A machine learning approach to integrate big data for precision medicine in acute myeloid leukemia

Area of Science:

• Oncology
• Genomics
• Pharmacology

Background:

• Pathologically similar cancers exhibit variable responses to chemotherapy.
• Personalized medicine approaches are needed to match patients with effective drug regimens.
• Acute myeloid leukemia (AML) is a complex disease with diverse therapeutic outcomes.

Purpose of the Study:

• To develop and validate a computational method for identifying robust molecular markers of drug sensitivity in AML.
• To improve the accuracy of predicting patient response to chemotherapy.
• To discover novel therapeutic targets for AML treatment.

Main Methods:

• Utilized genome-wide gene expression profiles and in vitro drug sensitivity data from 30 AML patients.
• Developed a computational approach incorporating multi-omic prior information to identify reliable gene expression markers.
• Validated marker identification and drug sensitivity prediction using independent datasets.
• Investigated the role of identified markers in drug response using cell line models.

Main Results:

• The developed computational method significantly outperformed existing state-of-the-art approaches.
• Identified molecular markers that were consistently replicated in validation data.
• Achieved accurate prediction of drug sensitivity based on molecular profiles.
• Discovered SMARCA4 as a marker and driver of sensitivity to topoisomerase II inhibitors (mitoxantrone, etoposide) in AML.

Conclusions:

• The novel computational method offers a promising strategy for identifying reliable molecular markers in AML.
• SMARCA4 expression is a predictive biomarker for response to specific topoisomerase II inhibitors in AML.
• This approach can advance targeted therapy development and personalized treatment strategies for AML patients.