Integrative single-cell and machine learning approach to characterize immunogenic cell death and tumor microenvironment in LUAD

作者：Han Zhang, Qiuqiao Mu, Yuhang Jiang, Xiaojiang Zhao, Xiaoteng Jia, Kai Wang, Xin Li, Daqiang Sun · 发表于：Journal of Translational Medicine · 年份：2025 · DOI：10.1186/s12967-025-06889-2 · 被引用次数：17 · 研究领域：Ferroptosis and cancer prognosis、Single-cell and spatial transcriptomics、Cancer Immunotherapy and Biomarkers

BACKGROUND: Immunogenic cell death (ICD) triggers antitumor immune responses and plays a critical role in shaping the tumor microenvironment (TME). However, its specific contribution to lung adenocarcinoma (LUAD) progression and immunotherapy response remains insufficiently explored. METHOD: We integrated single-cell RNA sequencing with machine learning to characterize ICD-related transcriptional features in LUAD. ICD activity was quantified across cell types using five scoring algorithms. To develop a robust prognostic model, we evaluated over 100 machine learning algorithm combinations and selected the CoxBoost + SuperPC approach based on the highest concordance index (C-index). The resulting ICD-related gene signature (ICDRS) was validated in six external cohorts. Downstream analyses included immune infiltration, mutation profiling, drug sensitivity, and immunotherapy response. SLC2A1 was selected for functional validation using qRT-PCR, CCK-8, Transwell, colony formation, and xenograft assays. RESULTS: Single-cell analysis revealed that macrophages exhibited the highest ICD activity and contributed significantly to intercellular communication. Based on ICD-associated genes, the ICDRS model consisting of 11 core genes was constructed and showed superior prognostic performance over 112 published LUAD signatures across multiple cohorts. The ICDRS stratified patients into distinct risk groups with significant differences in overall survival, immune infiltration patterns, and ...