Exploratory Study on Prognostic Prediction in Triple-negative Breast Cancer

Background: Triple-negative breast cancer (TNBC) accounts for approximately 15-20% of all breast cancers and is associated with a high risk of recurrence and poor prognosis. There exists a significant gap in personalized treatment strategies for this aggressive subtype, necessitating better prognostic tools to guide therapy.

Method: This exploratory study utilized a machine-learning algorithm to analyze pathological images from a cohort of 200 early-stage TNBC patients. The primary endpoint focused on the accuracy of prognostic predictions, comparing AI-generated assessments with traditional histopathological evaluations.

Result: The AI model demonstrated a prognostic prediction accuracy of 85%, surpassing the conventional histopathological assessment accuracy of 70%. Additionally, the prediction model identified high-risk patients with a 3-year progression-free survival rate of 55% versus 85% in low-risk patients (P<0.01). Safety events were not reported in the study.

Conclusion: This study supports the potential integration of AI-driven image analysis for prognostic predictions in TNBC, suggesting that this approach could significantly enhance treatment personalization. However, as a preliminary analysis with a limited sample size, further validation in larger, prospective studies is required.

Original citation address: https://www.besjournal.com/en/article/doi/10.3967/bes2025.119