Breaking the Heterogeneity Barrier: A Robust Prognostic Signature for Survival Stratification and Immune Profiling in Triple-Negative Breast Cancer

Haixing Shen¹Qing Zheng²Zhenyu Wang²Daitian Zheng³Zhenqi Gong²Huaiming Wang²Tianmiao Sun¹Jie Pan⁴Yukai Jin⁴Xiaohong Zheng⁴Jingzhi Wang^5*Jiongjiong Zhang^1*

• ¹Affiliated Cixi Hospital, Wenzhou Medical University, Ningbo, Zhejiang 315300, China
• ²Department of Gastrointestinal Surgery, The First Affiliated Hospital of Shantou University Medical College, Shantou, Guangdong 515041, China
• ³Second Department of Thyroid and Breast Surgery, The Third Affiliated Hospital of Shenzhen University, Shenzhen, Guangdong 518000, China
• ⁴Cixi Biomedical Research Institute, Wenzhou Medical University, Ningbo, Zhejiang 315300, China
• ⁵Department of Radiotherapy Oncology, The Affiliated Yancheng First Hospital of Nanjing University Medical School; The First People's Hospital of Yancheng, Yancheng, Jiangsu 224000, China

Background: Triple-negative breast cancer (TNBC), a highly heterogeneous breast cancer subtype, poses significant challenges to human health. Intra-tumor heterogeneity (ITH) limits the reliability of conventional prognostic models.Methods: Using multi-region RNA-seq, we quantified TNBC transcriptomic heterogeneity through an integrative heterogeneity score (IHS). After evaluating inter-patient heterogeneity (IPH) and ITH, prognostic and low-heterogeneity genes were identified and used to build a prognostic risk model with a random survival forest (RSF) algorithm. This model was combined with TNM staging into a nomogram for clinical applicability. We further revealed the distinct immune microenvironment features, somatic mutations, and chemotherapy responses between risk subgroups. Gene expression was validated via RT-qPCR.Results: Spatial characterization uncovered substantial ITH, evidenced by sharp shifts in PAM50 subtypes and immune infiltration. Two low-heterogeneity biomarkers, CYP4B1 and GBP1, were identified to develop a robust prognostic signature with consistent predictive performance across 3- to 9-year survival endpoints (AUC > 0.6). The high-risk subgroup exhibited reduced immune infiltration, reduced immune checkpoint molecule expression, and poor immunotherapy response rates. Integration of the risk signature with TNM staging created a clinically practical nomogram with superior predictive accuracy (C-index >0.67). Therapeutic vulnerability profiling identified six targeted agents showing increased efficacy in high-risk patients. Dysregulation of signature genes was demonstrated in two TNBC cell lines.Conclusions: This study established a transcriptomic heterogeneity-resilient prognostic model for TNBC, enabling precise survival stratification and immune microenvironment assessment. The integrative nomogram and risk-guided therapeutic predictions address clinical challenges in TNBC management, advancing personalized treatment strategies.