Multimodal integration using a machine learning approach facilitates risk stratification in HR+/HER2− breast cancer

• DOI: 10.1016/j.xcrm.2024.101924
• Publication Date: 2025-01-22T15:34:41Z
• Authors: Hang Zhang; Fan Yang; Ying Xu; Shen Zhao; Yi-Zhou Jiang; Zhi-Ming Shao; Yi Xiao
• Keywords: Risk Stratification; Stratification (seeds)

Hormone receptor-positive (HR+)/human epidermal growth factor receptor 2-negative (HER2-) breast cancer is the most common type of cancer, with continuous recurrence remaining an important clinical issue. Current relapse predictive models in HR+/HER2- patients still have limitations. The integration multidimensional data represents a promising alternative for predicting relapse. In this study, we leverage our multi-omics cohort comprising 579 (200 complete across 7 modalities) and develop machine-learning-based model, namely CIMPTGV, which integrates information, immunohistochemistry, metabolomics, pathomics, transcriptomics, genomics, copy number variations to predict risk cancer. This model achieves concordance indices (C-indices) 0.871 0.869 train test sets, respectively. population predicted by CIMPTGV encompasses those identified single-modality models. Feature analysis reveals that synergistic complementary effects exist different modalities. Simultaneously, simplified mean area under curve (AUC) 0.840, presenting useful approach applications.