Molecular Mechanisms of Primary Resistance to Azacitidine in MDS/AML Patients - Data of the Hellenic MDS Study Group

Introduction: Azacitidine (AZA) is a hypomethylating agent that at low doses acts by inhibiting DNA methyltranferase activity. AZA is approved and widely used for the treatment of MDS patients and patients with AML not candidate for intensive chemotherapy. Unfortunately, even after an initial response, almost all patients relapse and so far -with the exception of a few clinical parameters and genetic mutations weakly correlated with favorable AZA response- the exact mechanisms underlying primary AZA resistance remain largely unknown. On the other hand, over the last years accumulated data suggest that hypoxia, an important regulatory factor of both, physiological and malignant, hematopoiesis, is also involved in MDS pathogenesis (Hayashi et al., 2018), while high Hif-1α levels in MDS have been previously correlated with poor overall survival and disease progression (Tong et al., 2012). Moreover, our group recently investigated the association between Hif-1α and response to AZA therapy and found that AZA-responders present with higher Hif-1α mRNA expression compared to non-responders/stable disease patients, while logistic regression analysis showed that Hif-1α mRNA expression is an independent predictor of response to AZA therapy (unpublished data).

Aims: The current study focused on investigating the mechanisms underlying the observed association of Hif-1α over-expression with response to AZA-therapy, by examining the methyltransferase activity and mitochondrial dysfunction due to inactivation of complex II, which is reported to lead to increased Hif-1α expression.

Methods: A total of 54 patients with a median age of 76 (52-89) years, and 10, age matched, healthy donors participated in the study. According to WHO 2016, 41 patients were classified as MDS (10 as MDS-EB-1, 24 as MDS-EB-2 and 7 as MDS-MLD) and 13 as AML. All patients received AZA treatment at the dose of 75mg/m² x7 days SC. BM-derived mononuclear cells were isolated before treatment using the Ficoll-paque method, followed by RNA extraction using TRIzol reagent, and cDNA preparation using Superscript II reverse transcriptase. Hypoxia-inducible factor 1-alpha (Hif-1α), succinate dehydrogenase complex subunit D (DSHd) and DNA methyltrasferase beta (DNMT3b) expression were estimated by real time PCR TaqMan gene expression assays, using the appropriate primers and probes. Relative gene expression was calculated by comparative threshold cycle (2^-ΔΔCt) method and normalized based on β-actin expression. Non-parametric tests were used for the statistical analysis of the results.

Results: Out of the 54 examined patients, 28 responded to azacitidine treatment (R), (including CR, PR and HI), 9 failed to respond (NR), and 17 achieved stable disease status 9 (SD). NR and SD patients were considered as one group (NR/SD) in all analyses. Using Rt-PCR we found that the 2^-ΔΔCt ratio of Hif-1α/β-Actin median expression for control samples was 1.18 (95% CI: 0.617-1.687), for AZA-responders 1.59 (95% CI: 1.029-3.18), while for NR/SD patients 0.754 (95% CI: 0.640-0.840), with a statistical significance between R and NR/SD patients (Mann-Whitney test, p=0.003). Moreover, the 2^-ΔΔCt ratio of SDHd/β-Actin median expression for control samples was 1.2 (95% CI: 0.360-1.954), for R patients 0.81 (95% CI: 0.294-1.401), and for NR/SD patients 0.73 (95% CI: 0.542-0.793). Finally, for DNMT3b, the 2^-ΔΔCt median expression ratio in control samples was 0.75 (95% CI: 0.637-1.526), for R patients 2.188 (95% CI: 1.547-3.630), while for NR/SD patients 1.338 (95% CI: 0.824-2.250).

Conclusions: Our data suggest that both AZA-R and NR/SD patients present with low levels of SDHd mRNA, compared to control, in line with previous reports in MDS. For AZA-responders, this could be related to the observed Hif-1α mRNA over-expression, since the SDH inactivation (decreased Complex II activity) is known to cause HIF stabilization (Frezza et al., 2011; Selak et al., 2005). Nevertheless, NR/SD patients also appear with decreased SDHd activity, despite the observed low Hif-1α expression. Therefore, in those patients, Hif-1α- related AZA-therapy response seems to be independent from mitochondrial dysfunction and possibly relies on other hypoxia regulatory mechanisms. Moreover, our data suggest that AZA-responders appear with an increased DNMT3b expression compared to both control and NR/SD patients, which could also explain their better response to therapy.