Lung adenocarcinoma (LUAD) is the most common histological type of lung cancer, characterized by high mortality, recurrence, and metastasis. Despite advancements in therapies such as surgery, targeted treatment and immunotherapy, therapeutic resistance and immune evasion remain significant challenges. In our study, we integrated multi-omics data, including spatial transcriptomics, single-cell RNA sequencing, H3K18la ChIP-seq, CRISPR data and bulk transcriptomics, to explore the metabolic heterogeneity of LUAD, particularly focusing on glycolysis and histone H3K18 lactylation (H3K18la). Our findings revealed significant intra- and inter-tumoral metabolic heterogeneities, with glycolysis and H3K18la-related genes being more active in tumor regions. We also identified H3K18la-related gene activities as a marker of LUAD progression, demonstrating its strong correlation with glycolysis and tumor cell phenotypes. Based on these insights, we developed a machine learning-based prognostic model (termed as “Kla.Sig”) that predicts patient survival and immunotherapy response, with validation across multiple cohorts. The model highlighted the immunosuppressive tumor microenvironment in high-risk score patients, with lower immune cell infiltration and higher immune evasion ability. In addition, we developed an online R shiny application “LUAD-Kla.Sig” to facilitate users’ estimation of survival based on Kla.Sig model. In-silico drug screening suggests that targeting Polo-like kinase 1 (PLK1) with BI-2536 could be an effective strategy for high-risk LUAD patients. This study offers a deeper understanding of LUAD metabolism and immune evasion at single-cell and spatial resolution, proposing potential therapeutic targets and a risk-stratified treatment strategy for precision medicine.
扫码关注我们
求助内容:
应助结果提醒方式:
