miRNA Predictors of Response to DNA Methyltransferase Inhibitors

Background

Myelodysplastic syndrome (MDS) is a common disease of the elderly characterized by ineffective maturation of hematopoietic cells manifesting as low blood counts, morphologic atypia, and predisposition to the development of acute leukemia. The only FDA-approved drugs for the treatment of all subytpes of MDS are the DNA methyltransferase inhibitors (DNMTis), azacytidine and decitabine. However, only a minority of patients achieve a hematologic response or better on these agents. We have previously identified a miRNA signature which distinguishes MDS patients from normal controls. We sought to investigate whether miRNA expression might also be predictive of how patients respond to therapy with DNA methyltransferase inhibitors (DNMTis).

Design

We collected paired fresh frozen bone marrow mononuclear cells (BM-MNCs) from 28 patients with MDS both pre and post treatment with DNMTis as well as from 30 normal controls. Following enrichment for miRNAs and other small RNA species using the TruSeq Small RNA sample preparation kit (Illumina, San Diego, CA), miRNAs were sequenced on an Illumina HiSeq 2500. Alignment was performed using Bowtie 2 against reference sequences from mirBASE and differential expression analysis was analyzed using a rank sum score from the following analysis programs: DESeq, edgeR and baySeq. The response to the DNMTis was characterized based upon the following scale: 1 = complete remission, 2 = marrow complete remission, 3 = partial remission, 4 = hematologic improvement, 5 = stable disease, and 6 = progressive disease. Linear regression analysis was conducted using the response score as outcome and miRNA expression value as the predictor.

Results

Several differences were found between the control samples and the pre-treatment MDS samples, many of which have been described previously, including hsa-miR-125b, hsa-miR-342, hsa-miR-140, and hsa-miR-150. Not unexpectedly, the expression of numerous miRNAs was significantly altered as a result of treatment with DNMTis. However, there were no significant miRNA expression differences between responders and non-responders in the post-treatment samples and no significant differences between the changes in miRNA expression (pre versus post treatment changes) in responders and non-responders. Interestingly, 5 miRNAs demonstrated significant association between their pre-treatment expression and the degree of response (FDR < 0.05): hsa-miR-125a, hsa-miR-342, hsa-miR-146b, hsa-miR-146a, hsa-miR-423, and hsa-miR-150. These miRNAs share striking overlap in their mRNA targets including key pathway regulators in the p53 pathway as well as numerous growth factors implicated in hematopoietic stem cells as well as hematopoietic stem cell markers.

Conclusion

These studies confirm that miRNA expression differences differentiate between MDS and normal controls. The administration of DNMTis results in widespread expression changes regardless of the response to therapy. However, the expression of a small handful of miRNA are positively correlated with the degree of response in MDS pre-treatment samples to DNMTis and therefore may be useful as markers which predict responders to DNMTi therapy in addition to suggesting biological pathways implicated in the mechanism of response. These results await further prospective studies for confirmation.