2025-07-16 ジョンズ・ホプキンス大学(JHU)
<関連情報>
- https://hub.jhu.edu/2025/07/16/breast-cancer-microenvironments/
- https://www.cell.com/patterns/fulltext/S2666-3899(25)00026-1
2値グラフ学習が乳房多重デジタル病理診断における予後に関連する腫瘍微小環境パターンを明らかにする Bi-level graph learning unveils prognosis-relevant tumor microenvironment patterns in breast multiplexed digital pathology
Zhenzhen Wang ∙ Cesar A. Santa-Maria ∙ Aleksander S. Popel ∙ Jeremias Sulam
Patterns Published:February 11, 2025
DOI:https://doi.org/10.1016/j.patter.2025.101178
Graphical abstract
The bigger picture
Breast cancer is one of the most common cancers in women and, unfortunately, telling how or when a patient will respond to a given treatment continues to be poorly understood——particularly in the cases of most aggressive breast cancer subtypes. The complex biology reflected in the tissue surrounding and including the tumor, known as the tumor microenvironment, has an important impact on the evolution of the disease and on the overall prognosis of the patient. However, this microenvironment is very complex and heterogeneous, making it difficult to extract informative quantitative measures that can be indicative of better patient outcomes. This work provides a method that learns, in an automatic manner, patterns of cell interactions in the tumor microenvironment. We show that these cell patterns characterize different groups of patients with distinct outcomes, leading to the discovery of prognostic biomarkers. This work highlights the power of data-driven methods in uncovering biologically meaningful features using breast cancer as a case study, but it is applicable to other diseases more broadly.
Highlights
- BiGraph systematically links local cell patterns with patient survival outcomes
- It provides data-driven and interpretable insights into cancer prognosis
- It complements and enhances risk stratification of classical clinical subtyping
- This approach is broadly applicable to other diseases and other spatial single-cell data
Summary
The tumor microenvironment (TME) is widely recognized for its central role in driving cancer progression and influencing prognostic outcomes. Increasing efforts have been dedicated to characterizing it, including its analysis with modern deep learning. However, identifying generalizable biomarkers has been limited by the uninterpretable nature of their predictions. We introduce a data-driven yet interpretable approach for identifying cellular patterns in the TME associated with patient prognoses. Our method relies on constructing a bi-level graph model: a cellular graph, which models the TME, and a population graph, capturing inter-patient similarities given their respective cellular graphs. We demonstrate our approach in breast cancer, showing that the identified patterns provide a risk-stratification system with new complementary information to standard clinical subtypes, and these results are validated in two independent cohorts. Our methodology could be applied to other cancer types more generally, providing insights into the spatial cellular patterns associated with patient outcomes.
