Decreased Expression of Anti-DNMT1 Tumor-Suppressor microRNAs in Azacitidine (AZA)-Resistant Cells Independently Predicts Survival in Patients Treated with AZA for Higher Risk Myelodysplastic Syndrome (HRMDS) and Oligoblastic Acute Myeloid Leukemia (AML)

Background: AZA is the standard of care for patients with either HRMDS or AML with 20-30% blasts (Oligoblastic AML). The response rate is ~50% with a median response duration of 12-15 months. The mechanisms underlying primary and secondary resistance are poorly understood and it remains difficult to accurately predict which patients will respond and in responders, to predict secondary resistance. The main factors demonstrated or suggested to interfere with response to AZA and/or survival include marrow blast percentage, performance status, IPSS cytogenetic risk, presence of peripheral blasts, transfusion dependency, IPSS-R risk, mutations of TET2, P53, IDH1/2, and DNMT3A, elevated expression of BCL2L10, Fas, or PI-PLCbeta1. Deregulation of microRNAs is a hallmark of MDS, yet its consequences on AZA efficacy have not been specifically addressed in HRMDS.

Methods: We measured the expression of 754 miRNAs in SKM1 MDS cells resistant or sensitive to AZA. miRNAs of interest were ectopically expressed or specifically inhibited in HEK293T cells which were assayed for AZA sensitivity (MTT). Seven miRNAs were quantified in bone marrow mononuclear cells deriving from 75 patients with HRMDS (n= 50) or oligoblastic AML treated with AZA [median age 74, 24 females, median cycle number 6, (1-39)].

Results: Seven miRNAs (miR-125a-5p, miR-99b-5p, miR-126, miR-126*, hsa-let-7c, miR-34b-3p, and miR-10b*) that include 6 tumor-suppressor miRNAs, were found differentially expressed between AZA-sensitive versus -resistant cells; all being transcriptionally repressed in resistant cells. In silico, 5 of these miRNAs were found to target the 3' UTR of DNA methyltransferase 1 (DNMT1) while Zhao et al. [Arthritis & rheumatism, 2011; 63(5)] have shown that miR-126 directly inhibits DNMT1 translation. DNMT1 is one of the main AZA molecular target, the higher the level of DNMT1 expression, the lower the AZA sensitivity [Li et al., Cancer biology & therapy, 2010; 9(4)]. Microarray and western blotting showed that levels of DNMT1 protein, but not mRNA, were significantly higher in AZA-resistant cells. In HEK293T cells, specific endogenous miR-126/126* inhibition significantly augmented DNMT1 protein expression and triggered AZA-resistance, as measured through MTT proliferation assay. In the same cells, ectopic expression of miR-126/126* decreased DNMT1 protein expression in a concentration-dependent manner. In the 75 patients treated with AZA, a decreased expression level of all but miR-10* was associated with decreased response rate (IWG 2006 criteria) and poor outcome; miR-126/126* having the strongest prognostic impact. When compared with miR-126*^High MDS, miR-126*^Low MDS had significantly lower overall response rate (37% versus 60%, p=0.04), higher relapse rate (100% versus 71%, p=0.034), shorter progression-free (PFS) (8±8% versus 49±12% at 3 years, p=0.004, log rank test), and overall survival (OS) (16±6% versus 46±9%, p=0.004). Baseline patient characteristics of the 2 groups were similar. Multivariate analyses were performed using the Cox proportional hazards model. Table 1 shows that age, miR-126* expression, and IPSS-R risk independently predicted PFS and OS. miRNAs expression was analyzed over time in 15 patients. The mean expression level of 5/7 miRNAs increased over time in the 10 responders without statistically significant difference (DNS) between the first and latest values. In contrast, the expression of 7/7 miRNAs decreased over time in the 5 patients with treatment failure, the difference being statistically significant for 2 miRNAs (p²0.043, Wilcoxon test). Secondary resistance occurred in 7/10 responders and was found associated with a decreased expression of 6/7 miRNAs (p²0.044 for 4 miRNAs) versus 2/7 (DNS) in the 3 long-term responders.

Conclusion: Our results suggest that in HRMDS and oligoblastic AML, i. primary or secondary resistance to AZA is associated with the decreased expression of anti-DNMT1 tumor-suppressor microRNAs that trigger AZA resistance in vitro; ii. measuring miRNAs expression before and under treatment might help to predict primary or secondary AZA resistance and thereby to rapidly offer alternative treatment; iii. increasing AZA exposure might help to circumvent AZA resistance in case of either low or decreased anti-DNMT1 miRNA concentration while using anti-DNMT1 microRNAs as a therapeutic tool might increase or restore AZA sensitivity.