In 377 patients from Guatemala and Romania, a locked DELFI classifier detected hepatocellular carcinoma across stages and causes, outperforming AFP, the standard blood biomarker for liver cancer.
Using less than 1 mL of plasma and low-coverage whole-genome sequencing, the test read genome-wide cfDNA fragmentation patterns rather than single mutations, aiming to catch cancers current ultrasound-based surveillance often misses.
Adding AFP plus clinical risk factors such as age and sex improved diagnostic performance further, yielding high sensitivity for both early- and late-stage HCC in the validation cohorts.
MethID, a new tissue-of-origin algorithm, found increased hepatocyte-derived cfDNA in HCC and also picked up vascular and immune-cell signals, while fragmentome analysis captured chromosome changes and an aflatoxin-linked mutation signature.
HCC makes up about 75% of primary liver cancers and has rising U.S. mortality; the authors said prospective and cost-effectiveness studies are still needed before the multimodal liquid biopsy enters routine surveillance.