Molecular profiling of pre-and post-5-azacytidine myelodysplastic syndrome samples identifies predictors of response

Treatment with the hypomethylating agent 5-azacytidine (AZA) increases survival in high-risk (HR) myelodysplastic syndrome (MDS) patients, but predicting patient response and overall survival remains challenging. To address these issues, we analyzed mutational and transcriptional profiles in CD34+ hematopoietic stem/progenitor cells (HSPCs) before and following AZA therapy in MDS patients. AZA treatment led to a greater reduction in the mutational burden in both blast and hematological responders than non-responders. Blast and hematological responders showed transcriptional evidence of pre-treatment enrichment for pathways such as oxidative phosphorylation, MYC targets, and mTORC1 signaling. While blast non-response was associated with TNFa signaling and leukemia stem cell signature, hematological non-response was associated with cell-cycle related pathways. AZA induced similar transcriptional responses in MDS patients regardless of response type. Comparison of blast responders and non-responders to normal controls, allowed us to generate a transcriptional classifier that could predict AZA response and survival. This classifier outperformed a previously developed gene signature in a second MDS patient cohort, but signatures of hematological responses were unable to predict survival. Overall, these studies characterize the molecular consequences of AZA treatment in MDS HSPCs and identify a potential tool for predicting AZA therapy responses and overall survival prior to initiation of therapy. Overall design: To investigate whether transcriptional profiles from paired pre-/post AZA MDS samples predict treatment responses to AZA, we generated RNA-seq data from 14 MDS patients before and after treatment with AZA. We also generated RNA-seq data from 6 age-matched healthy donors as control. Bone marrow mononuclear cells (BMMCs) were isolated from BM samples using Ficoll-Paque Plus (GE Healthcare, Massachusetts, US) density gradients per manufacturer's instructions. CD34+ cells were isolated by immunomagnetic separation using an autoMACs Pro Separator (Miltenyi Biotec GmbH, Bergisch Gladbach, Germany) following the manufacturer´s instructions, which were used for library preparation and sequencing.