Abstract
Triple-negative breast cancer (TNBC) is a highly heterogeneous tumor that lacks effective treatment and has a poor prognosis. Exosomes carry abundant genomic information and have a significant role in tumorigenesis, metastasis, and drug resistance. However, further exploration is needed to investigate the relationship between exosome-related genes and the heterogeneity and tumor immune microenvironment of TNBC. Based on the exosome-related gene sets, multiple machine learning algorithms, such as Cox boost, were used to screen the risk score model with the highest C-index. A 9-gene risk score model was constructed, and the TNBC population was divided into high- and low-risk groups. The effectiveness of this model was verified in multiple datasets. Compared with the low-risk group, the high-risk group exhibited a poorer prognosis, which may be related to lower levels of immune infiltration and immune response rates. The gene mutation profiles and drug sensitivity of the two groups were also compared. By screening for genes with the most prognostic value, the hub gene, CLDN7, was identified, and thus, its potential role in predicting prognosis, as well as providing ideas for the clinical diagnosis, treatment, and risk assessment of TNBC, was also discussed. This study demonstrates that exosome-related genes can be used for risk stratification in TNBC, identifying patients with a worse prognosis. The high-risk group exhibited a poorer prognosis and required more aggressive treatment strategies. Analysis of the genomic information in patient exosomes may help to develop personalized treatment decisions and improve their prognosis. CLDN7 has potential value in prognostic prediction in the TNBC population.