Identifying breast cancer subtypes and exploring spatial expression patterns of the key subtype-specific gene based...

Area of Science:

• Oncology
• Pathology
• Genomics

Background:

• Breast cancer (BRCA) is a heterogeneous disease requiring accurate prognosis for personalized treatment.
• Identifying molecular subtypes is crucial for tailoring BRCA therapies.
• This study aimed to discover prognostic subtypes using pathological images and molecular analyses.

Purpose of the Study:

• To identify prognostic subtypes of breast cancer (BRCA) from pathological image features.
• To explore the molecular and cellular mechanisms underlying these subtypes using spatial transcriptomics (ST) and single-cell RNA sequencing (scRNA-seq).
• To uncover potential therapeutic targets by analyzing key gene expression and cellular interactions.

Main Methods:

• Extracted pathological features from H&E-stained TCGA-BRCA images using PyRadiomics.
• Performed feature selection via Cox regression and machine learning, defining subtypes through consensus clustering.
• Identified key genes using WGCNA and Lasso regression, integrating ST and scRNA-seq for spatial and cellular interaction analysis.

Main Results:

• Identified 25 core pathological features defining two BRCA subtypes; C1 subtype associated with advanced disease and poor prognosis.
• ESRP1 identified as a key upregulated gene in the C1 subtype, highly expressed in tumor lesions.
• ST analysis revealed ESRP1 co-localization with epithelial and fibroblast cells, with fibroblasts central to TGF-β signaling in ESRP1-enriched regions.

Conclusions:

• Pathology-derived prognostic BRCA subtypes complement existing molecular classifications.
• The key gene ESRP1 shows distinct spatial expression in BRCA, localized to tumor cores with fibroblasts.
• Findings offer insights into BRCA heterogeneity and potential therapeutic strategies targeting ESRP1 and associated cellular interactions.