Comprehensive Analysis of Epigenetic Signatures in Non‐Small Cell Lung Cancer: Development and Validation of an Epigenetics‐Based Prognostic Model for Drug Sensitivity Prediction

Non-small cell lung cancer (NSCLC) exhibits complex epigenetic dysregulation that impacts treatment response and prognosis, yet comprehensive analysis linking epigenetic signatures to clinical outcomes remains limited. We integrated single-cell RNA sequencing data from 42 NSCLC and 11 normal samples with bulk transcriptomics from multiple cohorts (TCGA-NSCLC [n = 993], GSE13213 [n = 110], GSE42127 [n = 176]). Cell types were annotated using scMayoMap and validated through marker gene analysis. Epigenetic patterns across 15 cell types were characterized using single-sample Gene Set Enrichment Analysis (ssGSEA). Through weighted gene co-expression network analysis (WGCNA), we identified key epigenetic regulatory modules and their associated genes. We systematically evaluated 111 machine learning algorithms to develop an epigenetic-based risk stratification model, with Random Survival Forest (RSF) emerging as the optimal approach. Experimental validation confirmed significant upregulation of key model genes at both mRNA (qRT-PCR: SLC2A1, LAD1, LYPD3; all p < 0.01) and protein levels (immunohistochemistry) in tumor tissues compared to adjacent normal tissues. Furthermore, functional experiments demonstrated that overexpression of LYPD3 significantly promoted NSCLC cell proliferation, migration, and invasion in vitro, corroborating computational findings and providing strong mechanistic validation of the model. Drug sensitivity analysis revealed differential therapeutic vulnerabilities, with high-risk patients showing increased sensitivity to EGFR-TKIs, including Gefitinib (p < 0.001). The model demonstrated significant prognostic value in kidney chromophobe (p = 0.040), kidney clear cell carcinoma (p < 0.001), and cervical squamous cell carcinoma (p = 0.004). Our study establishes a robust epigenetic-based prognostic model for NSCLC and identifies LYPD3 as a novel oncogenic driver, providing insights into tumor biology and treatment response, offering potential clinical utility for personalized therapeutic strategies.