The cytosine analogue decitabine can induce both apoptosis and epigenetic/differentiation effects. Although the regimen commonly used to treat myelodysplastic syndrome has de-escalated doses with an epigenetic mechanism of action in mind, therapy continues to resemble pulse-cycled therapy for apoptosis objectives. This contrasts with the lower dose and one to three times per week schedule of decitabine used for non-cytotoxic epigenetic-differentiation therapy of non-malignant disease. Non-cytotoxic differentiation therapy could have substantial advantages, such as sparing of normal hematopoietic stem cells (HSC), decreased therapy related cytopenia that enables more frequent treatment exposure (a critical consideration with S-phase specific therapy), and a p53-independent mechanism of action. These possibilities were assessed in vitro and in vivo. Concentrations of decitabine that deplete DNMT1 in normal HSC without causing measurable DNA damage or apoptosis were determined. Treatment with equimolar AraC was used as a control. These concentrations of decitabine and AraC (0.5 μM) were used to treat p53 wild-type AML cells produced by retroviral insertion of MLL-AF9 into human CD34+ cells. Unlike AraC, decitabine did not induce apoptosis, but nonetheless terminated AML cell proliferation, accompanied by morphologic changes of differentiation, increased CD14 expression, and late and substantial upregulation of key proteins associated with myeloid cell cycle exit by differentiation, CEBPe and CDKN1B/p27. Decitabine produced an identical effect in p53 null MLL-AF9 leukemia cells (THP1 cells). In contrast, the p53 null cells did not demonstrate apoptosis, differentiation or proliferation inhibition in response to AraC. To determine if the non-cytotoxic differentiation terminated the self-renewal ability of leukemia initiating cells, p53 wild-type MLL-AF9 cells and normal HSC were treated with the identical regimen of decitabine or PBS in vitro then assayed for engraftment ability in NOD/SCID mice. Mice receiving the combination of mock treated normal and mock treated MLL-AF9 cells died of overwhelming leukemia by week 6. Mice receiving the combination of decitabine-treated normal and decitabine-treated MLL-AF9 cells remained healthy and after greater than twice the period of survival of the control group, were documented to have normal human hematopoietic cell engraftment, comparable to that seen in mice receiving normal human CD34+ cells without leukemia cells. To confirm that 0.2 mg/kg of decitabine administered sub-cutaneously on a weekly basis depletes DNMT1 without causing cytotoxicity or severe cytopenia in vivo, NSG mice were treated for 8 weeks. There was no treatment associated cytopenia or bone marrow cell apoptosis although DNMT1 was substantially depleted in bone marrow cells. This decitabine regimen, conventional AraC or vehicle was then used to treat xeno-transplant models of p53 wild-type and p53 null human AML (n=5 per group). In the p53 wild-type model, decitabine treatment was associated with significantly longer median survival than vehicle (>50% increase in survival, median survival 92 versus 61 days, Log-Rank p=0.0188), with one decitabine treated mouse without evidence of disease when the experiment was terminated on day 150. In the p53-null model, decitabine treatment was associated with significantly longer median survival (>20% increase) than AraC and vehicle treated mice (median survival 51, 45, and 42 days respectively, Log-Rank p=0.0004). To complement the above experiment in which AML cell lines were used, a xenotransplant model was established using fresh AML cells from a patient with relapsed treatment refractory AML. These AML cells contained complex chromosome abnormalities. Mice treated with decitabine (n=7) had significantly longer median survival (>100% increase) than AraC or vehicle treated mice (median survival 113, 56, and 50 days respectively, Log-Rank p<0.0001). These observations provide the foundation for AML therapy that is mechanistically distinct and a true alternative to conventional apoptosis-based therapy. This approach to therapy was non-toxic and highly effective in the pre-clinical in vivo models of human AML, as expected from its non-apoptosis based, p53-independent, and normal HSC sparing mechanism of action, and warrants further pre-clinical and clinical study.
No relevant conflicts of interest to declare.
Asterisk with author names denotes non-ASH members.