ABCC2 作为肺腺癌的新型治疗靶点：一项机器学习驱动的发现，将氨代谢与预后和耐药性联系起来

Ammonia, long regarded as a metabolic waste product, has recently been recognized as a pivotal oncometabolite in the tumor microenvironment, contributing to cancer progression and immune evasion. However, its prognostic value and therapeutic relevance in lung adenocarcinoma (LUAD) remain insufficiently characterized.

Transcriptomic data from multiple LUAD cohorts were obtained from public databases. Ten machine learning algorithms were integrated into 101 combinations to construct predictive models and identify key genes associated with prognosis. The tumor immune microenvironment (TIME) and immunotherapy sensitivity were evaluated using established computational methods. The functional impact of the top candidate gene was validated through in vitro and in vivo experiments. Additionally, candidate agents whose efficacy correlates with ABCC2 expression were screened, and molecular docking was performed to analyze binding affinity and interaction modes.

The Ammonia Metabolism Score (AMs) emerged as an independent prognostic index for LUAD. A high AMs was associated with a suppressed TIME, characterized by fewer tumor-infiltrating lymphocytes such as CD8⁺ T cells, and showed resistance to immunotherapy. Consistently, ABCC2 itself also demonstrated significant potential as a prognostic biomarker in LUAD. Overexpression of ABCC2 altered the expression of core ammonia-metabolism genes and drove the proliferation of lung adenocarcinoma cell lines both in vitro and in vivo.

This study identifies tumor ammonia metabolism as a critical determinant of prognosis and immunotherapy resistance in LUAD. ABCC2 is further established as a key driver of this adverse phenotype, positioning it as a novel prognostic biomarker and a promising therapeutic target.