Transcriptomic Features Associated With Kinase Inhibitor Sensitivity in KMT2A-Rearranged AML

KMT2A-rearranged (KMT2A-r) acute myeloid leukemia (AML) has a transcriptional program, but its relationship to drug sensitivity remains incompletely characterized. This study investigated whether a KMT2A-r-like transcriptomic state is associated with ex vivo kinase inhibitor sensitivity.

TARGET-AML RNA-sequencing data with fusion and cytogenetic annotations were used to define a predefined, biology-informed, continuous KMT2A-r identity score based on KMT2A-rearrangement-associated genes. The score was applied to Beat AML primary samples with matched normalized expression, ex vivo drug-response area under the curve (AUC) values, malignant cell-type scores, and FLT3-internal tandem duplication status. Associations were assessed using Spearman correlation and covariate-adjusted linear regression.

KMT2A-r AML exhibited increased expression of HOXA genes, MEIS1, PBX3, MEF2C, and FLT3 in TARGET-AML. The identity score differentiated KMT2A-r AML from other fusion groups in TARGET-AML; when applied to Beat AML, higher KMT2A-r identity scores were associated with lower AUC values for kinase inhibitors. After adjustment for malignant cell-type scores and FLT3-internal tandem duplication status, cabozantinib, sorafenib, and pelitinib retained significant negative associations with AUC. Further adjustment for FLT3 expression attenuated these associations, whereas an FLT3-excluded score preserved their direction.

A KMT2A-r-like transcriptomic state is associated with candidate kinase inhibitor sensitivity patterns in primary AML datasets and appears linked to, but not fully explained by, FLT3-related biology.