M-PACT: AI Accurately Classifies Pediatric Brain Tumors from Liquid Biopsy DNA | Nature Cancer Study

A new artificial intelligence-powered tool is offering hope for more accurate and timely diagnoses for children with brain tumors. Researchers have developed Methylation-based Predictive Algorithm for CNS Tumors, or M-PACT, a deep neural network capable of classifying these tumors from remarkably small samples of circulating tumor DNA (ctDNA) found in cerebrospinal fluid.

The development, detailed in a February 17, 2026 publication in Nature Cancer, addresses a significant challenge in pediatric neuro-oncology: the limited amount of ctDNA available from liquid biopsies. Liquid biopsies, which analyze body fluids for signs of cancer, offer a less invasive alternative to traditional tissue biopsies, but their effectiveness has been hampered by the difficulty of obtaining sufficient material for analysis in young patients.

“This is a next-generation assay and computational framework that we’ve optimized and applied across a range of pediatric [patients with] brain tumors,” said Paul A. Northcott, PhD, Director, Center of Excellence in Neuro-Oncology Sciences (CENOS), and Member, Department of Developmental Neurobiology, St. Jude Children’s Research Hospital. “M-PACT is about taking liquid biopsy to another level in pediatric neuro-oncology and applying the technology across many different clinical scenarios.”

Overcoming the Challenges of Limited Samples

Traditionally, analyzing DNA methylation patterns – chemical modifications to DNA that can influence gene expression – required larger tissue samples. Researchers recognized the need for a method specifically tailored to the small quantities of ctDNA present in cerebrospinal fluid. M-PACT was designed to overcome this hurdle.

“Traditionally, methylation-based diagnostics for circulating tumor DNA use classifiers designed for tumor tissue, which have higher DNA input,” explained Katie Han, a PhD student at St. Jude and MD candidate at University of Tennessee Health Sciences Center. “We reversed the usual flow and designed M-PACT for circulating tumor DNA itself with applicability to tissue, instead of the other way around.”

The team achieved this by computationally combining extensive reference datasets with datasets representing normal cell-free DNA. This approach allowed M-PACT to accurately identify tumor-specific methylation patterns even with minimal input material. Kyle Smith, PhD, of the Department of Developmental Neurobiology at St. Jude, explained that the algorithm was developed by “computationally mixing large reference datasets with normal cell-free DNA datasets.”

High Accuracy in Tumor Classification

The performance of M-PACT was rigorously evaluated in two cohorts of patients. In a benchmarking cohort of 79 patients, the algorithm achieved an impressive 92% accuracy in classifying embryonal central nervous system tumors. This accuracy was maintained in a validation cohort of 58 patients, where M-PACT correctly classified tumors 88% of the time.

Beyond simply identifying the presence of a tumor, M-PACT can also provide valuable information about its characteristics. The deep neural network is capable of performing methylation-based cellular deconvolution, allowing researchers to distinguish between different cell types within the tumor. It can also accurately detect copy-number variations – alterations in the number of copies of specific DNA segments – within the ctDNA.

Monitoring Treatment Response and Detecting Relapse

The potential applications of M-PACT extend beyond initial diagnosis. The algorithm can be used to monitor a patient’s response to treatment, tracking changes in the tumor’s DNA methylation pattern over time. This allows clinicians to assess whether a therapy is effective or if adjustments are needed.

Perhaps even more critically, M-PACT can help differentiate between a true tumor relapse and the development of a secondary malignancy – a new cancer that arises as a late effect of previous treatment. “If a tumor reoccurs years later, M-PACT can reliably determine whether it’s a true relapse or a second malignancy,” Dr. Northcott noted.

Potential Beyond Pediatric Brain Tumors

While M-PACT was initially developed for pediatric brain tumors, researchers believe its underlying principles could be applied to a wider range of cancers. Dr. Northcott and his team are actively exploring this possibility.

“Although we applied this to pediatric brain tumors, it will clearly be useful in other solid tumors and hematological malignancies as well,” Dr. Northcott said. “The informatics will need to grow to classify the full scope of cancer types diagnosed in children, but we’ve developed something quite powerful that is likely to be more broadly adopted in the community.”

The development of M-PACT represents a significant step forward in the field of liquid biopsy and promises to improve the diagnosis, treatment, and monitoring of pediatric brain tumors, and potentially other cancers as well. The study was supported by a number of organizations including the Verein unser_kind, the Forschungsgesellschaft für Cerebrale Tumore, and the St. Baldrick’s Foundation, among others.