Researchers at Johns Hopkins Kimmel Cancer Center have developed an artificial intelligence (AI)-powered liquid biopsy that analyzes genome-wide patterns in cell-free DNA (cfDNA) fragments circulating in the blood. This test looks at how these DNA fragments are broken down and where they appear throughout the genome. Using this information, the system can identify early signs of liver fibrosis and cirrhosis, and may also detect broader indicators of chronic disease.
The study was partially funded by the National Institutes of Health and published March 4 in the journal Science. scientific translational medicine. This is the first time that this type of DNA fragmentation analysis, known as fragmentome technology, has been systematically applied to detect chronic diseases unrelated to cancer. Previously, this approach was primarily studied as a way to detect cancer.
DNA fragment patterns across the genome reveal disease signals
Liquid biopsies that measure cfDNA have already shown promise in identifying cancer. However, scientists have not extensively investigated its potential for diagnosing other diseases. In this new study, researchers performed whole-genome sequencing on cfDNA samples taken from 1,576 individuals with liver disease and other medical conditions. By examining DNA fragments across the genome, they searched for patterns that could signal disease.
The researchers analyzed both the size of the DNA fragments and their distribution across the genome, including little-studied repetitive DNA regions. Each analysis included approximately 40 million fragments spanning thousands of genomic regions, generating a huge dataset compared to most liquid biopsy tests.
Machine learning algorithms processed this information to identify fragmentation patterns associated with the disease. Researchers used these patterns to create a classification system that sensitively detects early liver disease, advanced fibrosis, and cirrhosis.
“This builds directly on earlier fragmentomic studies in cancer, but now focuses on chronic diseases using AI and genome-wide fragmentation profiles of cell-free DNA,” said Victor Belculescu, MD, co-director of the Cancer Genetics and Epigenetics Program at the Johns Hopkins Kimmel Cancer Center and co-senior author of the study. “For many of these diseases, early detection can make a big difference, and liver fibrosis and cirrhosis are key examples. Liver fibrosis is reversible in its early stages, but if left undetected, it can progress to cirrhosis and ultimately increase the risk of liver cancer.”
Why DNA fragment analysis is different
Unlike many liquid biopsy methods that search for specific cancer-related gene mutations, fragmentomic approaches focus on how DNA fragments are cut, packaged, and distributed throughout the genome. The researchers say this broader perspective could eventually allow the method to be applied to conditions other than cancer, such as diseases that may increase the risk of cancer. The study was also co-led by Dr. Robert Scharpf, professor of oncology, and Dr. Jill Farren, assistant professor of oncology.
“The fact that we’re not looking for individual mutations is what makes this study so powerful,” says lead author Akshaya Annapragada, MD/PhD. A student in Belculescu’s laboratory. “We are analyzing whole fragments that contain vast amounts of information about a person’s physiological state. The scale of these data, combined with machine learning, enables the development of specific classifiers for different health conditions.”
Early detection could benefit millions at risk
Belculescu points out that about 100 million people in the United States have liver diseases that increase the risk of cirrhosis and liver cancer. Current blood-based fibrosis tests often have low sensitivity, especially in the early stages of the disease. Standard blood markers usually fail to detect early fibrosis and only identify cirrhosis about half of the time. Imaging techniques such as specialized ultrasound and magnetic resonance scans can be helpful, but these tools require equipment that is not always available.
“Many people at risk don’t know they have liver disease,” Belculescu says. “If we can intervene early, before fibrosis progresses to cirrhosis or cancer, the impact could be greater.”
Identifying these prodromal symptoms early could allow doctors to treat the underlying disease sooner, potentially preventing the development of cancer, he added.
Research origins and fragmentomic comorbidity index
This research started in 2023 cancer discovery A study led by Velculescu focused on fragments of liver cancer. While studying patients with liver tumors, scientists noticed that some patients with fibrosis or cirrhosis showed mostly normal fragmentation profiles but contained subtle DNA signals associated with the disease. This observation led the team to specifically examine fragmentomic patterns associated with liver fibrosis and cirrhosis.
In a separate analysis of 570 people with suspected serious illness, researchers created a fragmented comorbidity index. The measure distinguished between those with high and low Charlson Comorbidity Index scores, a widely used measure that estimates how additional health conditions affect a person’s risk of death. Fragmentome-based indices independently predicted overall survival and in some cases proved to be more specific than traditional inflammatory markers. Certain signs of fragmentation also appear to be associated with poor clinical outcomes.
“The fragmentome serves as a basis for building different classifiers for different diseases. Importantly, these classifiers are disease-specific and do not cross-react,” says Annapragada. “The liver fibrosis classifier is different from the cancer classifier; it is a unique disease-specific test built from the same underlying platform.”
Possibility to detect other chronic diseases
The study also included people at high risk for various medical conditions. Researchers observed fragmentomic signals associated with cardiovascular, inflammatory, and neurodegenerative diseases. However, the study population did not include enough cases to construct separate disease classifiers for each of these symptoms. Rather, the findings suggest that the technology may eventually have broader medical applications, which the researchers plan to investigate in future studies.
The liver fibrosis assay described in this study remains a prototype and has not yet been implemented as a clinical trial. The team’s next steps include refining and validating the liver disease classifier and investigating fragmentary features associated with other chronic diseases.
Researchers and funding
The research team includes Belculescu, Annapragada, Scharpf, and Farren, as well as Zachariah Foda, Hope Orjuela, Carter Norton, Shashikant Kuhl, Noushin Niknahus, Sarah Short, Keerti Boyapati, Adrianna Bartolomucci, Dimitrios Matios, Michael No, Chris Cherry, Jacob Carey, Alessandro Real, and Brian Cesnik participated. Nick Dracopoli, Jamie Medina, Nicolas Varpescu, Daniel Bloom, Sara Bakas, Vilmos Adlev, Amy Kim, Stephen Balin, Gregory Kirk, Andrei Solop, Razvan Iacov, Speranta Yacob, Liana Gheorghe, Simona Dima, Katherine McGlynn, Manuel Ramirez-Zea, Klaus Vertoft, Julia Johansen, Jon Groopman.
Funding for the study was provided in part by the Dr. Miriam and Sheldon G. Adelson Medical Research Foundation, the SU2C Intime Lung Cancer Interception Dream Team Grant, the Stand Up to Cancer Dutch Cancer Society International Translational Cancer Research Dream Team Grant, the Gray Foundation, the Tina Brozman Honorary Foundation, the Commonwealth Foundation, the Mark Cancer Research Foundation, the Danaher Foundation, the ARCS Metro Washington Chapter, and the Dan Y. Chang Family of the AACR. Training Scholar Awards, Cole Foundation and National Institutes of Health grants CA121113, CA006973, CA233259, CA062924, CA271896, T32GM136577, T32GM148383, and DA036297.
