Network-based analysis with primary cells reveals drug response landscape of acute myeloid leukemia

Abstract

Acute myeloid leukemia (AML) is one of the most common, complex, and heterogeneous hematological malignancies in adults. Despite progresses in understanding the pathology of AML, the 5-year survival rates still remain low compared with CML, CLL, etc. The relationship between genomic features and drug responses is critical for precision medication. Herein, we depicted a picture for response of 145 drugs against 33 primary cell samples derived from AML patients with full spectrum of genomic features assessed by whole exon sequencing and RNA sequencing. In general, most of the samples were much more sensitive to the combinatorial chemotherapy regimens than the single chemotherapy drugs. Overall, these samples were moderately sensitive to the Traditional Chinese Medicine (TCM) and the targeted drugs. In the weighted gene coexpression network analysis (WGCNA), the TCM and targeted therapies displayed similar genetic signatures in the gene module correlation. Meanwhile, the expression of miRNAs, lncRNAs, and mRNAs did not display apparent gene module correlations among those different types of therapies. In addition, the combinatorial chemotherapy bear more module correlations than the single drugs. Interestingly, we found that the gene mutations and drug response were not enriched in any WGCNA module analysis. Most of the sensitive drug response biomarkers were enriched in the ribosome, endocytosis, cell cycle, and p53 associated signaling pathways. This study showed that gene expression modules might show better correlation than gene mutations for drug efficacy predictions.