J Breast Cancer.  2024 Oct;27(5):305-322. 10.4048/jbc.2024.0107.

p27 Cell Cycle Inhibitor and Survival in Luminal-Type Breast Cancer: Gene Ontology, Machine Learning, and Drug Screening Analysis

• Park IA
• Noh YK
• Min KW
• Kim DH
• Lee JY
• Son BK
• Kwon MJ
• Han MH
• Hur JY
• Pyo JS

• KMID: 2560788
• DOI: http://doi.org/10.4048/jbc.2024.0107

Abstract

Purpose
A widely distributed cell cycle inhibitor, p27, regulates cyclin-dependent kinase-cyclin complexes. Although the prognostic value of p27 has been established for various types of carcinomas, its role in luminal breast cancer remains poorly understood. This study aimed to explore the functional enrichment of p27 and identify potential drug targets in patients with luminal-type breast cancer.
Methods
Clinicopathological data were collected from 868 patients with luminal-type breast cancer. Additionally, publicly available data from the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) dataset (1,500 patients) and the Gene Expression Omnibus database (855 patients) were included in the analysis. Immunohistochemical staining for p27, differential gene expression analysis, disease ontology analysis, survival prediction modeling using machine learning (ML), and in vitro drug screening were also performed.
Results
Low p27 expression correlated with younger age, advanced tumor stage, estrogen receptor/progesterone receptor negativity, decreased cluster of differentiation 8+ T cell count, and poorer survival outcomes in luminal-type breast cancer. The METABRIC data revealed that reduced cyclin-dependent kinase inhibitor 1B (CDKN1B) expression (encoding p27) was associated with cell proliferation-related pathways and epigenetic polycomb repressive complex 2. Using ML, p27 emerged as the second most significant survival factor after N stage, thereby enhancing survival model performance. Additionally, luminal-type breast cancer cell lines with low CDKN1B expression demonstrated increased sensitivity to specific anticancer drugs such as voxtalisib and serdemetan, implying a potential therapeutic synergy between CDKN1B-targeted approaches and these drugs.
Conclusion
The integration of ML and bioinformatic analyses of p27 has the potential to enhance risk stratification and facilitate personalized treatment strategies for patients with breast cancer.

Keyword

Breast Neoplasms; Cyclin-Dependent Kinase Inhibitor p27; Gene Ontology; Machine Learning; Prognosis