Categories
Nevin Manimala Statistics

Machine Learning-Based Predictive Modeling Maximizes the Efficacy of mTOR/p53 Co-Targeting Therapy Against AML

Cancer Sci. 2025 Aug 11. doi: 10.1111/cas.70170. Online ahead of print.

ABSTRACT

Although mTOR signaling plays a key role in acute myeloid leukemia (AML), mTOR inhibitors have shown limited efficacy against AML in clinical trials. In this study, we found that the anti-leukemic effect of mTOR inhibition was mediated in part through the TP53 pathway. mTOR inhibition by rapamycin and TP53 activation by DS-5272 collaboratively induced the downregulation of MYC and MCL1 partly through miR-34a, thereby inducing cell cycle arrest and apoptosis in AML cells. Joint non-negative matrix factorization (JNMF) and statistical regression analysis using public AML databases revealed that monocytic AMLs with distinctive gene expression profiles were highly sensitive to mTOR inhibition, leading to the generation of an 11-gene score (Rapa-11) to predict the rapamycin sensitivity of each monocytic AML. Consistent with our in silico prediction, mouse AML cells expressing MLL-AF9, the monocytic AML with a low Rapa-11 score, were highly sensitive to rapamycin, whereas those expressing RUNX1-ETO or SETBP1/ASXL1 mutations were not. Co-treatment with rapamycin and DS-5272 had a dramatic in vivo effect on MLL-AF9-driven AML, curing 85% of the leukemic mice. Thus, machine learning-based predictive approaches identified monocytic AML with wild-type TP53 and low Rapa-11 score as a rapamycin-sensitive AML subtype and an ideal target for mTOR/p53 co-targeting therapy.

PMID:40785506 | DOI:10.1111/cas.70170

By Nevin Manimala

Portfolio Website for Nevin Manimala