A lactylation-driven prognostic model for lung adenocarcinoma: cellular lactylation heterogeneity and immune insights.

BACKGROUND: Lactylation, a recently identified post-translational modification, plays a critical role in tumor progression and immune regulation. However, its cellular heterogeneity and functional impact in lung adenocarcinoma (LUAD) remain poorly understood. This study was designed as exploratory biological research to characterize lactylation-associated patterns at the single-cell level and to propose a potential lactylation-related prognostic model. METHODS: Single-cell transcriptomic data from LUAD and normal lung tissues were analyzed to quantify lactylation activity using AUCell based on 332 lactylation-related genes. Cell-cell communication was inferred using CellChat to identify ligand-receptor interactions among subpopulations. Candidate genes were selected by integrating ligand-receptor pair genes, marker genes from highly lactylated subtypes, and previously reported lactylation-related genes. A total of 101 machine learning model combinations were evaluated to construct the prognostic model. Selected genes were further validated by quantitative reverse transcription polymerase chain reaction (qRT-PCR), and the potential relationship between ANGPTL4 expression and lactylation was examined. Immune characteristics were evaluated using multiple established computational approaches for estimating immune infiltration, dysfunction, and immunogenicity. RESULTS: Lactylation activity was higher in tumor epithelial and stromal cells, with particularly elevated levels in specific epithelial subpopulations. A 12-gene signature was identified, comprising nine risk genes (e.g., ANGPTL4, SOD1, and VEGFC) and three protective genes. The random survival forest (RSF) model demonstrated strong predictive performance [area under the curve (AUC) =0.99 for training, 0.68 for testing], and qRT-PCR validation largely confirmed the predicted gene expression patterns. High-risk patients exhibited poorer survival, reduced immune infiltration, increased immune exclusion, and higher enrichment of immunosuppressive cells. Moreover, ANGPTL4 expression positively correlated with lactylation-associated indicators. Cell-cell communication analysis further highlighted signaling pathways involving the identified genes. CONCLUSIONS: This study presents a lactylation-based prognostic model for LUAD and uncovers potential immune-related mechanisms by which highly lactylated epithelial cells may contribute to immune evasion and tumor progression.