A novel artificial intelligence (AI) blood testing technology—known as DELFI (DNA evaluation of fragments for early interception)—may successfully detect more than 80% of hepatocellular carcinomas, according to a new study published by Foda et al in Cancer Discovery. The findings were also presented at the American Association for Cancer Research (AACR) Special Conference: Precision Prevention, Early Detection, and Interception of Cancer.

DELFI Background

The DELFI test was developed and used to detect lung cancer in a 2021 study. It measures the way DNA is packaged inside the nucleus of a cell by studying the size and amount of cell-free DNA present in the circulation from different regions across the genome. Healthy cells carefully package different regions of the genome into various compartments. The nuclei of cancer cells, by contrast, haphazardly combine items from across the genome. When cancer cells die, they release DNA fragments in a chaotic manner into the bloodstream. The DELFI test is capable of identifying the presence of cancer by examining millions of cell-free DNA fragments for abnormal patterns—an approach that only requires low-coverage sequencing, enabling the screening technology to be cost-effective.

In the most recent study, investigators used the DELFI technology on blood plasma samples obtained from patients across the United States, the European Union, and Hong Kong to detect hepatocellular carcinoma. The researchers believe the study may be the first genome-wide fragmentation analysis to be independently validated in two high-risk populations and across different racial and ethnic groups who had different carcinogenesis-associated factors.

Researchers estimated that 400 million people worldwide are at higher risk of developing hepatocellular cancer because of cirrhosis from chronic liver diseases—including chronic viral hepatitis or nonalcoholic fatty liver disease. “Increased early detection of liver cancer could save lives, but currently available screening tests are underutilized and miss many cancers,” highlighted co–lead study author Victor E. Velculescu, MD, PhD, Professor of Oncology and Pathology, and Co-Director of Cancer Genetics and Epigenetics at the Sidney Kimmel Cancer Center at the Johns Hopkins University School of Medicine.

Research Methods and Findings

Among the 724 plasma samples studied were 501 samples from 75 patients in the United States and European Union with hepatocellular carcinoma, whose samples were used to train and validate the machine learning model. For validation, an additional 223 plasma samples from patients in Hong Kong were analyzed—which included samples from 90 patients with hepatocellular carcinoma, 66 patients with hepatitis B virus, 35 patients with hepatitis B virus–related liver cirrhosis, and 32 patients with no underlying risk factors.

After researchers performed the DELFI test on cell-free DNA fragments isolated from the plasma samples, they analyzed the patterns of fragmentation across each sample to develop a DELFI score. The scores were low for cancer-free patients with viral hepatitis (with a median DELFI score of 0.078) or cirrhosis (0.080), but tended to be 5 to 10 times higher for the 75 patients with hepatocellular cancer—with high scores observed across all cancer stages, including early-stage disease (with median DELFI scores of 0.46 for stage 0, 0.61 for stage A, 0.83 for stage B, and 0.92 for stage C). In addition, the test detected fragmentation changes in the content and packaging of hepatic cancer genomes, including from genome regions associated with liver-specific activity.

The DELFI technology detected hepatic cancers at their earliest stages, with an overall sensitivity of 88% and a specificity of 98% among people at average risk. In samples collected from those at high risk of hepatocellular carcinoma, the test had 85% sensitivity and 80% specificity.

Conclusions

“Currently, less than 20% of the high-risk population get[s] screened for liver cancer due to accessibility and suboptimal test performance. This new blood test can double the number of liver cancer cases detected—compared [with] the standard blood test available—and increase early cancer detection,” emphasized co–senior study author Amy K. Kim, MD, Assistant Professor of Medicine at the Johns Hopkins University School of Medicine.

The findings from this study may provide a biological basis for changes in cell-free DNA fragmentation in patients with hepatic cancer and provide an accessible approach for noninvasive cancer detection in the future. The researchers hope to validate this novel approach in larger studies for clinical use.

Disclosure: The research in this study was supported by the Dr. Miriam and Sheldon G. Adelson Medical Research Foundation, Stand Up to Cancer–Dutch Cancer Society International Translational Cancer Research Dream Team Grant, the Gray Foundation, the Commonwealth Foundation, Stand Up to Cancer–Intercept Lung Cancer Through Immune, Imaging, and Molecular Evaluation Lung Cancer Interception Dream Team Grant, the Mark Foundation for Cancer Research, a Delfi Diagnostics research grant, research grants from the National Institutes of Health, and a Department of Defense Congressionally Directed Medical Research Programs award. For full disclosures of the study authors, visit aacrjournals.org.