Genome-Wide DNA Methylation Analysis Reveals Epigenetic Deregulation of Key Biological Pathways in Therapy-Related Myeloid Neoplasms

Therapy-related myeloid neoplasms (tMN), including therapy-related myelodysplastic syndromes (tMDS) and therapy-related acute myeloid leukemia (tAML) are associated with resistance to chemotherapy and a more unfavorable prognosis than their de novo counterparts. While many studies have explored the epigenome of de novo MDS, little is known about the epigenetic status of tMN. Recently, we reported that tMN showed distinct mutational profiles whether they arose after an initial solid tumor or a lymphoproliferative disorder, indicating that aside from the type of therapy received it is possible the type of primary neoplasm might also help determine the phenotype of t-MN. Therefore, we sought to perform a comprehensive epigenomic study of tMN that would help improve our current understanding of epigenetic deregulation in this disorder. Using next-generation bisulfite sequencing we analyzed the DNA methylation status at ~3M CpG sites of 15 cases, including 5 de novo MDS (dnMDS), 5 tMN with a prior history of Hodgkin lymphoma (HL-tMN) treated with MOPP/BEACOPP-based regimens, and 5 tMN arising after breast cancer (BC-tMN) treated with combined chemo-radiotherapy. There were no significant differences in median age, median bone marrow blast percentage, or median latency to t-MN for the HL and BC groups. Karyotype mainly consisted of abnormalities of chromosomes 5 or 7, or complex karyotypes in all groups. None of the patients had balanced translocations or abnormalities of the 11q23 locus. We first sought to determine whether tMDS and dnMDS are epigenetically distinct. For this we compared the DNA methylation profiles of the tMDS (n=6) vs. dnMDS (n=5). We identified 1504 statistically significant (FDR<0.1 and mean methylation difference ≥25%) differentially methylated regions (DMRs). Relative to dnMDS, tMDS showed a markedly hypomethylated profile. Pathway analysis using DAVID revealed that these DMRs were significantly enriched for genes in the Wnt signaling (FDR=3.8x10^-6) and cadherin pathways (FDR=5.3x10^-7), indicating that these key pathways maybe play different roles in these two forms of MDS. Next we asked whether among tMNs the type of neoplasm preceding the development of tMN might also influence the epigenome. To address this we performed a direct comparison of the DNA methylation profiles for the HL-tMN (n=5) vs. BC-tMN (n=5), which identified 457 statistically significant DMRs, the vast majority of which localized to intronic and intergenic regions and were hypomethylated in HL-tMN relative to BC-tMN. Finally, in order to determine the role that epigenetic deregulation may play in different forms of AMLs not arising de novo, we compared the epigenetic profiles of tAML (n=4) to a cohort of AMLs arising after progression from a myeloproliferative neoplasm (MPN) (n=5). Notably, despite their shared features of chemoresistance and poor prognosis, the epigenetic profiles of these two entities were vastly different, with 14,887 statistically significant DMRs detected. These DMRs were depleted at promoter regions (DMRs 4% vs. Background [BG] 12%, Fisher p-value[p]=0.038) and CpG islands (DMRs 9% vs. BG 30%, p=0.0003), but were enriched at regions outside of CpG islands (DMRs 73% vs. BG 54%, p=0.008). Gene ontology analysis showed these DMRs were significantly enriched for genes involved transcriptional regulation (FDR=0.02), nucleotide binding (FDR=0.00025), regulation of RNA metabolic process (FDR=0.02), and protein kinase activity (FDR=0.009). In addition, these DMRs were significantly enriched at enhancer regions (p<0.0001), indicating they may play key regulatory roles. In summary, we have completed the first comprehensive analysis of the epigenome-wide abnormalities associated with tMN. Our findings demonstrate that tMN are epigenetically distinct from dnMDS and that these abnormalities target biological pathways known to play a key role in self-renewal and differentiation of hematopoietic stem and cancer cells, as well in MDS pathogenesis. Moreover, we demonstrate that similar to our findings at the mutational level, epigenetic deregulation in tMN also has distinct profiles strongly correlated with the type of first malignancy preceding the tMN. Finally, AML arising as a progression of MPN or after prior chemotherapy show robust epigenetic differences that clearly distinguish these two entities.