Comprehensive Mutational Screening Of 5-Azacitidne Treated Myelodysplastic Syndrome (MDS) Patients Fails To Identify a Specific Mutational Profile Predicting Response To Therapy

5-azacitdine (aza) treatment in myelodysplastic syndrome (MDS) induces a response rate of 40-45% and prognostic factors for response and survival remain still largely unknown. Presence of mutant TET2 has been shown to predict better response to aza and the recently described French Azacitidine prognostic score segregates patients into 3 groups with varying median overall survival. Although a plethora of somatic mutations have been described in MDS, none has been consistently shown to be prognostically important in the context of response to hypomethylating agent(s).

To identify the mutation signature associated with aza response, we undertook screening of 24 myeloid genes (splicing, epigenetic, transcription factors, STAG2, TP53) in 66 MDS patients treated with aza at our institution over a period of 2004-2012. Mutation analysis was done by deep (454 FLX) and Sanger sequencing. SNP-6 karyotyping was also performed in a subset to correlate with mutation status. Responses were assessed as per the international working group for MDS criteria.

The median age was 67 years (range 36–87 years), median number of courses 7(range 2–42), 79% of patients belonged to int-2/high risk IPSS category. WHO category subtypes were; RA/RARS-2; RCMD-9; RAEB-39; s-AML (evolved from pre-existing MDS) -5, therapy related myeloid neoplasm (t-MDS/t-AML) -8 and CMML-3.IPSS cytogenetic subgroups were, good risk: 22, intermediate: 7, and poor risk: 37. One fourth of patients had received prior therapy, with only two receiving low dose cytarabine. Median time from diagnosis to aza treatment was 8.6 months. The overall response rate (ORR) to Aza was 47% (31/66) with complete response (CR) 17%, partial response (PR) 11%, marrow CR (mCR) 12% and stable disease with hematological improvement (SD-HI) 7%.

Candidate mutations were seen in 82% (54/66) of patients, with the more than half harboring ≥2 mutations each. The most frequently (≥5 %) mutated genes being ASXL1 (29%), TP53 (23%), TET2 (14%), DNMT3A (12%), SRSF2 (12%), EZH2 (11%), NRAS (8%), U2AF1 (8%), IDH2 (8%), RUNX1 (8%), CCBL (6%) and FLT3-ITD (5%).

On univariate analysis presence of EZH2 mutations predicted for a better ORR compared to wild type EZH2 (21% vs. 3%, p<0.04). TET2 mutations did not confer a better response (13% vs. 14%, p=0.1). None of other mutations predicted response. In multivariate logistic regression analysis, only shorter duration of disease (< than 8 months) predicted a better response (p=0.33,HR 0.29,CI 0.07-0.9).

Patients with any mutation had a worse outcome compared to patients with no mutations (p=0.03, med OS 12.4 vs. 20.2 months). TP53 mutations were detected and quantified by high throughput sequencing in poor risk IPSS cytogenetic group (15/37) and were absent in patients with normal cytogenetics and isolated chromosome 7 abnormalities. Sequential samples were available in 9 patients and changes in clone sizes were correlative of treatment response, with 3 patients achieving cytogenetic response with concurrent reduction in clone size but the responses were transient. The TP53 mutation status correlated strongly with finding aberration of chromosome 17p by SNP array and positive p53 antibody staining on bone marrow trephine. On univariate analysis, TP53 mutations had an adverse impact on OS (P=0.0001, med OS 9.7 vs. 20.2 months).

Age, IPSS cytogenetic risk group, WHO subtypes, overall response rate (ORR),progression to AML, TP53 mutation status and presence of EZH2 ,ASXLI,TET2,DNMT3A, SRSF2,NRAS mutations were used as co-variables in multivariate analysis.TP53 mutation status (HR=7.9(3.9-17.8), p=0.0001) and DNMT3A mutations (HR=2.5 (1.03-6.2), p=0.04) were established as independent prognostic variables influencing OS. Additionally, ORR to aza treatment imparts a better survival advantage (HR=0.3 (0.16-0.57),p=0.0001)

This study shows that gene mutations are frequent in all cytogenetic subgroups of MDS. TET2 mutations did not predict response to aza contrary to published reports, which could be due to differences especially as low dose cytarabine was infrequently used in our cohort. TP53 mutations do not correlate with response rate but imparts a dismal prognosis. In conclusion, although we have screened comprehensively we do not identify a mutational profile predictive of response to azacitdine.

No relevant conflicts of interest to declare.