PIK3CG and DMD play a certain role in tumor immunity and may be potential therapeutic targets.