Researchers at the Johns Hopkins Kimmel Cancer Center validated an artificial intelligence-powered blood test that accurately detected liver cancer in people from two geographically and biologically distinct populations and also uncovered the underlying biological signals that make the test effective, according to a July 31 announcement.
The findings, published in Cell Press Blue, build on earlier development of the DELFI (DNA Evaluation of Fragments for Early Interception) liquid biopsy platform. This technology analyzes millions of fragments of cell-free DNA circulating in the bloodstream. The new research validates a previously developed liver cancer classifier in independent high-risk populations and provides new insights into its biological basis. The study also advances research from March 2026, where a similar genome-wide fragmentome technology was shown to detect liver fibrosis and cirrhosis—conditions often preceding liver cancer.
"Our earlier studies showed that fragmentome analyses could detect liver cancer and, more recently, chronic liver diseases that increase cancer risk," said Victor Velculescu, M.D., Ph.D., co-senior author of the study. "This study demonstrates that the approach works with high performance across different patient populations while revealing the biological signals in the bloodstream that make this type of detection possible."
To assess whether DELFI could reliably detect liver cancer regardless of its cause, investigators analyzed blood samples from 377 people from Guatemala and Romania with and without hepatocellular carcinoma—the most common form of liver cancer. Most Romanian participants had disease related to viral hepatitis or alcohol use; Guatemalan participants primarily had metabolic disease, obesity or diabetes, with many exposed to aflatoxin—a toxin linked to liver cancer. Despite these differences, researchers found consistent detection across both groups. When combined with AFP testing and clinical risk factors such as age and sex, early- and late-stage cancers were identified with greater sensitivity than existing blood tests alone.
Using a method called MethID for tracing DNA fragment origins, researchers demonstrated that DELFI captures not only tumor cell signals but also those from surrounding cells responding to cancer—including immune cells and blood vessels. "As a result, these DNA fragments contain much more information than whether cancer is present," said Zachariah Foda, M.D., Ph.D., co-senior author. "It tells us where these fragments originate and how they change during cancer development... allowing us to better understand the biology of the disease and improve our ability to detect it." Molecular signatures differing between regions were observed—for example, distinctive mutation patterns associated with aflatoxin exposure—but overall classifier effectiveness remained robust regardless of underlying cause.
The study suggests genome-wide fragmentome analysis can capture both universal features of liver cancer as well as region-specific molecular changes—making it adaptable globally. Researchers indicated future studies will focus on prospective clinical validation and refining multimodal approaches combining fragmentome analysis with protein biomarkers.
