Stem cells play an important role in acute myeloid leukemia (AML). However, their precise effect on AML tumorigenesis and progression remains unclear.
The present study aimed to characterize stem cell-related gene expression and identify stemness biomarker genes in AML. We calculated the stemness index (mRNAsi) based on transcription data using the one-class logistic regression (OCLR) algorithm for patients in the training set. According to the mRNAsi score, we performed consensus clustering and identified two stemness subgroups. Eight stemness-related genes were identified as stemness biomarkers through gene selection by three machine learning methods.
We found that patients in stemness subgroup I had a poor prognosis and benefited from nilotinib, MK-2206 and axitinib treatment. In addition, the mutation profiles of these two stemness subgroups were different, which suggested that patients in different subgroups had different biological processes. There was a strong significant negative correlation between mRNAsi and the immune score (r= -0.43, p<0.001). Furthermore, we identified eight stemness-related genes that have potential to be biomarkers, including SLC43A2, CYBB, CFP, GRN, CST3, TIMP1, CFD and IGLL1. These genes, except IGLL1, had a negative correlation with mRNAsi. SLC43A2 is expected to be a potential stemness-related biomarker in AML.
Overall, we established a novel stemness classification using the mRNAsi score and eight stemness-related genes that may be biomarkers. Clinical decision-making should be guided by this new signature in prospective studies.