Robust machine-learning based prognostic index using fatty acid metabolism genes predicts prognosis and therapy responses in glioblastoma

<b>Background</b>: Glioblastoma (GBM) is the most prevalent and aggressive type of primary brain tumor in adults. Fatty acid metabolism plays a crucial role in promoting tumorigenesis, disease progression, and therapeutic resistance through the regulation of lipid synthesis, storage, and catabolism. However, its potential for predicting both prognosis and treatment response in glioblastoma is unexplored. <b>Methods</b>: We systematically compiled fatty acid metabolism-related genes (FAMGs) from published literature and databases. A fatty acid metabolism signature (FAMS) was developed using a machine learning-based framework. The predictive performance of the FAMS was rigorously validated across multiple independent cohorts. Additionally, we investigated the associations between FAMS and clinical characteristics, mutation profiles, tumor microenvironment features, and biological functions. <b>Results</b>: Our analysis revealed distinct FAMGs expression patterns in patients with GBM, which correlated with varying survival outcomes. Leveraging a robust machine learning framework, we established a fatty acid metabolism-based prognostic model. The FAMS emerged as an independent predictor of overall survival and other survival endpoints. Patients with lower FAMS exhibited enrichment in mitosis- and DNA repair-related pathways, which is linked to better survival. Conversely, higher FAMS scores were associated with enhanced immune activation, cellular proliferation, and chemotaxis, suggesting a greater likelihood of benefitting from immunotherapy. <b>Conclusion</b>: We developed a reliable fatty acid metabolism signature capable of stratifying GBM patients on the basis of prognosis. The FAMS serves as an independent prognostic indicator and may offer clinical utility in guiding personalized treatment strategies for patients with GBM.