A research team at the University of Michigan has developed a new machine-learning approach to mapping tumor metabolism in brain cancer patients, which may help doctors tailor treatments for individual cases of glioma. The study, published in Cell Metabolism, tested the accuracy of this computer-based "digital twin" by comparing its predictions with patient data and results from mouse experiments.
The digital twin was designed to predict how a patient's brain tumor would respond to different treatment strategies, including dietary changes and drugs that target tumor metabolism. Previous studies have shown that some gliomas can be slowed if patients avoid certain amino acids, but other tumors are able to produce these amino acids themselves and continue growing. Until now, there was no straightforward way to determine which patients might benefit from dietary interventions.
According to Deepak Nagrath, professor of biomedical engineering at U-M and co-corresponding author, "Typically, metabolic measurements during surgeries to remove tumors can't provide a clear picture of tumor metabolism-surgeons can't observe how metabolism varies with time, and labs are limited to studying tissues after surgery. By integrating limited patient data into a model based on fundamental biology, chemistry and physics, we overcame these obstacles."
The digital twin integrates patient blood samples, metabolic measurements from tumor tissue, and genetic profiles to estimate the rate at which cancer cells process nutrients—a measure known as metabolic flux.
Baharan Meghdadi, doctoral student in chemical engineering and co-first author of the study, said: "This is the first time a machine learning and AI-based approach has been used to measure metabolic flux directly in patient tumors."
To build their model—a type of deep learning called a convolutional neural network—the researchers trained it on synthetic data grounded in established biology and chemistry. They then constrained it using measurements from eight glioma patients who received labeled glucose during surgery. Comparing predictions with additional data from six patients showed that the digital twins could accurately forecast metabolic activity. Mouse experiments confirmed that only those identified by the digital twin as good candidates benefited from dietary interventions.
Daniel Wahl, Achtenberg Family Professor of Radiation Oncology and co-corresponding author on the study stated: "These results are exciting. The ability to measure metabolic activity in patient tumors could allow us to predict which metabolic therapies might work best for each patient."
The technology also predicted how tumors would react to mycophenolate mofetil—a drug targeting DNA synthesis in cancer cells—by identifying tumors capable of bypassing its effects through alternative nutrient pathways. These predictions were again supported by mouse experiments.
Wajd N. Al-Holou, assistant professor of neurosurgery and co-first author added: "This amazing tool could help doctors avoid prescribing treatments that a specific tumor is already equipped to resist, and is a way for us to move towards more targeted and personalized treatments for our patients."
With this technology, physicians could use a patient's digital twin before making recommendations about diets or medications intended to starve cancer cells.
Funding for this research came primarily from the National Institutes of Health (NIH), especially the National Cancer Institute; additional support was provided by organizations such as Damon Runyon Cancer Foundation, Forbes Scholar Award, Rogel Scholar Award (U-M), Sontag Foundation, Ivy Glioblastoma Foundation, Alex's Lemonade Stand Foundation, Chad Tough Defeat DIPG Foundation, National Institute of Neurological Disorders and Stroke (NIH), American Cancer Society and B*Cured Foundation.
Researchers from University of Alabama at Birmingham and Mayo Clinic also contributed. The team has applied for patent protection with U-M Innovation Partnerships’ assistance and is seeking partners for commercialization.
Nagrath holds appointments in chemical engineering; Wahl is also an associate professor of neurosurgery; Lyssiotis serves as professor of molecular/integrative physiology/internal medicine as well as co-director at Rogel/Blondy Center for Pancreatic Cancer.
