Noninvasive detection of brain cancers

Current methods to diagnose brain cancer based on molecular information rely upon invasive surgical techniques to obtain tissue samples. A noninvasive way to make a diagnosis would be a game-changer for both patients and healthcare practitioners. Another challenge in treating brain cancers is the accurate diagnosis of subtypes of brain cancers, information that is used to determine prognosis and assist in treatment planning.

Scientists led by professor Gelareh Zadeh, the head of surgical oncology at University Health Network, head of the Toronto Central Regional Cancer Program at Cancer Care Ontario, and also the program medical director for the Krembil Neuroscience Centre at Toronto Western Hospital developed a simple—but highly sensitive—blood test has been found to both accurately diagnose and classify different types of brain tumors, resulting in what may be less invasive methods and better treatment planning for patients. The test is based on recent work that shows that DNA-methylation profiles from plasma reveal highly specific signatures to detect and accurately discriminate intracranial tumors that share cell-of-origin lineages. The findings are now published in the journal Nature Medicine.

The research team used DNA methylation-based liquid biopsy approach to profile hundreds of thousands of these epigenetic alterations in DNA molecules circulating in the blood. Combining this technology with machine learning, they were able to develop a highly sensitive and accurate test to detect and classify multiple solid tumors.

Together, the scientists and their teams applied this approach to intracranial brain tumor classification. They tracked the cancer origin and type by comparing patient tumor samples of brain cancer pathology, with the analysis of cell-free DNA circulating in the blood plasma from 221 patients.

Using this approach, they were able to match the circulating tumor DNA (ctDNA) to the tumor DNA, confirming their ability to identify brain tumor DNA circulating in the blood of these patients. Then, using a machine learning approach, they developed a computer program to classify the brain tumor type based solely on the ctDNA.

Previously it was not thought possible to detect any brain cancers with a blood test because of the impermeable blood-brain barrier, however the test they developed is so sensitive it can even pick up even small amounts of highly specific tumor-derived signals in the blood.

De Carvalho added that the field of identifying tumor-specific alterations in ctDNA with new, more sensitive tests in various body fluids—such as blood and urine—is now at a turning point because advanced technologies can detect and analyze even the smallest traces of cancer-specific molecular signatures from the vast quantities of circulating non-tumor DNA fragments.

Indeed molecular characterization of tumors by profiling epigenetic alterations in addition to genetic mutations gives us a more comprehensive understanding of the altered features of a tumor, and opens the possibilities for more specific, sensitive, and tumor agnostic tests.