Multiple Secondary Genetic Events Inducing Histologic Progression (HP) of Non-MALT Marginal Zone Lymphoma (MZL): An Integrated Genomic Analysis of Genome-Wide Copy-Number-Changes (CNC) and Transcriptomic Data.

Abstract

HP of indolent B cell lymphoma to a high grade lymphoma is usually associated with a rapidly progressive clinical course and a short survival. The risk of HP in MZL is evaluated around 10–20%, and can be detected at diagnosis or during the evolution of the disease. No criteria have been yet formally adopted to range HP, even in the more recent OMS classification, although therapeutic impact is important. In this study we wished to determine the genetic alterations implicated in the HP of non-MALT MZL, using an integrated analysis of genome-wide CNC and transcriptomic data. Fresh-frozen tumor biopsies and clinical data were obtained from 66 untreated non-MALT MZL patients of 3 institutions (SLS, HD, CHLS, France). Features reviews including morphologic aspect, immunophenotype, and conventional cytogenetics, classified 52 cases as MZL and 14 cases as MZL with HP (HP-MZL). HP was assessed by the presence of large cells, the presence of cohesive sheets of larges cells, and elevated Ki67. CNC were analysed in 20 splenic MZL (SMZL), in 6 HP-SMZL cases including 2 matched cases, and in 3 germ-line DNA of these cases using array comparative genomic hybridization (array-CGH) (Agilent, 105K). Normalization of the array-CGH data was realized by Lowess correction. Transcriptomic analysis of 32 SMZL and 8 HP-MZL, including 2 matched cases, was performed with a nylon DNA microarray of 9,216 human genes. After Lowess correction, data were visualized by hierarchical clustering allowing to distinguish patterns of gene expression corresponding to precise functions, cell or tissue subtype. The ability of individual transcripts to distinguish MZL and HP-MZL subtypes was calculated using discriminating score and bootstrap resampling. The most common imbalances detected by cytogenetics were gains of 3/3q (n=11/61, 18 %) and 18q (n=13/61, 21%), and deletions in 7q (n=18/61, 17%), and were detected in both groups, SMZL and HP-SMZL. Array-CGH analysis showed that no recurrent CNC was specific of all HP-SMZL. Only 1 CNC was present in 4 of the 6 HP-SMZL cases, 3 in 3 cases, and 29 in 2 cases. These CNC could also be seen in SMZL. Comparing the matched pairs, we identified secondary CNC changes. One was located in 4p and includes 9 genes. Among them, was Cyclin G associated kinase. This CNC was also present in 2 other HP-MZL cases. Another was located in 6p and includes 4 genes, comprising IRF4. Another HP-SMZL case exhibited a HP-SMZL specific CN gain of the oncogene Myc in 8q. These 3 genes are known to be involved in proliferation. Gene expression profiling showed overexpression of genes involved in proliferation and underexpression of p53 in HP-MZL cases compared to SMZL cases. The proliferation signature included PCNA, genes involved in glycolysis (GAPD, LDHA), in cell cycle (CDC10, CDK4), and cell proliferation (TNFRSF13B, MCL1). Two other signatures included genes overexpressed specifically in HP-MZL: one specific to the stroma with MIG, STAT1, Cathepsin B and C, CREG, IGFR2R; and a cluster of unknown function regrouping several oncogenes (DEK, DJ-1), genes related to cell machinery and cell motility (HMMR, PRPF4, PPOX) and genes involved in signal transduction (ZNF146, FLJ10618, PBXL2). In addition to p53, HP-MZL underexpressed genes included PMSB1 in the proteasome, KLF2 and PFC. In conclusion, integrated genetic and transcriptomic analyses of non-MALT MZL and HP-MZL showed that histologic progression was related to the alteration of cell proliferation. Beside chromosomal alterations typical to SMZL, secondary CNC could be detected in HP-SMZL cases. These CNC included genes associated with cell proliferation: Cyclin G associated Kinase, IRF4, and Myc. These results highlight that deregulation of different pathways drive the MZL cells toward a higher proliferative rate charateristic of histologic progression.