Different methylation signatures at diagnosis in patients with high-risk myelodysplastic syndromes and secondary acute myeloid leukemia predict azacitidine response and longer survival

Abstract

Background: Epigenetic therapy, using hypomethylating agents (HMA), is known to be effective in the treatment of high-risk myelodysplastic syndromes (MDS) and acute myeloid leukemia (AML) patients who are not suitable for intensive chemotherapy and/or allogeneic stem cell transplantation. However, response rates to HMA are low and there is an unmet need in finding prognostic and predictive biomarkers of treatment response and overall survival. We performed global methylation analysis of 75 patients with high-risk MDS and secondary AML who were included in CETLAM SMD-09 protocol, in which patients received HMA or intensive treatment according to age, comorbidities and cytogenetic.

Results: Unsupervised analysis of global methylation pattern at diagnosis did not allow patients to be differentiated according to the cytological subtype, cytogenetic groups, treatment response or patient outcome. However, after a supervised analysis we found a methylation signature defined by 200 probes, which allowed differentiating between patients responding and non-responding to azacitidine (AZA) treatment and a different methylation pattern also defined by 200 probes that allowed to differentiate patients according to their survival. On studying follow-up samples, we confirmed that AZA decreases global DNA methylation, but in our cohort the degree of methylation decrease did not correlate with the type of response. The methylation signature detected at diagnosis was not useful in treated samples to distinguish patients who were going to relapse or progress.

Conclusions: Our findings suggest that in a subset of specific CpGs, altered DNA methylation patterns at diagnosis may be useful as a biomarker for predicting AZA response and survival.

Keywords: Azacitidine; DNA methylation; Epigenetic drugs; Hypomethylating agents; Myelodysplastic syndromes; Prognostic factors; Secondary acute myeloid leukemia.

Fig. 1

DNA methylation profiles distinguish patients and healthy controls. a Principal component analysis shows a separation between patient and control samples after unsupervised analysis. We have analyzed a total of 25 bone marrow samples taken at diagnosis from patients with secondary acute myeloid leukemia (sAML) and 48 from patients with myelodysplastic syndrome (MDS). b Heatmap showing major methylation in neoplastic samples compared to healthy controls

Fig. 2

Methylation profile of 200 probes at diagnosis allow distinguishing responders and non-responders to azacitidine (AZA) treatment. Heatmap showing the hierarchical clusterization of the 200 probes obtained with combined rank analysis that are differentially methylated between responders and non-responders in bone marrow samples at diagnosis. Selected genes associated with differentially methylated CpGs are indicated

Fig. 3

Methylation profile of 200 probes at diagnosis predict overall survival in AZA treated patients. Heatmap showing the hierarchical clusterization of the 200 probes obtained with combined rank analysis that are differentially methylated between longer and shorter survivals in bone marrow samples at diagnosis. Selected genes associated with differentially methylated CpGs are indicated

Fig. 4

Methylation pattern at time of disease progression after initial response is not similar to methylation pattern from those patients that have never respond (non-responders). Heatmap showing the hierarchical clusterization of samples from patients in progression (n =12) in comparison with responder (n =21) and non-responder (n =16) patients at diagnosis based on DNA methylation of the sites identified in Fig. 2

Fig. 5

Differential methylation associated with initial response does not provide information on response duration or progression. Heatmap showing the hierarchical clusterization of samples of response after more than 6 AZA cycles (n =9) and samples of progression (n =7) using the methylation status of CpG sites identified in Fig. 2