MultiOmics analysis of metabolic dysregulation and immune features in breast cancer

Abstract

Metabolic disorders and diminished immune response are hallmark characteristics of tumors. However, limited studies have comprehensively integrated metabolic and immunological factors to evaluate or predict the prognosis of cancer patients. In this study, we utilized 72 metabolic pathway gene sets from the MsigDB database to conduct GSVA, univariate regression, and prognostic analyses on 247 breast cancer samples sourced from the TCGA and GEO databases. Consequently, five metabolic pathways with significant research value were identified. Based on these findings, unsupervised clustering was performed on the breast cancer samples to compare differences in gene expression, clinicopathological features, immune infiltration levels, and prognosis across different clusters. This process led to the identification of nine metabolism-related characteristic genes. Additionally, single-cell sequencing analysis was employed to assess the spatial expression patterns of these characteristic genes, revealing significantly higher expression indices in tumor cells compared to non-tumor cells. Subsequently, machine learning algorithms were applied to reconstruct metabolic risk models for evaluating the prognosis of breast cancer patients. The results indicated that the high metabolic risk group exhibited higher gene mutation scores, a greater proportion of unfavorable clinicopathological parameters, and lower chemokine and immune scores compared to the low-risk group. In conclusion, the metabolic risk model constructed using metabolism-related characteristic genes can accurately distinguish and predict the survival prognosis and immunotherapy outcomes of breast cancer patients, offering novel targets and insights for personalized treatment strategies.