Cancer treatment has made major advances. Yet one of the biggest challenges remains the return of tumours after treatment.
Modern therapies can destroy millions of cancer cells. However, a small population of cells can survive. These surviving cells can allow tumours to return, spread to distant organs and develop resistance to treatment.
Scientists have long focused on rare cancer stem-like cells because of their role in tumour recurrence, metastasis and treatment failure. Detecting these cells, however, has been difficult. They are extremely rare and can constantly change their identity.
Researchers from the S. N. Bose National Centre for Basic Sciences (SNBNCBS), an autonomous institute of the Department of Science and Technology (DST), Government of India, along with Ashoka University, have now developed an artificial intelligence framework to address this challenge.
The new AI framework can identify hidden cancer stem-like cell states using tumour gene-expression data. The work was led by Dr. Shubhasis Haldar.
AI Looks For Hidden Cancer Cells
The research builds on the team’s earlier AI platform, OncoMark.
OncoMark was developed to decode biological hallmarks linked to cancer progression. According to the research team, the platform achieved more than 99% predictive accuracy while analysing cancer-related information across millions of cells.
The platform showed how artificial intelligence can identify complex biological information hidden inside large genomic datasets.
The researchers have now applied this approach to another difficult problem in cancer biology.
Their focus is on identifying cancer stem-like cells that can contribute to tumour evolution and survival.
New Framework Identifies Three Cell States
The new framework is called ACSCeND.
Its full name is AI-based Cancer Stem-like Cell Profiler and Neoplasm Deconvoluter.
Unlike conventional approaches that give a tumour a single “stemness” score, ACSCeND identifies three different developmental states of cancer stem-like cells.
These are:
- Pluripotent-like
- Multipotent-like
- Unipotent-like
This distinction provides a more detailed view of the different cell states present inside a tumour.
The system combines information learned from high-resolution single-cell sequencing with deep learning. It then uses this information to analyse conventional bulk tumour RNA sequencing.
This is important because single-cell experiments are not available for every patient sample.
The framework can therefore help researchers study hidden cell populations across thousands of tumour samples using available bulk tumour RNA sequencing data.
Tested Across Different Datasets
The researchers also tested ACSCeND against existing computational methods.
The framework was evaluated using independent datasets and different sequencing platforms. According to the research findings, ACSCeND consistently performed better than the existing approaches used for comparison.
The researchers then applied the framework to more than 25,000 tumour samples.
These samples came from major international cancer databases, including TCGA and PRECOG.
The analysis provided information about the relationship between different cancer stem-like cell states and patient outcomes.
Link With Survival And Recurrence
The analysis found that tumours containing higher levels of highly potent, pluripotent-like cancer stem cells were associated with poorer patient survival.
The same tumours were also associated with a higher likelihood of tumour recurrence.
The research further found an association with reduced response to modern immunotherapies.
These findings give researchers a way to examine cancer stem-like cells in much larger numbers of tumour samples.
ACSCeND also identified molecular programmes linked to the ability of these cells to survive, adapt and evade the immune system.
Such information could help scientists investigate possible drug targets.
It could also help identify patients who may be more likely to experience a relapse.
AI And Precision Cancer Research
The development highlights the growing role of artificial intelligence in biomedical research.
Cancer research involves extremely large genomic datasets. Finding complex biological patterns across these datasets can be difficult through manual analysis.
AI-based systems can analyse such information and identify patterns that may otherwise remain hidden.
The researchers’ work with OncoMark and ACSCeND demonstrates how AI can be applied to different challenges in cancer biology.
The new framework focuses specifically on identifying different states of cancer stem-like cells.
By detecting these hidden cell populations in large numbers of tumour samples, the approach could support further research into tumour growth, recurrence, metastasis and treatment resistance.
The findings also provide information that may help researchers study treatment response and develop more targeted approaches to cancer therapy.
The research team says ACSCeND can bring precision medicine closer to reality, particularly in areas where advanced health facilities and single-cell experiments are limited.
The findings add to the growing use of artificial intelligence to understand complex biological processes and analyse large-scale cancer data.
Subscribe Deshwale on YouTube
