Abstract
SMZL usually has an indolent clinical course. Transformation to aggressive lymphoma occurs in only 4-5% of cases. Tools to identify patients facing a more aggressive clinical course and targeted therapies are needed. In the last years, we have characterized SMZL by DNA profiling (Blood 2011) and RNA/miRNA expression profiling (Mod Path 2013), and shown that somatic mutations of NOTCH2 and other genes involved in the pathway are frequent and mutually exclusive with mutations affecting NKFB pathway (JEM 2012). Here, we report genome-wide DNA promoter methylation profiling of 137 SMZL samples with its clinical and biologic implications.
Methods
A test series of 101 samples, including 3 SMZL cell lines (Karpas1718, VL51, SSK41) and 98 clinical SMZL cases from 10 Centers and, a validation set of 36 SMZL cases from 2 additional Centers were analyzed with the Illumina Infinium HumanMethylation27 arrays, as described (BJH 2013). Gene GEP was available for a subset of 11/101 cases. Methylation profiles were correlated with the presence of the copy number aberrations (CNAs) and somatic mutations recurrent in SMZL, IGHV mutational status, HCV infection and clinical data. Probes mapped outside CpG islands were discarded. Supervised analyses were performed using a t-test on the continuous beta values, followed by multiple test correction. For Fisher’s test, probes were classified into unmethylated (beta-values <0.3) or methylated (beta-value >0.3).
Results
Unsupervised clustering analysis of 101 SMZL samples identified 2 clusters.
Cluster one (23/101, 23%) comprised the 3 cell lines and included cases characterized by a generalized higher degree of promoter methylation (High-M), and had a significantly poorer overall survival (OS) than the remaining 79/101 (77%) cases (Low-M) (P=0.048; HR 2.4; 95% CI, 1-5.8).
To better understand the biology underlying this clustering, supervised analysis comparing High-M to Low-M cases was performed. By combining t- and Fisher's tests, 3200 gene probes showed differential methylation status between High-M and Low-M lymphomas. This probes signature was significantly enriched of promoters of promoter regions of PRC2-complex targets, genes involved in chromatin remodeling, and HOX and SOX family genes, and genes down-regulated by CDH1. Genes coding for subunits of the PRC2-complex (EZH2, EED, SUZ12) appeared unmethylated in their promoter regions and over-expressed by GEP in the High-M set, while EZH2-target genes and genes harboring the H3K27me3 mark had methylated promoters and down-regulated expression. A number of reported tumor-suppressor genes were highly methylated in High-M cases, including KLF4, DAPK1, CDKN2A/B, CDKN1C, CDH1, CDH2 and TIMP3. Known pro-survival lymphoma genes, such as SYK, MCL1, CARD11, BIRC5, BTK, BCL10, FOXP1, CD40, TACI and TCL1A/B, were un-methylated and over-expressed in High-M cases.
The prognostic relevance of the High-M phenotype was confirmed in an independent validation set. An unsupervised clustering using the top-100 differentially methylated probes obtained by t-test in the test series identified two clusters with one (11/36, 30%) bearing methylation profile similar to High-M and a poorer OS (P=0.038).
All the samples were then pooled to improve the statistical power of further analyses regarding the impact of methylation on the clinical course and on the association with biologic variables. High-M was significantly associated with poorer OS (P=0.008; HR 2.58; 95% CI, 1.2-5.4), transformation into aggressive lymphoma (33% vs 0.02%, P<0.001) and with 7q31-33 loss (38% vs 16%, P=0.026). In addition, High-M showed a higher frequency of NOTCH-pathway somatic mutations (39% vs 16%, P=0.063). HCV was more frequent in Low-M (20% vs 4%, P=0.058).
In a Cox regression model adjusted for the variables significantly associated with outcome at univariate analysis [High-M, age>60, Interguppo Italiano Linfomi (IIL) SMZL score, HCV infection], High-M (HR 16; 95% CI, 1.3-201) and IIL SMZL score (HR 4.5; 95% CI, 1.3-15) retained their prognostic significance on OS (P<0.001).
Conclusions
Genome-wide promoter-methylation profiling in splenic MZL identified a subset of patients with poorer OS, characterized by a high promoter methylation and by specific genetic and biologic features. These data may provide the rational to explore epigenetic drugs in SMZL. *, AJA and AR equally contributed.
Disclosures:
No relevant conflicts of interest to declare.
Methylation profiling of normal individuals reveals mosaic promoter methylation of cancer-associated genes