High-Resolution Genomic Methylation Analysis Using Next Generation Sequencing Identifies Loci Associated With Differential Prognosis In Mantle Cell Lymphoma Patients Treated With Bortezomib + DA-EPOCH-R

Forty-three newly diagnosed MCL patients were treated with single agent Bortezomib followed by DA-EPOCH-R (Dunleavy et al, ASH 2012). To understand the genomic and epigenomic basis of differences in patient outcomes, particularly disease free survival (DFS), we carried out high-resolution genome-wide methylation analysis using enhanced RRBS (ERRBS) and correlated cytosine methylation to gene expression (GE) and patient outcomes. For ERRBS, genomic DNA was extracted from CD19+ selected cells from lymph node biopsies or peripheral blood prior to therapy. Library fragment lengths of 150-250 bp and 250-400 bp were prepared and gel isolated per Akalin et al (PLOS Genetics, 2012) and sequencing was performed on an Illumina Hi-seq 2000. 24 out of 24 patient samples passed quality control with methylation assayed at >3.3 million CpG dinucleotides per sample on average 80x coverage per cytosine.

Patients were divided into two groups based on 2 year DFS. There were 39,373 differentially methylated (DM) loci (25% difference in methylation, q<0.01) correlating to 3,968 genes between these two groups. The hyper and hypomethylated loci were annotated with hg19 to find gene associations. The annotated analysis represented 74% and 26% hypermethylated loci located in gene body and promoter regions respectively. Similarly, 70% and 30% of hypomethylated loci were located in gene body and promoter regions. Genomic Regions Enrichment of Annotations Tool (GREAT), a next-generation software aimed at the interpretation of genome-wide cis-regulatory data sets, was used to understand the functional significance of identified DM cytosines. Pathway analysis by GREAT for the DM genes showed significant (p<0.05) enrichment for Wnt and Cadherin signaling pathways. 67% of the DM genes had DNA binding transcription factor activity with tumor suppressors BCOR, HIC1, TP73 from hypomethylated genes and oncogenes GATA3 and MAFB from hypermethylated genes.

Simultaneously, Cox proportional hazard models evaluating time-to-progression based on RNA expression using Affymetrix U133 Plus 2 arrays identified 1,933 genes as significantly (p<0.05) associated with DFS. The top canonical pathways enriched by these genes included Protein Kinase A Signaling, PI3K/AKT Signaling, and Protein Ubiquitination Pathway. Top molecular functions of prognostic loci by GEP were cell growth and proliferation. Our analysis has shown that high expression of CDC25A, NFKB2, GLI3, FOXO3, and USP9X were likely associated with increased hazard risk in MCL patient after Bortezomib + DA-EPOCH-R treatment. The gene sets identified by the two platforms were distinct, with only 343 genes in common between DM and DE genes.

Based on these findings, we conclude that genomic methylation analysis by next-generation sequencing can identify functionally important cis-regulatory cytosines associated with prognosis in MCL. Prognostic gene sets identified by methylation and gene expression were distinct with transcriptional regulation being the major function of DM genes while cell growth and proliferation being the major function of DE genes. We are developing an integrated prognostic platform for patient outcomes following Bortezomib + DA-EPOCH-R combining genomic methylation and gene expression.