Several subtypes of acute myeloid leukemia (AML) are resistant to treatment. Abnormal gene splicing—the process of joining DNA fragments to create new genetic sequences—occurs in many cancers, including AML.

In a collaborative study published online on May 7 in Science Translational Medicine, co-corresponding author, Amit Verma, M.B.B.S., and colleagues, looked at gene expression data from over one thousand AML patients to determine whether categorizing AML subtypes by their abnormally spliced gene arrays could lead to better treatments. The researchers developed a machine learning algorithm, dubbed Oncosplice, that identified splicing patterns across both adult and pediatric AML. By focusing on one major splicing subtype, the researchers determined that the gene Protein Arginine Methyltransferase 5 (PRTMD5) was a potential therapeutic target and showed that a PRTMD5 inhibitor could correct splicing defects in AML cells. The finding indicates that PRTMD5 inhibitors could be useful against cases of AML characterized by this gene-splicing subtype.

These studies were done in collaboration with co-senior authors Nathan Salomonis, Ph.D., Daniel Starczynowski, Ph.D., and H. Leighton Grimes, Ph.D., at the Cincinnati Children's Hospital Medical Center. Dr. Verma is associate director of translational science at the National Cancer Institute-designated Montefiore Einstein Comprehensive Cancer Center, interim chair of the department of oncology, the Susan Resnick Fisher Academic Chair in Brain Cancer Research, and professor of oncology, of medicine and of developmental & molecular biology at Einstein.