Analysis of Molecular Predictors of Response to 5-Azacitine Treatment in AML and MDS Patients Preemptively Treated for Molecular Relapse of Disease

Background: Relapse of disease remains the major cause of treatment failure in patients with acute myeloid leukemia (AML) or advanced myelodysplastic syndrome (MDS), even after allogeneic hematopoietic stem cell transplantation (HSCT). Treatment of relapsed AML or MDS is difficult, especially after HSCT, and long-term prognosis of patients suffering from relapse is dismal. One approach to overcome this problem is to use sensitive molecular diagnostic strategies to detect recurring disease already at the level of minimal residual disease (MRD), thus avoiding the development of overt hematologic relapse by treatment of patients at the stage of molecular relapse. We have recently implemented preemptive treatment with the demethylating drug 5-Azacitidine (AZA) in patients with molecular evidence of recurrent disease in a prospective Phase II study (RELAZA). In this study, 80% of the patients showed responses, with reduction of MRD and prolonged leukemia free survival, 20% of patients even showed molecular clearance of their leukemia and long-term disease free survival. More recently, results from several groups studying demethylating agents in MDS or AML suggested that patients with mutations in genes involved in epigenetic DNA-modification, such as TET2, DNMT3A or IDH1 or IDH2 might be more responsive to treatment with these drugs. Since we observed varying clinical response in the patients treated preemptively with AZA for molecular evidence of recurrent disease, we correlated the clinical response in these patients with the presence of mutations in epigenetic regulator genes in order to identify potential predictors of response.

Patients and Methods: A cohort of 44 patients (23 f/21 m), median age 55.6 years (range 21-75 years), in hematological remission with AML (N=40) or MDS (N=4) were given AZA to treat molecular relapse defined by mutant NPM1 (N=23) or CD34+ chimerism (N=21). Patients were monitored post allogeneic HSCT (N=26) or standard chemotherapy (N=18). The cohort received a median of 5 cycles of AZA (ranging from 1-18 cycles). DNA taken at first diagnosis was analyzed using amplicon based resequencing on a MiSeq next generation sequencing system for the following genes, either analyzing the complete coding region (EZH1, EZH2, DNMT3A, TET1 and TET2) or hot-spot regions (ASXL1, ASXL2, IDH1, IDH2). First diagnosis samples were unavailable for 4 patients. In these, DNA from sorted CD34+ cells taken at the time of molecular relapse was used as a substitute.

Results: Amplicon sequencing revealed mutations in one or more genes in 25/44 patients (56.8%). With 15 mutations (34%), DNMT3A was the most frequently mutated gene, the majority of the alterations (9; 60%) were located in exon 23. Mutations in TET2 were found in 8 patients, IDH1 was mutated twice, ASXL2, EZH2 and TET1 were mutated once each. In 20 of the 44 patients (45.5%), no mutations in the investigated genes were found. A comparison of primary response to AZA-treatment (defined as stabilization or decrease of the MRD-marker) between patients with and without mutations revealed no significant difference (79.2 vs 66.6%; P=.48). Likewise, the rate of hematologic relapse was comparable in both cohorts (54% vs. 56%). However, a more detailed look at the patients with mutations revealed differences. The highest initial response rate was observed in patients with DNMT3A mutations (87%), whereas patients with isolated TET2 mutations were less likely to respond (50%). Also, the rate of hematologic relapse was highest in patients with TET2-mutations (75%) compared to patients with DNMT3A-mutations alone (41.6%). In support of a role of TET2-mutations in mediating resistance, an analysis of matched diagnosis and relapse samples in three patients indicated persistence of TET2-loss of function mutations in one patient as well as an acquisition of a second mutant TET2- allele or a switch to a loss-of function-mutation in two patients, indicating that a clonal evolution favoring a subclone with an inactivating TET2-allele under treatment with AZA occurred.

Conclusions: Our data confirm that mutations in epigenetic regulator genes are common in patients with AML. Although based on small numbers, these preliminary data do not support that mutations in these genes are associated per se with an improved response to treatment with AZA, but might indicate a differential effect of certain alterations, i.e. DNMT3A-mutations or mutations of TET2.