Integrative multi-omics clustering for identifying novel breast cancer subtypes with distinct molecular and clinical characteristics

Breast cancer is a heterogeneous disease that requires refined classification frameworks that can effectively guide targeted therapies. However, traditional methods fail to capture the comprehensive molecular insights needed for this purpose. To comprehensively capture breast cancer heterogeneity, we employed integrative clustering that incorporates six molecular features from 670 breast cancer samples. Ten distinct clustering algorithms were combined to ensure robust subtype identification, and the identified subtypes were validated in four independent datasets. Subsequently, we constructed a survival support vector machine prognostic model based on key molecular features to enhance survival prediction and clinical applicability. Five novel subtypes were identified: consensus subtypes 1–5 (CS1–CS5). CS2 is an aggressive subtype with elevated TP53 mutation rates, high tumor mutational burden, and strong sensitivity to YM-155 and ispinesib. Conversely, CS5 exhibits stable genomics with enhanced nucleotide excision repair and favorable prognoses. CS2 and CS4 show enriched immune checkpoint expression, indicating potential immunotherapy responsiveness, while CS1 and CS5 exhibit immune-cold profiles. The survival support vector machine model effectively predicts survival outcomes across independent datasets. The refined breast cancer classification framework developed in this research uncovers new insights into molecular heterogeneity, enhances risk stratification, and enables the identification of promising therapeutic targets. The potential of this framework to optimize personalized treatment strategies warrants further clinical validation.