AI-Guided Spatial Proteomics Reveals Recurrence-Risk Niches in Triple-Negative Breast Cancer
An intriguing study recently published on arXiv (2608.03145) introduces a cutting-edge spatial pathology framework for triple-negative breast cancer (TNBC). This system combines AI-generated recurrence risk heatmaps with mass spectrometry-based spatial proteomics. The research involved 156 patients and achieved an AUC and C-index of 0.77 in a separate test cohort by pooling high-scoring patches. When examining bulk proteomics, it was found that high-risk images were linked to cell cycle and genome maintenance, while low-risk areas showed immune activation. Notably, both high and low-risk patches coexisted within the same tumor, illustrating intratumoral diversity. The heatmaps helped isolate and profile 46 tumor regions defined by AI, which could improve molecular characterizations and lead to more targeted treatments for TNBC patients.
Key facts
- Framework integrates AI-generated recurrence risk heatmaps with mass spectrometry-based spatial proteomics.
- Cohort of 156 patients; independent test cohort achieved AUC 0.77 and C-index 0.77.
- High risk associated with cell cycle and genome maintenance programs; low risk with immune activation.
- High and low risk patches coexist within same tumor compartment, showing distinct nuclear and architectural features.
- Heatmaps used to physically isolate and profile 46 AI-defined tumor regions.
- Study published on arXiv with ID 2608.03145.
- Focus on triple-negative breast cancer (TNBC).
Entities
Institutions
- arXiv