Spatially interpretable artificial intelligence framework to tailored neoadjuvant dual HER2 blockade in HER2-positive breast cancer
https://www.nature.com/articles/s41392-026-02734-0
Publish Date: 2026-06-18 19:31:00
Source Domain: www.nature.com
- The study systematically integrated multiple cohorts of HER2-positive breast cancer patients to develop and validate a predictive model termed HER2-LADDER.
- The model leverages digital pathology data, including spatial features extracted from tissue slides, to predict treatment responses in neoadjuvant and adjuvant settings.
- It demonstrated robust predictive performance across different treatment regimens and validation cohorts, indicating its generalizability and clinical utility.
- HER2-LADDER scores provide insights into treatment optimization, suggesting appropriate treatment de-escalation or tailored alternatives based on patient stratification.
- The model’s prognostic value extends to long-term outcomes, showing associations with overall and disease-free survival in adjuvant-treated patients.
- Biological interpretability of HER2-LADDER features reveals spatial determinants, including HER2 expression heterogeneity and tumor microenvironment interactions, that inform its predictive power.
