A new machine-learning tool can rapidly screen thousands of compounds to find those that push cancer cells into permanent senescence, offering fresh hope for tough-to-treat cancers, a new study reveals.
The research published in Aging-US describes a machine-learning method developed by Ryan Wallis and Cleo L Bishop that uses cell-shape analysis to rapidly identify pro-senescence drugs, accelerating discovery for hard-to-treat p16-positive tumours.
Cellular senescence is a natural state in which damaged or old cells permanently stop dividing.
In cancer treatment, deliberately pushing tumour cells into senescence offers a way to halt their growth without killing them outright.
The challenge has been reliably confirming senescence in cancers that already look “aged” (so-called Sen-Mark+ tumours, such as basal-like breast cancer), where conventional biomarkers often fail.
To solve this, researchers Wallis and Bishop developed SAMP-Score, a machine-learning tool that bypasses traditional markers entirely.
Instead, it recognises senescence by analysing microscopic changes in cell shape and structure, distinct patterns called senescence-associated morphological profiles (SAMPs).
Trained on thousands of images, the model can now accurately separate true senescence from mere toxicity or normal variation, providing a fast, visual way to screen compounds that drive cancer cells into permanent retirement.
"This technique builds upon our previous observation that senescent cells develop distinct senescence-associated morphological profiles (SAMPs), which can be assessed readily in traditionally challenging contexts for senescence identification, including high-throughput screens,” the researchers said in the study.
As the team used SAM-Score to screen more than 10,000 experimental compounds, a compound called QM5928 was identified that induced senescence in multiple cancer cell types but did not kill them.
The researchers believe this compound is worth further study. QM5928 was unique in that it was effective against cancers resistant to known drugs – such as palbociclib – that don't always work in cancers with high p16 expression.
"Through application of SAMP-Score, we have identified QM5928, a novel pro-senescence compound, that is able to induce senescence in a variety of Sen-Mark+ cancers and has potential utility as a tool molecule to explore the mechanisms and pathways through which senescence induction occurs in these cells," the research study said.
By combining machine learning with high-resolution imaging, the researchers offer a novel way to detect and measure cancer therapies.
SAMP-Score may pave the way to the emergence of treatments that utilise the body's natural ageing process to fight cancer, primarily for patients with treatment-resistant tumours.
