Integrated analysis of bulk and single-cell RNA sequencing data to identify an oxidative phosphorylation-based prognostic signature for lung adenocarcinoma.

e20059 Background: Lung adenocarcinoma (LUAD) remains a leading cause of cancer-related mortality globally. Despite treatment advances, challenges for drug resistance and low survival rates persist, highlighting the need for novel prognostic biomarkers. Oxidative phosphorylation (OXPHOS) has been implicated in cancer progression, drug resistance, and immune modulation. This study aims to develop an OXPHOS-based prognostic model for LUAD and explore its underlying mechanisms. Methods: We utilized RNA-seq data from TCGA-LUAD as the discovery cohort, with two Gene Expression Omnibus (GEO) datasets as validation cohorts. OXPHOS-related genes were obtained from MitoCarta3.0. Prognostic biomarkers were identified through LASSO-Cox regression. Immune cell infiltration was assessed by CIBERSORT. Single-cell RNA-seq data from two GEO datasets was analyzed to explore biomarkers expression and their regulatory mechanisms at a single-cell resolution, using Gene set enrichment (GSEA) and CellChat. Results: A total of 43 OXPHOS-related genes associated with overall survival were identified based on discovery cohort. Among them, 13 genes— ATP5F1D , CMC2 , COA6 , COX6B2 , CYCS , DMAC2 , FMC1 , NDUFA10 , PET100 , RAB5IF , TIMMDC1 , UQCC3 and UQCRB —were selected as biomarkers for constructing a prognostic risk model using LASSO-Cox regression analysis. Higher risk scores were associated with poor prognosis, with the AUC scores for 1-, 3-, and 5-year survival in validation cohorts as follows: 0.69, 0.62, and 0.53 for one cohort, and 0.61, 0.64, and 0.51 for the other. Additionally, a nomogram incorporating risk scores and clinical data demonstrated strong predictive power. Furthermore, we found that low-risk group had higher immune cell infiltration levels, including activated B cells, CD4+ T cells, eosinophils, immature B cells, mast cells and natural killer cells (p < 0.0001). UQCC3 and RAB5IF expression levels in epithelial cells was associated with metastatic lesions and higher CNV scores. Both UQCC3+ and RAB5IF+ epithelial cells showed significantly enriched in pathways E2F targets and G2M checkpoint. Cell chat analysis revealed stronger communication between both UQCC3+ and RAB5IF+ epithelial cells and immune cells in TGF-β and Galectin pathways, especially with B cells, mast cells and T cells via LGALS9 binding to CD45 and CD44. Conclusions: We constructed an OXPHOS-based prognostic signature and simultaneously explored the functional mechanisms of potential biomarkers at a single-cell resolution, suggesting UQCC3 and RAB5IF expressed in epithelial cells and their immune regulatory roles in the tumor microenvironment. These findings may offer insights into the development of robust OXPHOS-related biomarkers and aid in the diagnosis, outcome prediction, and targeted therapies for lung cancer.