M-PACT sets new standard for pediatric brain tumor diagnostics, treatment monitoring

Liquid biopsies, which test body fluids that contain cancerous material, including circulating tumor DNA (ctDNA), are a noninvasive way to learn about a cancer's biology. However, technological limitations with the small amount of ctDNA available from pediatric brain tumor liquid biopsies have previously stymied broad use of the approach for those patients. To address this, St. Jude Children's Research Hospital scientists, in collaboration with scientists at the Hopp Children's Cancer Center Heidelberg (KiTZ), German Cancer Research Center (DKFZ) and other international centers, created Methylation-based Predictive Algorithm for CNS Tumors (M-PACT). M-PACT uses AI to sift through ctDNA in cerebrospinal fluid and molecularly classify tumors based on their DNA methylation pattern. The resource, published today in Nature Cancer, sets a new standard for pediatric brain tumor diagnostics, treatment monitoring and surveillance.

In a compelling demonstration of functionality, M-PACT successfully identified 92% of brain tumors in a benchmarking test; it can also differentiate relapse from secondary tumors and can track if a cancer is getting more aggressive or responding to treatment with no extra input. Beyond brain tumors, M-PACT has the potential to be broadly applicable to many cancer types.

This is a next-generation assay and computational framework that we've optimized and applied across a range of pediatric brain tumor patients. M-PACT is about taking liquid biopsy to another level in pediatric neuro-oncology and applying the technology across many different clinical scenarios."

Paul Northcott, PhD, corresponding author, Center of Excellence in Neuro-Oncology Sciences (CENOS) director and Department of Developmental Neurobiology member

"Taking liquid biopsy to another level"

A key function of tissue-based biopsies is to describe DNA methylation patterns, the chemical modifications to DNA that help regulate gene activity. In cancer, these patterns often become abnormal in ways that act like a fingerprint for specific tumor types, guiding clinicians to the cancer's identity. While this approach is potent using tissue biopsy samples, the same classifiers fall short in liquid biopsies.

"Traditionally, methylation-based diagnostics for ctDNA use classifiers designed for tumor tissue, which have higher DNA input," said co-first author Katie Han, a PhD student in the St. Jude Graduate School of Biomedical Sciences and Department of Developmental Neurobiology and MD candidate at University of Tennessee Health Sciences Center. "We reversed the usual flow and designed M-PACT for ctDNA itself with applicability to tissue, instead of the other way around."

M-PACT utilizes a novel deep neural network training strategy using more than 5,000 DNA methylation profiles across roughly 100 tumor entities. This brings methylation-based ctDNA analysis up to, and beyond, current standards seen from tissue biopsies. 

"We developed M-PACT by computationally mixing large reference datasets with normal cell-free DNA datasets," said co-first author Kyle Smith, PhD, Department of Developmental Neurobiology. "We trained it extensively and showed that even tiny amounts of ctDNA can be accurately classified."

As proof of concept, the researchers used M-PACT to make diagnoses at the time of surgery using cerebrospinal fluid only and demonstrated its potential use during treatment and follow-up. "If a tumor reoccurs years later, M-PACT can reliably determine whether it's a true relapse or a second malignancy," Northcott said.

M-PACT gives unmatched insight into cancer microenvironment

M-PACT's sensitivity enables it to look beyond tumor cells to identify noncancerous cell types contributing to the tumor microenvironment. "Most DNA in cerebrospinal fluid is from something else, the 'negative space' of the tumor, which we previously ignored," Smith said. "Now we can predict what fraction comes from T cells, B cells, or other sources."

This opens questions about how cancers manipulate normal cells and microenvironments - everything involved in creating the perfect storm. "M-PACT provides us with a new lens to monitor disease evolution, especially during therapy, when tissue sampling isn't typically done," said Han. "Now we can start to see how both the tumor and its microenvironment change with therapeutic pressure."

While M-PACT is immediately applicable to pediatric brain tumors, Northcott is confident its robust framework offers a wide range of potential use cases. "Although we applied this to pediatric brain tumors, it will clearly be useful in other solid tumors and hematological malignancies as well," he 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 power of team science

This body of work relied on tightly knit partnerships between St. Jude investigators, scientists at the KiTZ, DKFZ and several other participating institutions, which were integral to building the large sample cohort of clinically annotated liquid biopsy samples. "Our study is a prime example of what can be accomplished when we approach science as a team and bring together complementary skills and expertise to achieve a common goal," said Northcott. "The technical and computational innovations that were fundamental to the success of this study would not have been possible without our international network of collaborators."