Establishment of an amino acid metabolism related signature for prognostic and therapeutic sensitivity prediction in breast cancer by machine learning

Amino acid metabolism plays a critical role in tumor growth and immune regulation, yet its comprehensive function in breast cancer remains underexplored. We developed an amino acid metabolism-related gene signature (AAMRGS) to predict prognosis and therapeutic response in breast cancer. The AAMRGS was constructed using a machine-learning framework integrating ten algorithms and validated across multiple independent cohorts. It served as an independent prognostic factor and outperformed existing amino acid metabolism-related signatures and clinical variables. Moreover, the prognostic utility of AAMRGS was further validated across pan-cancer datasets, and an AAMRGS-based nomogram was constructed to facilitate clinical application. Functional enrichment and protein-protein interaction analyses revealed that AAMRGS genes were primarily involved in metabolic reprogramming and cell proliferation. Experimental validation confirmed the downregulation of key genes such as SAV1 and IGF2R in breast cancer cells. Integrative analyses revealed that the high-AAMRGS subgroup exhibited a greater copy number variation burden, higher tumor mutation burden (TMB), enrichment of immunosuppressive cell populations, and increased sensitivity to most chemotherapeutic drugs. In contrast, the low-AAMRGS subgroup displayed higher immune scores, stronger immune activation, enrichment of anti-tumor immune cells, and greater responsiveness to immunotherapy. Collectively, our findings establish AAMRGS as a reliable prognostic signature and a potential tool to guide individualized therapeutic strategies for breast cancer patients.