An ordinary blood sample may soon help doctors find liver cancer much earlier than is possible today.
Scientists from the Johns Hopkins Kimmel Cancer Center have successfully tested an artificial intelligence system that searches for hidden DNA patterns in blood, offering a promising new way to detect one of the world’s deadliest cancers. The findings were published in Cell Press Blue.
The technology is based on a liquid biopsy platform called DELFI. Instead of looking for cancer cells themselves, it examines millions of tiny DNA fragments that are released into the bloodstream as cells naturally break down. Artificial intelligence then studies these patterns to identify signs of disease.
Finding liver cancer early is extremely important because treatment is much more successful before the disease spreads. Unfortunately, many patients are diagnosed only after symptoms appear. Existing screening methods, including ultrasound scans and alpha-fetoprotein blood tests, often fail to detect the earliest cancers.
To test whether the new method worked in different parts of the world, researchers analyzed blood samples from 377 people living in Romania and Guatemala. The two groups represented very different types of liver disease and cancer risk, allowing scientists to see whether the technology could perform reliably under different conditions.
The results were encouraging. The AI blood test successfully detected liver cancer in both populations, despite major differences in lifestyle, environmental exposure and underlying liver disease. Accuracy improved even further when the researchers combined the DNA analysis with simple clinical information such as age, sex and AFP results.
The research also answered an important scientific question. Using a new technique called MethID, the team discovered that the DNA fragments came from many sources, including cancer cells, healthy liver cells, blood vessels and immune cells. This explains why the blood test contains much richer biological information than scientists previously realized.
The investigators also observed DNA signatures linked with aflatoxin exposure among many Guatemalan participants. Even so, the overall AI model remained accurate regardless of the cause of liver cancer, suggesting the approach can adapt to different populations around the world.
The work builds on earlier studies from the same research group showing that genome-wide DNA fragment analysis can identify liver fibrosis and cirrhosis, diseases that often appear before liver cancer develops. Related DELFI technology has also been adapted to screen for lung cancer, showing the platform may have many future uses.
The researchers plan to continue testing the technology in larger real-world studies. They also hope to improve performance by combining DNA fragment analysis with protein markers and other clinical information to create even more accurate screening tools.
This research represents an important advance because it combines artificial intelligence with biological signals from the whole body rather than relying on one marker alone.
The study strengthens confidence that the technology may work across diverse populations, but widespread clinical use will require further validation. If successful, this approach could transform liver cancer screening by making early detection simpler, less invasive and more accurate.
If you care about liver health, please read studies about a diet that can treat fatty liver disease and obesity, and coffee drinkers may halve their risk of liver cancer.
For more health information, please see recent studies that anti-inflammatory diet could help prevent fatty liver disease, and results showing vitamin D could help prevent non-alcoholic fatty liver disease.
Source: Johns Hopkins University.
