Related Experiment Video
Updated: Jun 9, 2025
Quantifying the Brain Metastatic Tumor Micro-Environment using an Organ-On-A Chip 3D Model, Machine Learning, and Confocal Tomography
Published on: August 16, 2020
Single-cell omics and machine learning integration to develop a polyamine metabolism-based risk score model in breast
Xiliang Zhang1, Hanjie Guo2, Xiaolong Li1
1Department of General Surgery, Xinqiao Hospital, Army Medical University, No. 83 Xinqiao Main Street, Shapingba District, Chongqing, 400037, China.
Background:
Breast cancer remains the leading malignant neoplasm among women globally, posing significant challenges in terms of treatment and prognostic evaluation. The metabolic pathway of polyamines is crucial in breast cancer progression, with a strong association to the increased capabilities of tumor cells for proliferation, invasion, and metastasis.
Methods:
We used a multi-omics approach combining bulk RNA sequencing and single-cell RNA sequencing (scRNA-seq) to study polyamine metabolism. Data from The Cancer Genome Atlas, Gene Expression Omnibus, and Genotype-Tissue Expression identified 286 differentially expressed genes linked to polyamine pathways in breast cancer. These genes were analyzed using univariate COX and machine learning algorithms to develop a prognostic scoring algorithm. Single-cell RNA sequencing validated the model by examining gene expression heterogeneity at the cellular level.
Results:
Our single-cell analyses revealed distinct subpopulations with different expressions of genes related to polyamine metabolism, highlighting the heterogeneity of the tumor microenvironment. The SuperPC model (a constructed risk score) demonstrated high accuracy when predicting patient outcomes. The immune profiling and functional enrichment analyses revealed that the genes identified play a crucial role in cell cycle control and immune modulation. Single-cell validation confirmed that polyamine metabolism genes were present in specific cell clusters. This highlights their potential as therapeutic targets.
Conclusions:
This study integrates single cell omics with machine-learning to develop a robust scoring model for breast cancer based on polyamine metabolic pathways. Our findings offer new insights into tumor heterogeneity, and a novel framework to personalize prognosis. Single-cell technologies are being used in this context to enhance our understanding of the complex molecular terrain of breast cancer and support more effective clinical management.
