NCI-funded researchers developed and recently validated a genome-wide artificial intelligence technology for detecting hepatocellular carcinoma—the most common type of liver cancer.

The researchers used a new omics approach called DELFI (i.e., DNA evaluation of fragments of early interception) to analyze small pieces of DNA (cfDNA, or “cell-free DNA”) in the blood stream. These fragments carry unique signatures, making them ideal biomarkers for disease detection among other applications.

The next step will be to test this approach in larger, clinical samples. As noted by NCI Program Officer Dr. Miguel R. Ossandon, “If this technology is successful in large validation studies, it could be rapidly translated to clinical practice, leading to commercial tests and offering a noninvasive, cost-effective way for early detection of cancer.” Earlier detection would improve outcomes for a disease that results in more than 800,000 deaths worldwide.

DELFI may have other applications as well. According to the study’s investigator, Dr. Victor E. Velculescu, of the Sidney Kimmel Comprehensive Cancer Center, Johns Hopkins University School of Medicine, DELFI was successful in detecting small-cell lung cancer, as reported in a study in Nature Communications in 2021 and in two ongoing nation-wide prospective clinical trials.

With greater numbers of patients, genome-wide cfDNA fragmentation profiles could improve with machine learning algorithms, enabling researchers to detect the origins of many types of cancers. “In the future, this promising next generation liquid biopsy approach, which combines low-coverage, whole genome sequencing with machine learning algorithms, could make widely accessible early detection for multiple cancer types a reality,” said Dr. Velculescu.

For more information, see the article “Detecting liver cancer using cell-free DNA fragmentomes,” published in Cancer Discovery.