Identification and validation of molecular subtypes and a 9-gene risk model for breast cancer

The long-term efficacy of treatment, heterogeneity, and complexity in the tumor microenvironment remained a clinical challenge in breast cancer (BRCA). There is a need to classify and refine appropriate therapeutic intervention decisions. A stable subtype classification based on gene expression associated with neoadjuvant chemotherapy (NAC) prognosis and assessment on the clinical features, immune infiltration, and mutational characteristics of the different subcategories was performed using ConsensusClusterPlus. We constructed a prognostic model by the least absolute shrinkage and selection operator regression (LASSO) and univariate Cox regression method and further investigated the association between the risk model and clinical features, mutation and immune characteristics of BRCA. We constructed 3 molecular clusters associated with NAC. We found that cluster 1 had the best prognosis, while cluster 3 showed a poor prognosis. Cluster 3 were associated with the advance stage, higher mutation score, activated oncogenic, and lower tumor immune dysfunction and exclusion (TIDE) score. Subsequently, we constructed a prognosis-related risk model comprising 9 genes (RLN2, MSLN, SAPCD2, LY6D, CACNG4, TUBA3E, LAMP3, GNMT, KLHDC7B). The higher-risk group exhibited lower immune infiltration and demonstrated improved overall survival (OS) in both the independent validation cohort. Finally, by combining clinicopathological features with the NAC-related prognostic risk model, we enhanced the accuracy of survival prediction and model performance. Here, we revealed 3 new molecular subtypes based on prognosis-related genes for BRCA NAC and developed a prognostic risk model. It has the potential to aid in the selection of appropriate individualized treatment and the prediction of patient prognosis.