Integrated Bioinformatics Analysis to Identify a Novel Four-Gene Prognostic Model of Breast Cancer and Reveal Its Association with Immune Infiltration.

ABSTRACT

Liquid-liquid phase separation (LLPS) impact immune signaling in cancer and related genes have shown prognostic value in breast cancer (BRCA). However, the crosstalk between LLPS and immune infiltration in BRCA remain unclear. Therefore, we aimed to develop a novel prognostic model of BRCA related to LLPS and immune infiltration. BRCA-related, liquid-liquid phase separation (LLPS)-related genes, and differentially expressed genes (DEGs) were identified using public databases. Mutation and drug sensitivity analyses were performed using Gene Set Cancer Analysis database. Univariate cox regression and LASSO Cox regression were used for the construction and verification of prognostic model. Kaplan-Meier analysis was performed to evaluate overall survival (OS). Gene set variation analysis was conducted to analyze key pathways. CIBERSORT was used to assess immune infiltration and its correlation with prognostic genes was determined through Pearson analysis. A total of 6056 BRCA-associated genes, 3775 LLPS-associated genes, and 4049 DEGs, resulting in 314 overlapping genes. Twenty-eight prognostic genes were screened, and some of them were mutational and related to drug sensitivity Subsequently, a prognostic model comprising L1CAM, EVL, FABP7, and CST1 was built. Patients in high-risk group had shorter OS than those in low-risk group. The infiltrating levels of CD8+ T cells, macrophages M0, macrophages M2, dendritic cells activated, and mast cells resting was altered in high-risk group of breast cancer patients compared to low-risk group. L1CAM, EVL, FABP7, and CST1 were related to these infiltrating immune cells. L1CAM, EVL, FABP7, and CST1 were potential diagnostic biomarkers and therapeutic targets for BRCA.