The two hypomethylating agents DAC and azacitidine have received FDA approval for the treatment of MDS. However, it remains unclear why some patients are resistant to treatment. Our results show that primary resistance to DAC could be related to a higher ratio of CDA/DCK in a subset of patients, which means DAC is less activated through mono-phosphorylation by DCK and more inactivated through deamination by CDA in non-responders. Secondary resistance is likely due alternate progression pathways as we found less aberrant DNA methylation than at diagnosis, and there were no significant changes in DAC metabolism gene expression.
Mechanisms of in-vivo resistance to nucleoside analogues are complex and remain unresolved. One possibility might result from insufficient intracellular triphosphate, which has been tested for a number of drugs such as cytarabine, fludarabine, and 2-CdA in different trials 
. However, it remains experimentally very difficult to test this for DAC because clinical treatment is at low doses and its incorporation is at very low levels. Unlike the traditional cytotoxic therapies that induce rapid responses in MDS (mostly after one cycle), DAC has a different pattern of responses, which are rare after one cycle and improve over time 
. In humans, DAC has a short half-life (minutes) due to rapid inactivation in the liver by cytidine deaminase 
. Therefore, an alternate way to study DAC incorporation/activation is to measure gene expression related to its metabolic pathways as in our previous study in-vitro in cancer cell lines 
. Here, we found that the CDA/DCK ratio was statistically higher in non-responders than responders. These data favor a pharmacological mechanism of primary resistance for a subset of patients. The data on DCK are particularly relevant clinically given that azacitidine uses a different enzyme for initial mono-phosphorylation; thus, some patients with primary resistance to DAC could benefit from a therapeutic trial with azacitidine. However, multiple mechanisms must be active in different patients as we also found low CDA/DCK levels in some patients with primary resistance, that might not be able to overcome downstream pathways to resistance to DAC such as aberrant chromosome changes or defective induction of apoptosis, and others.
Secondary resistance to hypomethylating agents is emerging as a serious clinical problem. Survival at relapse after an initial response is poor. Here, we investigated secondary resistance using paired diagnosis/relapse samples and find that it is unlikely to be due to pharmacological mechanisms. We previously found that in-vitro acquired resistance to DAC in an HL60 cell line was due to DCK gene mutations 
, which also give rise to resistance to other NAs in other cell lines 
. However, DCK mutations were not detected in patients after relapse. Similarly, DCK mutations were rare in clinical resistance to other NAs 
. Although we found that the CDA/DCK ratio was higher in primary resistance to DAC, there was no significant difference in expression of these or other relevant genes between diagnosis and relapse in this study. The role of gene expression related to metabolic pathways in secondary resistance to NAs remains controversial. Some have observed a significant correlation between these gene expression or protein expression and clinical outcome to NA with relapsed and/or refractory leukemia. Conversely, other authors did not find this kind of relationship 
Another line of evidence against a pharmacologic mechanism for secondary resistance is the absence of hypermethylation at relapse. In fact, we observed that patients had significant hypomethylation at relapse compared to diagnosis, which cannot be explained by differential blast counts or other obvious confounders. Previously, we found that hypermethylation is accentuated in AML after relapse 
when patients received traditional chemotherapy containing cytarabine combinations. Thus, it is likely that hypomethylation induction by DAC does not recover in the face of continuing treatment, and that hypomethylation does not prevent patients' relapse and progression. Indeed, one cannot exclude the possibility that hypomethylation itself might eventually lead to progression and resistance to DAC either through ectopic gene reactivation or by mutagenesis and induction of chromosomal instability. Moreover, clinical responses to hypomethylating drugs in-vivo are complex and may involve differentiation and immune activation components. The bone marrow microenvironment is also an important factor to modulate response to chemotherapy 
. Thus, secondary resistance to DAC may also arise by complex mechanisms not entirely related to initial drug disposition.
Cytogenetic analysis showed that MDS patients after relapse showed evolution in 20% patients with abnormalities such +8, deletion of 16q, and −7. Cytogenetic evolution in MDS has been associated with progression to AML, and the new abnormalities we observed are already recognized as accompanying patients with poor prognosis, especially those involving loss or rearrangements of chromosome 7 and gain of chromosome 8. There are two broad critical regions of deletion on the long arm of chromosome 7 at bands 7q22 and 7q34-q36, which may contain important tumor suppressor genes that could be related to prognosis and resistance to DAC. This issue should be studied further using high resolution chromosomal analysis (for example by SNP-arrays) and/or genome sequencing to identify novel mutations in this setting. Overall, our data suggest that evolution to a more aggressive clone that is perhaps less dependent on DNA hypermethylation for survival may be a common mechanism of secondary resistance to decitabine.
In conclusion, we found that a high CDA/DCK ratio may be a marker of primary resistance to DAC in a subset of patients. If confirmed in other studies, this may help predict response to DAC treatment based on the value of CDA/DCK, or may steer patients towards azacitidine therapy, which does not depend on DCK for activity. By contrast, secondary resistance to DAC is likely independent of DNA methylation and pharmacologic pathways. It is more likely that genetic activation of oncogenic survival and progression pathways contribute to secondary resistance to DAC.