Scientists at Johns Hopkins Kimmel Cancer Center have successfully used DELFI, an AI-powered blood test that detects lung cancer, to detect liver cancer.
In 2021, scientists at the facility developed DELFI, which stands for “DNA evaluation of fragments for early interception.” It is a blood test that identifies cell-free DNA (cfDNA) or fragmentation changes in DNA from cancer cells in the bloodstream.
The system has successfully detected more than 80 percent of liver cancer in 724 people in the U.S. The researchers said this was the first study to be validated in two high-risk populations across ethnic groups with different causes of liver cancer.
DELFI detects cancer by evaluating millions of cell-free DNA fragments for abnormalities, including in the size and quantity of the DNA across genomic areas. The researchers said DELFI requires low-coverage sequencing, making this technology cost-effective for cancer screenings.
The study used cell-free DNA fragments recovered from plasma samples. The researchers created a DELFI score by analyzing fragmentation patterns across each sample. The findings were reported at the American Association for Cancer Research Special Conference: Precision Prevention, Early Detection, and Interception of Cancer.
Increased early detection of liver cancer could save lives
Victor Velculescu
“Increased early detection of liver cancer could save lives, but currently available screening tests are underutilized and miss many cancers,” Johns Hopkins Kimmel professor of oncology Victor Velculescu, M.D., Ph.D. said.
Study co-senior author Amy Kim, M.D. believes that compared to the usual blood test, this breakthrough method has the potential to boost early cancer detection. Currently, less than 20 percent of the high-risk population gets checked for liver cancer due to accessibility and test quality issues.
“This new blood test can double the number of liver cancer cases detected, compared to the standard blood test available, and increase early cancer detection,” Kim said.
The researchers are looking to validate the method for clinical use.
