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

Area of Science:

• Hematology
• Oncology
• Molecular Biology

Background:

• mTOR signaling is crucial in acute myeloid leukemia (AML), but mTOR inhibitors alone have limited clinical efficacy.
• The TP53 pathway's role in AML treatment response requires further elucidation.

Purpose of the Study:

• To investigate the combined therapeutic effect of mTOR inhibition and TP53 activation in AML.
• To identify predictive biomarkers for response to mTOR-targeted therapies in AML subtypes.

Main Methods:

• Utilized rapamycin (mTOR inhibitor) and DS-5272 (TP53 activator) in AML cell lines and mouse models.
• Employed joint non-negative matrix factorization (JNMF) and statistical regression on public AML databases.
• Developed an 11-gene score (Rapa-11) to predict rapamycin sensitivity.

Main Results:

• Combined mTOR/TP53 inhibition downregulated MYC and MCL1 via miR-34a, inducing apoptosis and cell cycle arrest in AML cells.
• Monocytic AMLs with specific gene expression profiles, identified by the Rapa-11 score, were sensitive to rapamycin.
• Co-treatment with rapamycin and DS-5272 cured 85% of MLL-AF9-driven AML mouse models.

Conclusions:

• Monocytic AML with wild-type TP53 and a low Rapa-11 score is a sensitive subtype to mTOR inhibition.
• mTOR/p53 co-targeting therapy represents a promising strategy for a subset of AML patients.
• Machine learning identified a predictive biomarker for targeted AML therapy.