Multi-omics machine learning-driven investigation of super-enhancers signatures and prognostic biomarkers in lung adenocarcinoma.

Lung adenocarcinoma (LUAD) is the most common histological subtype of malignant lung tumors, characterized by high incidence and mortality rates. Super-enhancers (SEs) are involved in regulating tumor transcription and promote tumorigenesis and progression. However, their role in LUAD remains underexplored. This study uses bulk, single-cell, and spatial transcriptomics analyses to uncover their tumor-promoting and tumor microenvironment-regulating features in LUAD. By integrating multi-omics data with 10 clustering algorithms, the study successfully identified molecular subtypes of SEs in LUAD. Analysis showed significant differences among patients with different subtypes in prognosis, genomic instability, and drug sensitivity. Subsequently, the Lasso + StepCox algorithm identified nine super-enhancer-related prognostic genes and built the super-enhancer-associated gene risk score (SEGRS) prognostic model. In-depth analysis revealed that SEGRS is closely linked to genomic mutations and oncogenic pathway enrichment in lung adenocarcinoma. SEGRS also plays a role in regulating the tumor microenvironment and immune response patterns. Finally, differential analysis, survival analysis, and spatial transcriptomics pinpointed the core gene HSPD1. In vitro experiments confirmed its high expression in LUAD and its role in promoting malignant progression. This work uncovers pro-tumorigenic SEs signatures in LUAD through a multi-omics and machine learning integrative approach, establishes new molecular subtypes and prognostic models, and clarifies the prognostic and cancer-promoting roles of HSPD1. These findings offer strategies and potential new targets for the precise diagnosis and treatment of LUAD.