AI Tool Measures Cancer Aggressiveness Using Protein Expression
Cancer diagnosis and treatment face increasing challenges as cases rise worldwide. A new AI-based model developed by researchers at the University of São Paulo and collaborators in Poland offers a promising approach to assessing tumor aggressiveness. The tool creates a stemness index by analyzing protein expression patterns, helping to predict how aggressive a tumor might be.
What Is the Stemness Index?
The stemness index measures the similarity of tumor cells to pluripotent stem cells—cells capable of transforming into almost any cell type. Tumors with higher stemness scores tend to be more aggressive, drug-resistant, and prone to recurrence. This index ranges from zero (low aggressiveness) to one (high aggressiveness).
As cancer progresses, malignant cells lose their original tissue characteristics, acquiring an undifferentiated phenotype and the ability to self-renew, similar to stem cells. This dedifferentiation contributes to the tumor’s malignancy and treatment resistance.
Building the PROTsi Model
The researchers developed the protein expression-based stemness index, called PROTsi, using data from the Clinical Proteomic Tumor Analysis Consortium (CPTAC). They analyzed over 1,300 tumor samples across 11 cancer types, including breast, ovarian, lung, kidney, brain, and pancreatic cancers.
By integrating PROTsi with proteomic data from 207 pluripotent stem cells, the team identified key proteins that drive tumor aggressiveness. These proteins represent potential targets for new or improved therapies, supporting both general and personalized treatment strategies.
Validation and Clinical Implications
PROTsi demonstrated consistent performance across multiple datasets, effectively distinguishing between tumor and non-tumor samples. It was particularly effective in predicting aggressiveness in uterine, head and neck, pancreatic, and pediatric brain cancers.
Some identified proteins are already targets of existing drugs, opening opportunities for repurposing or developing new treatments. This approach could guide clinicians in personalizing therapies based on a tumor’s stemness profile.
Context of Rising Cancer Incidence
The World Health Organization reports that 40 people are diagnosed with cancer every minute globally. Early-onset cancer in adults under 50 has increased by 79% from 1990 to 2019, with cancer-related deaths rising by 28% during the same period. In Brazil, the National Cancer Institute estimates over 700,000 new cases annually through 2025.
Common malignant tumors include non-melanoma skin cancer, breast, prostate, colon, lung, and stomach cancers. Tools like PROTsi are critical to improving diagnosis and treatment amidst this growing burden.
Future Directions
The research team is testing additional computational models to enhance prediction accuracy. They have made their data publicly available to encourage further development in this area.
This work reflects an important step in leveraging proteomics and AI to understand tumor biology and improve clinical outcomes.
About the São Paulo Research Foundation (FAPESP)
The São Paulo Research Foundation (FAPESP) supports scientific research across all knowledge fields by providing scholarships, fellowships, and grants to researchers in São Paulo, Brazil. It fosters international collaboration with institutions worldwide to enhance research quality.
Learn more at FAPESP’s official website and follow their updates at FAPESP News Agency.
Reference
Journal: Cell Genomics
DOI: 10.1016/j.xgen.2025.100851
Article Title: Proteomic-based stemness score measures oncogenic dedifferentiation and enables the identification of druggable targets
Publication Date: April 17, 2025
Your membership also unlocks:
