Tracking the Somatic Mutations in Azacitidine-Treated MDS Patients Documents Clonal Development and AZA Responsiveness

Introduction and hypothesis: Somatic gene mutations develop in ~78% Myelodysplastic syndrome (MDS) patients. MDS progresses into an unstable phase characterized by an accumulation of myeloblasts that is indicated for the DNA demethylation therapy with Azacitidine (AZA). To understand whether AZA is capable to eliminate tumor cells and whether a mutation pattern responds to AZA we herein tracked mutations in the bone marrow (BM) during AZA treatment. Two scenarios were postulated: 1) AZA treatment will/ will not eliminate clones characterized by specific mutations and this will relate to the clinical outcome, 2) during clinical progression on the AZA treatment either the new mutations will develop or the original mutations will be detected.

Patients: 40 int-2/high-risk MDS patients (176 samples, median age 70, 22F/18M) indicated for AZA (75mg/m², 5+2+2) were sequenced. The MDS subtypes included RAEB1, RAEB2, and MDS/AML. Half of the patients progressed from the 5q-syndrome. Libraries were prepared using TruSight DNA amplicon kit. Set of 54 myeloid genes (associated with MDS or AML) were sequenced by Illumina platform HiSeq 2500 with depth >100 per mutation, mutation should be heterozygous, non-synonymous, exonic with frequency >10%. As controls: 4 normal BM samples and 2 cord bloods (also applied for data filtering) and MOLM-13 cell line carrying FLT3-ITD mutation were used. Additional controls included CD3+ T cells isolated from the patient BM samples.

Results: 70% of patients were informative for at least one somatic mutation with high impact on the amino acid sequence. Mutations in TET2 (in 13%) were overall the most frequent before AZA was started (followed by mutations in BCOR, RUNX1, STAG2, and NOTCH genes). In 36% of informative patients, the mutation pattern developed on AZA. Interestingly, the most frequent mutations after AZA were in ASXL1 in 10% (followed by mutations in TET2, BCOR, and CUX1 genes). While mutations in ASXL1 and NOTCH genes developped only in the non-5q derived patients, the mutations in RUNX1 developped only in the 5q-derived higher-risk MDS patients. Average number of mutations before AZA and on AZA was 2 per patient. Complete elimination of the mutation pattern was noted in 57% of informative patients during the first 8 months of AZA treatment and this was associated with the therapy responsiveness (PR or CR). In contrast in 7% of informative patients the mutation pattern remained the same and this was associated with the stable disease (leading to progression). Upon progression on the AZA treatment we have observed appearance of the new mutations in 73% of informative patients. 20% of patients progressing after 14 cycles of AZA were non-informative which suggests that these patients may carry mutations in genes not included within the tested set. We have assessed overall survival (OS) according to the mutation status in the following subgroups: non-informative patients, informative patients with mutations that were eliminated by AZA, and those retaining or gaining the mutations upon AZA. The group that developed or retained mutations during first 4-8 cycles of AZA displayed significantly lower OS (p=0.004, median OS:11 months) compared to either non-informative patients (26 months) or those where AZA completely eliminated the mutation pattern (28 months).

Conclusions: Tracking the mutations in MDS patients during AZA therapy provides opportunity to detect clonal development in 2/3 of the MDS patients and relate these data to the clinical outcome. Moreover, progression on AZA therapy is usually associated with the development of a new mutation pattern and this coincides with the significantly lower OS.

Grants: GACR P305/12/1033, UNCE 204021, PRVOUK-P24/LF1/1.