LAUSR

Introduction: Acute Myeloid Leukemia (AML) is a heterogenous disease characterized by distinct clinical course and prognosis based on genomics, epigenomics or transcriptomics profile. However, the multi-omics profiles have not been systematically integrated to define the integrative molecular subtypes of AML. There is a great clinical interest to identify the patterns across multi-omics profiles that could be used for prognosis and guiding targeted therapy in AML. Methods: To identify integrative molecular subtypes (iSubtypes) of AML, we performed an integrative clustering (iCluster) analysis of the AML multi-omics data (n=160) generated by The Cancer Genome Atlas (TCGA), which consisted of somatic mutation, DNA copy number, DNA methylation and RNA-seq gene expression data for 160 AML samples. We identified the multi-omics signatures that drove the molecular classification of AML. Based on the methylation and gene expression signatures, we derived a gene panel for classification of AML and verified the prognostic power of the gene panel using three independent AML datasets (n=1382). Results: We identified four iSubtypes of AML that featured distinct multi-omics signatures. At the DNA level, the iSubtype 1 was characterized by chromosomal abnormalities and high-frequency mutation of TP53 (30%) and RUNX1 (27%); the iSubtype 2 was characterized by chromosomal abnormalities and high-frequency mutation of CEBPA (20%) and FLT3 (26%); while the iSubtypes 3 and 4 were characterized by chromosomal normality and high-frequency mutation of FLT3 (34%, 41%), NPM1 (37%, 57%) and DNMT3A (20%, 41%). At the epigenomics level, the iSubtypes 1, 3 and 4 were characterized by hypomethylation of subtype driver genes, while the iSubtypes 2 was characterized by hypermethylation of subtype driver genes. At the transcriptomics level, the 4 iSubtypes were distinguished by differentially expressed genes involved in immune process, regulation of immune process, angiogenesis, leukocyte cell activation and migration, wound healing and cell structure organization, etc. The iSubtypes 1 and 4 had the worse overall survival (OS), while the iSubtype 3 had the best and the iSubtype 2 had the middle OS. Based on the subtype-driving methylation and transcription signatures, we derived a 571-gene panel for classification of AML. Using 3 independent AML transcriptomics datasets, we demonstrated that the excellent power of the gene panel in classifying AML into 4 gene expression subtypes with similar survival outcomes. Conclusions: iSubtypes of AML are jointly determined by multi-omics profiles. The subtype-driving methylation and transcription signature have excellent power in classifying AML into clinically relevant subtypes. This integrative analysis provides insights into the biological processes underling AML. Citation Format: Qianxing (Quincy) Mo, Seongseok Yun, David A. Sallman, Nicole Vincelette, Guang Peng, Ling Zhang, Jeffrey E. Lancet, Eric Padron. Integrative molecular subtypes of acute myeloid leukemia. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 4307.