Cross-Platform Assessment Of Genomic Imbalance In Diffuse Large B-Cell Lymphoma Identifies Novel Candidate Loci and Genes With Prognostic Value and Roles In Lymphomagenesis

Diffuse large B-cell lymphomas (DLBCL) display marked clinical, pathologic, and genetic heterogeneity. With current frontline immunotherapy (RCHOP), only about 40% of patients are cured, with most relapses occurring within the first 2-3 years. Patients are currently risk-stratified based primarily on clinical features where the inclusion of molecular biomarkers into risk assessment could impact the potential to identify those patients most likely to have refractory disease or have an early relapse. Various cytogenomic studies have revealed the prognostic significance of genomic gain/loss in DLBCL, but their lack of utility and reproducibility across datasets can be attributed to not only different patient populations, but also the use of disparate platforms and analytical methods. The goal of the present study was to use a common analytical approach across different clinical datasets, to identify genomic loci with robust prognostic value in DLBCL.

To this end, GISTIC was applied to copy number data of newly-diagnosed DLBCL (GSE11318 [n=170], E-MEXP-3463 [n=49]) and together with another similarly analyzed published dataset (PMID: 22975378 n= [180]). Regions of known copy number variations were excluded from the analysis. A total of 35 overlapping minimal common regions (MCRs) and 21 overlapping peaks of genomic gain/loss with well-defined calling parameters were identified in at least 2 of the 3 datasets with a minimum occurrence of 5%. Overlapping peaks mapped to loci of genes with known involvement in DLBCL such as TP53, CDKN2A, and TNFAIP3 confirming the validity of the approach. Individual MCRs were tested for association with overall survival using the Kaplan-Meier method and log-rank statistic in three RCHOP datasets: E-MEXP-3463, GSE15127 [N=124], and an in-house dataset for which array-CGH was performed using a targeted oligonucleotide (Agilent Technologies) with DNA extracted from 1-3 CD20+ enriched cores of formalin–fixed paraffin-embedded de novo DLBCL; IH [n=46]. Nine MCRs were found to significantly associate (P ≤0.05) with poor outcome: gain of 3q, 9q, 19p, and loss of 1p, 8p, 9p, 13q, 15q, and 17p. Four remained significant after multiple testing correction: gain of chr3:187,651,865-196,853,350 (P<0.001), chr9:138,543,735-140,878,804 (P=0.0110), and loss of chr15:40,295,857-46,224,648 (P=0.012), chr17:1,000,000-16,936,602 (P=0.050). Genomic complexity assessed as previously described (PMID:22975378), failed to significantly correlate with outcome in any of the three RCHOP datasets tested.

Correlation of gene expression levels to copy number change was evaluated in the (GSE11318 [n=162]) dataset. A total of 397 differentially expressed genes (P ≤0.05, FDR) showing positive correlation with copy number status were identified within 16 MCRs of which, five mapped to three peak MCRs. Genes mapped to the peak MCRs were found to have roles in NFKB activation, iron transport, phosphatidylcholine biosynthesis and regulation of transcription all of which represent novel therapeutic targets in DLBCL. For the nine MCRs associated with poor outcome, 69 genes exhibited positive correlation with copy number and enrichment of the KEGG pathway using these genes in the DAVID 6.7 program identified the endocytosis, apoptosis, and glycosylphosphatidylinositol (GPI)-anchor biosynthesis pathways.

In summary, using this platform-agnostic approach, common and novel loci of genomic imbalance in DLBCL were identified, of which some were found to have clinical significance and could be included, with additional validation, in patient risk assessment. This analysis also afforded the identification of novel genes with possible roles in lymphomagenesis, representing potential therapeutic targets in DLBCL.