AI blood test sniffs out hidden brain tumors

Because AI detects previously invisible tumor signals, brain cancer will soon be absorbed a few months before symptoms appear. Researchers at John Hopkins have developed a liquid biopsy technology that recognizes brain cancer with unprecedented accuracy and has the potential to change the way these deadly tumors are diagnosed.

Brain cancer testing has long frustrated doctors. Most patients only get diagnosed after suffering headaches, seizures, or cognitive problems – usually the tumor has grown significantly. The brain’s protective blood barrier is essential to preventing pathogens and has historically prevented cancer markers from appearing in blood tests.

“Our next-generation AI liquid biopsy approach combines DNA fragments and repeat genomic patterns that may accelerate brain cancer diagnosis,” said Victor E. Velculescu, MD, Ph.D, co-director of Cancer Genetics and Epigenetics at the Kimmel Cancer Center.

The breakthrough lies in the dual approach of testing. The researchers not only focus on tumor DNA that may cross the blood-brain barrier, but rather create a machine learning system that not only identifies fragments of circulating tumor DNA, but also identify changes in the entire body in response to brain cancer.

“We finally have a way to detect brain cancer that can take advantage of its unique features, including genome-wide fragments of brain cancer and changes in immune cells,” explained lead study author Dimitrios Mathios, MD. “Some of the signals we found come from the cancer itself in the brain, some of which come from the immune system’s response to cancer.”

This dual detection strategy has proven to be very effective. The researchers tested their approach to 505 patients from the United States and South Korea, successfully identifying about 75% of the time, much better than the previous method of finding less than 10% of the cases. They then validated these results in about 95 patients from Poland.

For millions of headaches, there is enough headache every year that this test can provide critical early detection. The team conducted computer simulations showing that if only their liquid biopsy was used to screen about 10 million patients who had headaches in emergency rooms or primary care offices, nearly 1,700 cancer cases would be found than the current practice.

The technology builds on Velculescu’s previous work on liquid biopsy for lung cancer, a technology for clinical use in lung cancer screening. The brain cancer version uses more complex AI to identify patterns in DNA fragments and repeated genomic sequences associated with brain tumors.

It is particularly interesting to note that brain cancer patients are found to exhibit immunosuppression throughout the body and have a unique immune cell profile in the blood without crossing the blood-brain barrier. This immune signature becomes a key component of the test effectiveness.

Funded in part by the National Institutes of Health and Private Foundation, the study represents a potential paradigm metastasis for brain cancer diagnosis. Current regimens often rely on doctors who determine whether symptoms guarantee brain imaging—a subjective process that may delay diagnosis.

Although promising, the technology still needs to be validated in a large prospective trial that the team is planning. They will pay special attention to people with elevated risk of brain cancer.

Several organizations contributed to the study, including Johns Hopkins, Washington University in St. Louis, Lodz Medical University in Poland and Stanford University. Some researchers have revealed financial interests in Delfi Diagnostics, which has licensed some basic technologies.

If successful in larger trials, this blood test may eventually become part of a routine screening for people with persistent headaches or at a higher risk of brain tumors, and when treatment is most effective, it may save thousands of lives through early intervention.

Related