Transcriptomic analysis of non-leukemic cell subsets in azacytidine-responsive AML highlights pathways associated with adhesion, platelet aggregation, and angiogenesis in mice and humans.

Abstract

Background: Hypomethylating agents (HMAs), such as 5-azacytidine (AZA), are valuable treatment options for patients with acute myeloid leukemia (AML). Despite providing significant extensions in survival when used alone or in combination with BCL-2 inhibitors, resistance and eventual relapse is observed. Reported mechanisms of these outcomes are inconsistent when focusing on leukemic populations within bone marrow, indicating a need for studies on the impact of HMAs on non-leukemic cells in the blood and other tissue compartments.

Methods: Whole blood and spleens from vehicle- or AZA-treated mice implanted with the syngeneic AML line C1498 were transcriptionally profiled using a comprehensive panel of immune-related gene probes. Publicly available RNAseq data from blood of AZA-responsive, human AML patients were analyzed compared to matched, pre-treatment samples. Genes differentially expressed between vehicle- and AZA-treated (mouse) or pre- and post-AZA treatment (human) samples were analyzed for statistical overrepresentation in gene ontologies using Fisher's one-tailed t-test. Pathological analyses of various tissues in AML relapsed, AZA-responsive mice were compared to the corresponding tissues in vehicle-treated mice.

Results: We observed hematologic recovery in the peripheral blood of AZA-treated groups, versus vehicle control, that was associated with significant extensions in survival. Transcriptional analysis of AZA-treated samples revealed decreased cell type scores for suppressive subsets and increased pathway scores for T and B cell functions. Comparisons of gene ontology annotations enriched from genes differentially regulated by AZA in human and mouse blood samples revealed overlap in numerous biological pathways including adhesion, thrombosis, and angiogenesis. Consistently, C1498 permeated the liver at end-stage disease in vehicle-treated mice, while AZA treatment limited their spread to just outside the bone after relapse.

Conclusions: AZA-induced differences in C1498 spread could be a result of gene expression changes in adhesion, platelet aggregation and/or angiogenesis in non-leukemic compartments; however, further mechanistic studies must be done to confirm a direct link between modulated genes and disease manifestation. Overall, these studies provide rationale for expanding the exploration of biomarkers and therapeutic targets to include normal immune cells in blood, spleen, or other microenvironments of AML patients treated with HMA, rather than limiting studies to the bone marrow and leukemic blasts.