New AI Model Improves Detection of Breast Cancer Metastasis

Catalan researchers combine clinical and immunological data to identify axillary lymph node metastasis, paving the way for new biomarkers.

Generic image of a microscope with cells and immune markers.
AI

Generic image of a microscope with cells and immune markers.

Researchers from the Pompeu Fabra University and Catalan hospitals have created a new artificial intelligence (AI) model that improves the detection of luminal breast cancer metastasis in the axillary lymph nodes by combining clinical data with patients' immune response.

The research, led by the Pompeu Fabra University and the Hospital Universitari de Tortosa Verge de la Cinta, along with the Rovira i Virgili University and the Biomedical Research Institute of Catalonia South, highlights the importance of the immune response, a factor not routinely assessed in clinical practice, for determining the presence of metastasis.
The new machine learning system has demonstrated a reduction in false positives compared to previous models trained solely on clinical data. While it does not predict future metastasis development, it does identify clinical and immune variables associated with their presence in the axillary lymph nodes.
The study has also identified a potential new biomarker, CD21+, a cell within the primary tumor that could be key to detecting tumor spread and, in the future, serve as a therapeutic target. This finding opens new avenues for understanding the immune mechanisms influencing breast cancer progression.
According to data from the World Health Organization, breast cancer was diagnosed in over 2.4 million women globally in 2024, with a significantly increased mortality rate when tumor cells spread to the axillary lymph nodes. The research was disseminated as part of World Breast Cancer Day, observed on October 19th.
The research was funded by European projects such as BosomShield and SCARLET, and the findings have been published in the journal Breast Cancer Res.
Based on information from the official source: IISPV - Institut d'Investigació Sanitària Pere Virgili (19/10/2026)