[Construction of a prognosis forecasting model for breast cancer based on lipid metabolism-related genes and functional verification of ].

[OBJECTIVES] To investigate the expression patterns and prognostic value of lipid metabolism-related genes in breast cancer.

[METHODS] RNA sequencing data and clinical information were obtained from The Cancer Genome Atlas breast cancer-related gene (TCGA-BRCA) cohort, including 1100 breast cancer tissue samples and 112 normal breast tissue samples. Differentially expressed lipid metabolism-related genes were screened from a predefined set of 2043 genes using Bioconductor in R, with a false discovery rate <0.05 and |log(fold change)|>2. Breast cancer tissue samples were randomly divided into a training cohort (=651) and a validation cohort (=431) at a 6∶4 ratio. Prognostic lipid metabolism-related genes were identified using univariate Cox regression (<0.01) and further refined via least absolute shrinkage and selection operate (LASSO) regression. A risk score model was constructed using multivariate Cox regression, and patients were stratified into high- and low-risk groups based on the median risk score. The model's performance was evaluated using Kaplan-Meier survival analysis with the log-rank test and time-dependent receiver operator characteristic (ROC) curves. A nomogram integrating age, TNM stage, clinical grade, and risk score was developed and validated using calibration curves and the concordance index. Immune cell infiltration was quantified using an immune scoring algorithm, and weighted gene co-expression network analysis (WGCNA) was applied to identify key modules associated with immune cell infiltration. Finally, to validate the function of the key gene , small interfering RNA targeting was transfected into breast cancer cells (MDA-MB-231), and its effects on invasion and migration were assessed using Transwell invasion and wound healing assays.

[RESULTS] A total of 185 differentially expressed lipid metabolism-related genes were identified. Univariate Cox and LASSO regression analyses identified three genes- and -which were incorporated into the multivariate Cox model. The prognosis forecasting model based on these genes demonstrated good predictive performance in both cohorts: patients in the high-risk group had significantly shorter overall survival (both <0.01), and the areas under the ROC curve for predicting 1-, 3-, and 5-year survival rates were all greater than 0.64. Analysis of the tumor microenvironment revealed a dysfunctional state in the high-risk group, characterized by reduced infiltration of several anti-tumor immune cells and downregulation of key immune checkpoint molecules such as PDCD1 and CTLA-4. WGCNA suggested an association between and immune cell infiltration. Functional experiments confirmed that knockdown significantly enhanced the migration and invasion abilities of breast cancer cells.

[CONCLUSIONS] This study established and validated a prognosis forecasting model for breast cancer based on lipid metabolism-related genes. It revealed that reduced expression is closely associated with poor prognosis and immunosuppression.