A new artificial intelligence tool, M-PACT (Methylation-based Predictive Algorithm for CNS Tumors), is now able to classify pediatric brain tumors with 92% accuracy using liquid biopsies, researchers at St. Jude Children’s Research Hospital announced today.
The breakthrough, detailed in the February 17, 2026, issue of Nature Cancer, addresses a longstanding challenge in pediatric oncology: the limited amount of circulating tumor DNA (ctDNA) found in cerebrospinal fluid, which has previously hindered the widespread use of liquid biopsies for these cancers. Liquid biopsies offer a noninvasive method to analyze a cancer’s biology by testing body fluids for cancerous material.
M-PACT was developed through a collaboration between St. Jude and scientists at the Hopp Children’s Cancer Center Heidelberg (KiTZ), the German Cancer Research Center (DKFZ), and other international centers. The algorithm sifts through ctDNA in cerebrospinal fluid, classifying tumors based on their DNA methylation patterns.
“This sets a new standard for pediatric brain tumor diagnostics, treatment monitoring and surveillance,” according to a statement released by St. Jude. The tool’s capabilities extend beyond initial diagnosis; it can also differentiate between tumor relapse and secondary tumors, and track a cancer’s response to treatment without requiring additional testing.
Researchers highlighted the potential for M-PACT to be applicable to a wide range of cancer types beyond brain tumors, though initial validation has focused on pediatric cases. Co-first authors of the Nature Cancer study are Katie Han and Kyle Smith, PhD, with Paul Northcott, PhD, serving as the corresponding author. Anna Kostecka, PhD and Taha Soliman, PhD, also contributed to the research, all from the St. Jude Center of Excellence in Neuro-Oncology Sciences (CENOS) and Department of Developmental Neurobiology.
