A new liquid biopsy approach developed by Johns Hopkins Kimmel Cancer Center investigators could revolutionize brain cancer detection by identifying circulating DNA fragments from tumors and immune cells in blood samples, potentially enabling earlier diagnosis.
The research results were published in the April 29, 2025 edition of Cancer Discovery. The research was partially supported by the National Institutes of Health. [1]
Hard to detect
Brain cancers are notoriously hard to detect. As a result, many patients receive a diagnosis at a late stage after experiencing symptoms like headaches, seizures or cognitive problems. Late diagnosis limits treatment options and often leads to worse patient outcomes. Previous attempts to find biomarkers in a patient’s blood have failed because the blood-brain barrier that protects the brain from infection may also prevent potential biomarkers from reaching the blood.
To overcome this challenge, Victor E. Velculescu, M.D., Ph.D., co-director of cancer genetics and epigenetics at the cancer center, and his colleagues created a next generation liquid biopsy (blood-based assay) approach that they’ve previously shown effective for detecting lung cancer and is now in clinical use for lung cancer screening. The technique uses machine learning, a form of artificial intelligence (AI), to identify patterns of DNA fragments circulating in the blood linked to brain tumors. It also searches for repeating patterns in the genome linked to brain cancer.
In this study, the scientists analyzed genome-wide cell-free (cfDNA) fragmentomes, including fragmentation profiles and repeat landscapes, from the plasma of individuals with brain cancer (n=148) or without brain cancer (n=357).
Using machine learning analyses of cfDNA fragmentome features, investigators successfully detected brain cancer in about three-quarters of the time in a cohort of 505 patients from the United States and South Korea across all grade gliomas (AUC=0.90, 95% CI: 0.87-0.93) and these results were validated in an independent prospectively collected cohort of about 95 patients from Poland. Previous liquid biopsy approaches have only been able to detect brain cancers less than 10% of the time. [1]
“Our next-generation AI liquid biopsy approach combining DNA fragments and repeating genome patterns may accelerate brain cancer diagnosis,” says Velculescu, senior author of the study. “It may ultimately allow patients to get care earlier, improving their treatment outcomes.”
“One factor that led to the success of this approach is that it detected immune changes associated with brain cancer. Patients with brain cancer have immune suppression throughout their body and a unique immune cell profile in the blood. These immune differences do not have to pass through the blood-brain barrier to be detected,” explained the study’s lead author Dimitrios Mathios, M.D., who completed the research during his fellowship in cancer immunology, genetics and epigenetics at Johns Hopkins.
“We finally have a way of detecting brain cancer that takes advantage of its unique characteristics, including genome-wide fragmentation from brain cancer and immune cell changes,” added Mathios, who is now an assistant professor of neurosurgery and genetics at Washington University School of Medicine in St. Louis.
“Some of the signals we detect come from the cancer itself in the brain, and some of them come from the immune system’s response to the cancer.”
Next, the team ran a computer simulation of what would happen if their liquid biopsy method were used to screen the estimated 10 million patients presenting with headaches to the emergency room or primary care clinics each year. Usually, these individuals are only referred for brain imaging if their physician suspects an underlying problem. But in the simulation, sending patients who received a liquid biopsy result suggestive of brain cancer for imaging would result in the detection of nearly 1,700 more cancer cases than the usual approach.
The next step for the team will be designing a prospective trial to confirm these findings in larger patient populations who are at increased risk of brain cancer.
Reference
[1] Mathios D, Niknafs N, Annapragada AV, Bobeff EJ, Chiao EJ, Boyapati K, Boyapati K, Short S, Bartolomucci AL, Cristiano S, Koul S, Vulpescu NA, Ferreira L, Medina JE, Bruhm DC, Adleff V, Podstawka M, Stanisławska P, Park CK, Huang J, Gallia GL, Brem H, Mukherjee D, Caplan JM, Weingart J, Jackson CM, Lim M, Phallen J, Scharpf RB, Velculescu VE. Detection of brain cancer using genome-wide cell-free DNA fragmentomes. Cancer Discov. 2025 Apr 29. doi: 10.1158/2159-8290.CD-25-0074. Epub ahead of print. PMID: 40299319.
Featured image:Release of cell-free DNA and altered blood cells in patients with cancer.
Illustration credit: © 2025 Carolyn Hruban; Johns Hopkins Kimmel Cancer Center.
DOI
 Shows Promise in Detecting Brain Cancer.){kind=link}