DNA methylation predicts survival and response to therapy in patients with myelodysplastic syndromes

Abstract

Purpose: The current classification systems of myelodysplastic syndromes (MDS), including the International Prognostic Scoring System (IPSS), do not fully reflect the molecular heterogeneity of the disease. Molecular characterization may predict clinical outcome and help stratify patients for targeted therapies. Epigenetic therapy using decitabine, a DNA hypomethylating agent, is clinically effective for the treatment of MDS. Therefore, we investigated the association between DNA methylation and clinical outcome in MDS.

Patients and methods: We screened 24 patients with MDS for promoter CpG island methylation of 24 genes and identified aberrant hypermethylation at 10 genes. We then performed quantitative methylation analyses by bisulfite pyrosequencing of the identified genes in 317 patient samples from three independent studies and assessed relations between methylation and clinical outcome.

Results: In an initial training cohort of 89 patients with MDS, methylation frequencies of individual genes ranged from 7% to 70% and were highly concordant. Therefore, we defined a methylation z score based on all genes for each patient. We found that patients with higher levels of methylation, compared with patients with lower levels, had a shorter median overall survival (12.3 v 17.5 months, respectively; P = .04) and shorter median progression-free survival (6.4 v 14.9 months, respectively; P = .009). This methylation prognostic model was independent of age, sex, and IPSS group. Applied to two validation cohorts (228 patients), this model was confirmed as an independent prognostic predictor for survival. Although methylation at baseline did not correlate with clinical response to decitabine, we observed a significant correlation between reduced methylation over time and clinical responses.

Conclusion: DNA methylation predicts overall and progression-free survival in MDS.

Conflict of interest statement

Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.

Figures

Fig 1.

Kaplan-Meier survival estimates of overall and progression-free survival in patients with myelodysplastic syndromes. Overall survival in (A) training cohort, (B) first validation cohort, (C) second validation cohort, and (D) all patients. Progression-free survival in (E) training cohort, (F) first validation cohort, (G) second validation cohort, and (H) all patients. In each panel, patients are grouped into methylation low (gold) or methylation high (blue) groups according to their combined methylation z scores. Median survival (95% CI) of each group in each panel is shown. P values are based on the log-rank test.

Fig 2.

The DNA methylation prognostic model and cytogenetic risk groups. The Kaplan-Meier estimates show survivals for groups of patients with cytogenetic good risk (A: overall survival; D: progression-free survivals), intermediate risk (B: overall survival; E: progression-free survival), and high risk (C: overall survival; F: progression-free survival).

Fig 3.

Methylation changes at multiple time points after treatment. Average methylation changes (before and after 4 months on therapy) were compared between patients treated with decitabine (DAC) and supportive care (SUP). Methylation decreased by 11.2% in patients on DAC but increased by 20.1% in patients on SUP (P = .04 by Mann-Whitney U test). These methylation changes were then analyzed for correlation with response in 34 patients (DAC arm: two patients with complete remission [CR], three patients with partial remission [PR], four patients with hematologic improvement [HI], four patients with stable disease [SD], and one patient with progressive disease [PD]; supportive care arm: two patients with HI, six patients with SD, and 12 patients with PD). A greater decrease was observed in patients with CR or PR (40.6% ± 15.7%) compared with HI (9.8% ± 13.2%). Methylation increased by 15.4% in patients with SD and by 27.2% in patients with PD (P = .02 by Kruskal-Wallis test).