AI Finds Potential Treatment for Incurable Pediatric Brain Cancer | Psychology Today

Artificial intelligence (AI) machine learning is helping scientists and researchers find new treatments for diseases and conditions. A new study published in Cancer Discovery shows how AI machine learning can identify a new possible treatment for incurable pediatric brain cancer.

Diffuse intrinsic pontine glioma (DIPG) is an incurable infiltrating glioma, and the median overall survival of DIPG ranges from nine months to a year according to the scientists at The Institute of Cancer Research, London and The Royal Marsden NHS Foundation Trust who led the research study. DIPG is a type of malignant brain tumor found in the pons region of the brainstem that mainly affects children between the ages of five to seven years of age according to DIPG.org.

“25% patients with the incurable brainstem tumor DIPG harbor somatic activating mutations in ACVR1, however there are no approved drugs targeting the receptor,” wrote the researchers.

The researchers sought to find medications that can target ACVR1 mutations in diffuse intrinsic pontine glioma by using the over 40 million documents with more than a billion relationship edges in the BenevolentAI’s knowledge graph. The BenevolentAI knowledge graph enables scientists to computationally discover novel insights such as ways to repurpose existing medications for new treatments.

“In order to find approved drugs with activity against ACVR1 that could be rapidly explored in the clinical context, we searched the knowledge graph for compounds which may have inhibitory effects on the ALK2 protein,” the researchers wrote. “Inherent within this approach was the necessity to find drugs with known central nervous system (CNS) penetration.”

The BenevolentAI knowledge graph identified over 850 compounds that inhibit ALK2 enzyme activity, prioritized the potency, and identified the regulatory approval status. The AI platform found that there were not any approved medications that would penetrate the central nervous system at concentrations necessary to inhibit ALk2 with therapeutic doses. The AI machine learning algorithm did identify four compounds (vandetanib, dasatinib, crizotinib and nintedanib) that were promising, but lacked the ability to effectively cross the blood brain barrier.

“We therefore queried whether a combination of a putative ACVR1 inhibitor and another drug known to interfere with the drug transporter mediated efflux could be found which had been previously shown to be safe in humans,” the scientists shared.

In addition to using the AI platform to see if a combination would be effective, the researchers evaluated compounds for potential feedback mechanisms that could be thwarted by a combination of inhibitors.

“Using Artificial Intelligence, we identify and validate the novel combination of vandetanib and everolimus in these children based upon both signaling and pharmacokinetic synergies, experimentally and clinically,” reported the researchers.

The scientists evaluated the novel drug combination in the brains of mice with diffuse intrinsic pontine glioma and discovered that the survival in mice that received treatment was extended by 14 percent. Also, everolimus was found to increase the median concentration of vandetanib in the brains of mice by 56 percent, compared to just using vandetanib only.

According to the National Cancer Institute (NCI), Vandetanib is a drug that is already approved by the U.S. Food and Drug Administration (FDA) to treat medullary thyroid cancer that is locally advanced and cannot be removed by surgery or metastasized which means it has spread to other parts of the body. Everolimus has FDA approval to treat breast cancer, pancreatic cancer, gastrointestinal cancer, lung cancer renal cell carcinoma, and subependymal giant cell astrocytoma per the NCI.

The researchers assessed the AI-predicted drug combination of vandetanib and everolimus with four children with DIPG and confirmed ACVR1 mutations. The next step is to expand testing and enter full-scale clinical trials to evaluate is this can help children with diffuse intrinsic pontine glioma.