Genetic Landscape and Clonal Evolution Following 5-Aza Therapy in Patients with High-Risk Myelodysplastic Syndromes

Background:

DNA hypomethylating agents, such as 5-azacitidine (5-aza) and decitabine, comprise the current standard in therapy for patients with high-risk myelodysplastic syndromes (MDS), with dramatic responses in some patients. However, the responses are poorly predictable and their impact on clonal dynamics has not been fully elucidated.

Patients and Methods:

We enrolled a total of 119 patients with high-risk MDS who were treated with 5-aza . Bone marrow samples were collected before (n = 71) and both before and after (n = 48) treatment and analyzed by targeted-capture sequencing using RNA baits designed for 67 known or putative driver genes in myeloid neoplasms and 1,674 single nucleotide polymorphisms, which enabled detection of both mutations and copy number alterations on the same platform. In 9 of the 48 patients, pre- and post-therapy samples were further analyzed by whole exome sequencing (WES).

Results:

Average number of driver mutations before 5-aza was 2.8 per patient and 107 (90%) patients had multiple mutations. Most frequently mutated were TP53 (27%), followed by RUNX1, TET2, DNMT3A, and ASXL1. Reflecting high-risk disease subtypes of the subjects, splicing factor mutations were relatively rare (29 %) in the current cohort. Chromosomal abnormalities were identified in 65 (55%) patients, where 7q- and /or 5q- were the most frequent. Among 48 patients with serially collected samples, 46 had one or more mutations, enabling an evaluation of clone dynamics. In total 163 and 146 mutations were detected before and after treatment, respectively. About two thirds (110/163) of the mutations before 5-aza remained detectable after treatment. By contrast, the remaining one third showed a dynamic clonal behavior; 36 mutations in 22 cases were newly acquired, whereas 53 in 28 cases disappeared. Among those newly acquired, most frequently observed were mutations in STAG2 and EP300 (n = 3), of which STAG2 (7 cases) also represented the most frequent targets of disappeared mutations after treatment. In WES in 9 patients, a total of 112 mutations were identified either before or after 5-aza treatment with a mean of 10.4 or 8.9 mutations per sample, respectively. Among these, 63 were found at both pre- and post-therapy samples, whereas 17 and 32 mutations were newly acquired or disappeared during treatment, Given that only 4 newly acquired and 8 lost mutations had been detected by targeted-capture sequencing, respectively, WES enabled more sensitive detection of alternation of clones during 5-aza treatment, which were demonstrated in 8 (89%) subjects, rather than 5 (56%) in targeted-capture sequencing.

Clinical outcomes have been reported for 22 patients as of the time of abstract submission; 5 achieved complete remission (CR), 9 stable disease (SD), and 5 progressive disease (PD). Alteration in clone size was frequently associated with clinical response. The size of dominant clones significantly decreased in 4 of 5 cases with CR, whereas stable or increased in 12 of 14 patients with SD or PD. In patients with SD or PD, acquisition of new mutations was common (10/14) during 5-aza treatment and potentially implicated in the resistance to 5-aza-treatment. Of interest, newly acquired mutations were also found in 2 CR samples, albeit at low allele frequency, even though the clone size of dominant clones was substantially reduced, suggesting the evolution of alternative MDS subclones or expansion of preexisting non-leukemic hematopoietic clone. Although CR was achieved in 3 of 6 patients with TP53 mutations, the TP53-mutationsdid not totally disappeared but were still detectable in CR samples in 2 cases, suggesting that TP53 mutated clones have not been completely eradicated by 5-aza treatment.

Conclusion:

Our study successfully depicted the structure of clones and their dynamics in high-risk MDS on 5-aza treatment. Alteration in the size of the dominant clones was frequently associated with a clinical response. Clonal evolution was common even in patients who achieved CR. Tracking the mutations in MDS patients during 5-aza treatment provides the opportunity to detect clones resistant to 5-aza and might be used to guide 5-aza therapy.