Machine learning-based integration develops multi regulated cell death-related signatures for improving outcomes and immune response in acute myeloid leukemia

Abstract

Acute myeloid leukemia (AML) presents an unpredictable and complex prognosis. Researchers have been exploring different ways to determine the death of cancer cells, including various forms of regulated cell death such as apoptosis, autophagy, cuproptosis, disulfidptosis, ferroptosis, necroptosis, panoptosis, and pyrotosis. However, there have been limited studies on combining these cell death modalities in AML. This study aimed to identify clinically relevant molecular classification of AML based on genes related to multiple programmed cell death signaling pathways. Machine learning techniques were used to identify these subtypes and provide guidance on medication for each subtype using transcriptome data. The subtypes differed in immune and drug response. A robust prognostic model MPCDPG was developed based on gene pairs and validated in multiple test sets. In conclusion, this study identified three clinically relevant subtypes of AML and developed a reliable prognostic model based on gene pairs regulated cell death-related genes.