Automated Prognosis Marker Assessment in Breast Cancers Using BLEACH&STAIN Multiplexed Immunohistochemistry

Area of Science:

• Oncology
• Biotechnology
• Artificial Intelligence

Background:

• Routine breast cancer prognosis relies on RNA panels susceptible to tumor purity variations.
• Multiplex fluorescence immunohistochemistry (mfIHC) offers potential for more precise risk assessment.
• Automated analysis is needed to leverage mfIHC for prognostic marker detection.

Purpose of the Study:

• To develop and validate an AI-driven framework for automated detection of prognostic markers in breast cancer using mfIHC.
• To assess the prognostic relevance of nine biomarkers (PR, ER, AR, GATA3, TROP2, HER2, PD-L1, Ki67, TOP2A) in invasive breast cancer.
• To establish a novel five-marker prognosis score for improved breast cancer risk stratification.

Main Methods:

• Development of a thirteen-step AI framework for automated breast cancer identification and prognosis marker detection.
• Application of 11+1-marker-BLEACH&STAIN-mfIHC staining on 1404 invasive breast cancers of no special type (NST).
• Inclusion of an algorithm for cell distance analysis and strict limitation of analysis to malignant cells.

Main Results:

• The AI framework achieved 98.4% accuracy in discriminating normal from malignant glands.
• Five biomarkers (PR, ER, AR, GATA3, PD-L1) were significantly associated with prolonged overall survival (p ≤ 0.0095).
• PR and AR emerged as independent risk factors, and a combined five-marker score (PR-ER-AR-GATA3-PD-L1) demonstrated strong prognostic relevance (p < 0.0001).

Conclusions:

• Automated AI analysis of mfIHC provides rapid, reliable assessment of multiple prognostic parameters in breast cancer.
• The developed framework overcomes the limitation of fluctuating tumor purity inherent in RNA-based methods.
• The five-marker prognosis score offers independent prognostic value, enhancing clinical risk stratification for breast cancer patients.