Abstract
Breast cancer is the second most prevalent malignant tumor worldwide and is highly heterogeneous. Cuproptosis, a newly identified form of cell death, is intimately connected to lipid metabolism. This study investigated breast cancer heterogeneity through the lens of cuproptosis-related lipid metabolism genes (CLMGs), with the goal of predicting patient prognosis, immunotherapy efficacy, and sensitivity to anticancer drugs. By utilizing transcriptomic data from The Cancer Genome Atlas (TCGA) for breast cancer, we identified 682 CLMGs and applied the nonnegative matrix factorization (NMF) method to categorize breast cancer patients into four distinct clusters: cluster 1, ''immune-cold and stroma-poor''; cluster 2, ''immune-infiltrated''; cluster 3, ''stroma-rich''; and cluster 4, ''moderate infiltration''. We subsequently developed a risk model based on CLMGs that incorporates ACSL1, ATP2B4, ATP7B, ENPP6, HSPH1, PIP4K2C, SRD5A3, and ULBP1. This model demonstrated excellent prognostic predictive performance in both the internal (testing and entire sets) and external (GSE20685 and Kaplan-Meier Plotter sets) validation sets. High-risk patients presented lower expression levels of immune checkpoint-related genes and lower immunophenoscores (IPSs), whereas low-risk patients presented higher CD8(+) T-cell infiltration levels and IPSs. Furthermore, the risk index was positively correlated with tumor cell stemness and could predict sensitivity to anticancer drugs. We also confirmed that SRD5A3 was highly expressed in breast cancer and participated in promoting the proliferation and migration of breast cancer cells. In conclusion, the results of this study provide new insights and strategies for assessing prognosis and implementing precision treatment for breast cancer through the lens of CLMGs.