Myelodysplastic syndromes (MDS) are a group of heterogeneous hematopoietic malignancies characterized by the defective production of mature blood cells and the risk of progression to myeloid leukemia. Hypomethylating agents (HMA) are currently the first-line therapy for high-risk MDS and induce hematological improvement in 50% of the patients. However, complete responses occur in less than 20% of these patients, and a majority of MDS patients treated with HMA will eventually relapse, with a subsequently fatal prognosis. Although the presence of mutations in certain genes, such as Tet2, has been reported to predict response to HMA, the causes of resistance and relapse have yet to be described.

MDS research has been traditionally a slow-moving field owing to the lack of well-established MDS cell lines and animal models. This landscape recently changed with the development of transgenic mouse models that mirror the phenotype of high-risk MDS patients. In mice from the 4th/5th generations (G4/G5) of our group's telomere-dysfunctional TERTER/ER mouse model, the DNA-damage response triggered by telomere erosion causes profound cell-intrinsic abnormalities in hematopoietic stem and progenitor cells (HSPC) that lead to myeloid-skewed hematopoiesis. These mice exhibit peripheral blood (PB) cytopenias, age-related granulocytosis and myeloid infiltration of the BM and the spleen (Colla et al. Cancer Cell 2015). Similarly, the conditional deletion of Tet2 in mouse hematopoietic cells generates a phenotype of HSC expansion and myeloid skewing of hematopoiesis. These defects, which are also cell-intrinsic, induce progressive PB neutrophilia and granulocytosis, with myeloid infiltration of the bone marrow (BM) and spleen and severe splenomegaly (Moran-Crusio et al. Cancer Cell 2011).

Using the TERTER/ER and Tet2-/- mouse models as a platform for cellular and molecular studies, we have investigated the mechanism of action of HMA in MDS, with a focus on the characterization of response by the different HSPC populations and the mechanisms involved in resistance and relapse. Our results show that the administration of the HMA azacytidine (AZA) for 7 days induces in both mouse models a 20-40% drop in white blood cell counts and a 17-24% decrease in hemoglobin levels that mimic patient responses to AZA. After the last day of treatment and during the 3 last weeks of the cycle, cell counts and hemoglobin levels returned to normal values. These changes in PB were explained by an 80-86% decrease in the number of BM common myeloid progenitors (CMP; n = 8, P = 0.0046) and granulo-monocytic progenitors (GMP; n = 8, P <0.0001) after 1 week of AZA treatment, while megakaryocytic-erythroid progenitors remained largely unaffected. However, the number of CMP and GMP in the BM of both G5 and Tet2-/- mice recovered by the end of each cycle, and the characteristic myeloid skewing was not overcome after 2 cycles of treatment, which indicates that AZA cannot reverse the cell-intrinsic defects of these progenitors. Most importantly, BM analysis in both mouse models after one week of treatment and during the recovery period revealed that AZA has no effect on long and short-term HSC populations.

To study the effect of AZA in a BM context that mimicked the coexistence of MDS and normal cells in patients, we performed competitive transplantation experiments in both models. Our preliminary results indicate that, in mixed populations, AZA induces a 2.6-fold expansion of the wild type long-term HSC compartment (n = 4, P = 0.0018). The fact that AZA gives a repopulation advantage to normal HSC suggest that, after several cycles, there is a shift in the abundance of normal versus MDS cells in patients BM, and this explains why patients enter morphologic remission. However, in both transgenic mouse models, AZA failed to decrease the size of the MDS-like HSC compartment. This indicates that, although this drug allows a temporary hematologic recovery, it does not eliminate the cell populations that maintain the disease, and therefore it cannot eliminate MDS.

Ongoing unbiased transcriptomic and methylomic analyses in the sorted MDS-like HSC that survived AZA treatment will identify the HSC signature of resistance to HMA, and the integration of these results with similar analyses in MDS patients will provide valuable insights into the mechanistic bases of relapse and resistance to HMA and facilitate the design of better combination therapies for high-risk MDS.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.