Obesity is a known risk factor for breast cancer, yet the molecular mechanisms linking adipose tissue dysfunction to tumor progression remain incompletely understood. Extracellular vesicles derived from breast adipose tissue are a key mediator of cell-tocell communication within the tumor microenvironment, capable of influencing cancer cell metabolism and proliferation. In this study, we applied an integrative proteomic and computational pipeline to explore the differential expression and predictive significance of EV-associated proteins derived from breast adipose tissue in overweight and obese women.
An open-source proteomic dataset of EVs isolated from breast adipose tissue was reanalyzed and filtered for high-confidence peptides with experimental metadata. Preprocessing included log₂ transformation, imputation, and feature scaling. Dimensionality reduction was performed using a Principal Component Analysis (PCA) + UMAP hybrid strategy, followed by unsupervised K-means clustering, which revealed two molecularly distinct sample clusters not associated with any previously noted metric. Differential expression analysis identified 665 proteins, including SERPINA1, CAV1, MMP9, and AKR1C2, with significant enrichment in extracellular matrix remodeling, hormone metabolism, and immune modulation. A supervised machine learning approach using Decision Tree regression predicted breast cancer cells (BCC) proliferation from EV proteomic profiles (R² = 0.87, AUC = 0.98 for binary classification). Top-ranked predictive proteins, including PARP1, FCN2, and HLA-G, were further explored for functional relevance, although not all overlapped with canonical pathways explored by a comprehensive systems biology software (IPA-Ingenuity Pathway Analysis) which identified Estrogen-Dependent Breast Cancer Signaling and Tumor Microenvironment as canonical pathways, while CTNNB1-centered node network analysis revealed EVassociated proteins with known roles in metabolic rewiring and tumor progression. Further validation is needed to enhance the robustness of the machine learning model for early breast cancer risk stratification and molecular subtyping
Obesity is a known risk factor for breast cancer, yet the molecular mechanisms linking adipose tissue dysfunction to tumor progression remain incompletely understood. Extracellular vesicles derived from breast adipose tissue are a key mediator of cell-tocell communication within the tumor microenvironment, capable of influencing cancer cell metabolism and proliferation. In this study, we applied an integrative proteomic and computational pipeline to explore the differential expression and predictive significance of EV-associated proteins derived from breast adipose tissue in overweight and obese women.
An open-source proteomic dataset of EVs isolated from breast adipose tissue was reanalyzed and filtered for high-confidence peptides with experimental metadata. Preprocessing included log₂ transformation, imputation, and feature scaling. Dimensionality reduction was performed using a Principal Component Analysis (PCA) + UMAP hybrid strategy, followed by unsupervised K-means clustering, which revealed two molecularly distinct sample clusters not associated with any previously noted metric. Differential expression analysis identified 665 proteins, including SERPINA1, CAV1, MMP9, and AKR1C2, with significant enrichment in extracellular matrix remodeling, hormone metabolism, and immune modulation. A supervised machine learning approach using Decision Tree regression predicted breast cancer cells (BCC) proliferation from EV proteomic profiles (R² = 0.87, AUC = 0.98 for binary classification). Top-ranked predictive proteins, including PARP1, FCN2, and HLA-G, were further explored for functional relevance, although not all overlapped with canonical pathways explored by a comprehensive systems biology software (IPA-Ingenuity Pathway Analysis) which identified Estrogen-Dependent Breast Cancer Signaling and Tumor Microenvironment as canonical pathways, while CTNNB1-centered node network analysis revealed EVassociated proteins with known roles in metabolic rewiring and tumor progression. Further validation is needed to enhance the robustness of the machine learning model for early breast cancer risk stratification and molecular subtyping Read More
Related posts:
Corn Husk Ferulic Acid Induces Apoptosis of Breast Cancer Cells via ROS‐Mediated Signaling Pathways
In Vivo and In Vitro Study on the Mechanism of Anticervical Cancer Effects of Corilagin in Mice
Holothuria tubulosa Extract Represses Proliferation and HIF‐α Activity by Inhibiting Erk1/2 Phosphorylation in Liver Cancer Cell Line under Hypoxia
Functional characterization of genes mediating cell wall metabolism and responses to plant cell wall integrity impairment
Programmable ROS-Mediated Cancer Therapy via Magneto-Inductions
Anti‐adiposity and lipid‐lowering effects of schisandrol A in diet‐induced obese mice