Abstract
Introduction
Multiple myeloma is a clinically highly heterogeneous disease, which is reflected by both a complex genome and epigenome. Dynamic epigenetic changes are involved at several stages of myeloma biology, such as transformation and disease progression. Our previous genome wide epigenetic analyses identified prognostically relevant DNA hypermethylation at specific tumor suppressor genes (Kaiser MF et al., Blood 2013), indicating that specific epigenetic programming influences clinical behavior. This clinically relevant finding prompted further investigation of the epigenomic structure of myeloma and its interaction with genetic aberrations.
Material and Methods
Genome wide DNA methylation of CD138-purified myeloma cells from 464 patients enrolled in the NCRI Myeloma XI trial at presentation were analyzed using the high resolution 450k DNA methylation array platform (Illumina). In addition, 4 plasma cell leukemia (PCL) cases (two t(11;14) and two (4;14)) and 7 myeloma cell lines (HMCL) carrying different translocations were analysed.
Analyses were performed in R Bioconductor packages after filtering and removal of low quality and non-uniquely mapping probes.
Results
Variation in genome wide DNA methylation was analyzed using unsupervised hierarchical clustering of the 10,000 most variable probes, which revealed epigenetically defined subgroups of disease. Presence of recurrent IGH translocations was strongly associated with specific epigenetic profiles. All 60 cases with t(4;14) clustered into two highly similar sub-clusters, confirming that overexpression of the H3K36 methyltransferase MMSET in t(4;14) has a defined and specific effect on the myeloma epigenome. Interestingly, HMCLs KMS-11 and LP-1, which carry t(4;14), MM1.S, a t(14;16) cell line with an E1099K MMSET activating mutation as well as two PCLs with t(4;14) all clustered in one sub-clade. The majority (59/85) of t(11;14) cases showed global DNA hypomethylation compared to t(4;14) cases and clustered in one subclade, indicating a epigenetic programming effect associated with CCND1, with a subgroup of t(11;14) cases showing a variable DNA methylation pattern.
In addition to translocation-defined subgroups, a small cluster of samples with a distinct epigenetic profile was identified. In total 7 cases with a shared specific DNA methylation pattern (median inter-sample correlation 0.4) were identified. The group was characterized by DNA hypermethylation (4,341 hypermethylated regions vs. 750 hypomethylated regions) in comparison to all other cases. Intersection of regions hypermethylated in this subgroups with ENCODE datasets revealed mapping to poised enhancers and promoters in H1-hESC, indicating functionally relevant epigenetic changes. Gene set enrichment analysis (KEGG) demonstrated enrichment of developmental pathway genes, e.g. Hedgehog signaling (adj p=5x10exp-13), amongst others and all four HOX clusters were differentially methylated in this group. Of note, three of seven cases in this subgroup carried a t(11;14) and all t(11;14) or t(11;14)-like HMCLs clustered closely together with these patient cases, but not with the cluster carrying the majority of t(11;14) myeloma or t(11;14) PCLs. This potentially indicates that t(11;14) HMCL could be derived from a subgroup of patients with specific epigenetic characteristics.
Conclusion
Our results indicate that the recurrent IGH translocations are fundamentally involved in shaping the myeloma epigenome through either direct upregulation of epigenetic modifiers (e.g. MMSET) or through insufficiently understood mechanisms. However, developmental epigenetic processes seem to independently contribute to the complexity of the epigenome in some cases. This work provides important insights into the spectrum of epigenetic subgroups of myeloma and helps identify subgroups of disease that may benefit from specific epigenetic therapies currently being developed.
Walker:Onyx Pharmaceuticals: Consultancy, Honoraria.
Author notes
Asterisk with author names denotes non-ASH members.
