The EBV-Positive Tumor Methylome is Distinct from EBV-Negative in Diffuse Large B-Cell Lymphoma

1. Introduction

Epstein-Barr virus (EBV)-driven diffuse large B-cell lymphoma (DLBCL) represents a clinically aggressive tumor subtype that is often refractory to standard R-CHOP chemotherapy (Rituximab, Cyclophosphamide, Doxorubicin Hydrochloride, Vincristine Sulfate, Prednisone) [1,2,3]. Currently, no targeted treatment options exist for EBV(+) DLBCL. While EBV is generally accepted to play a role in the pathogenesis of DLBCL, the specific disease mechanisms and the significance of the viral contribution in vivo have not been clearly defined. Furthermore, no consensus standard exists for diagnosing DLBCL as EBV-positive using the conventional in situ hybridization tissue stain for EBV-encoded small RNA (EBER-ISH) [4,5,6,7,8,9]. Until EBV diagnostics and pathogenic mechanisms are elucidated, the development of targeted treatments for EBV(+) DLBCL will remain a challenge.

The role of EBV in modifying the epigenetic landscape of DLBCL offers promising avenues for understanding viral pathogenesis, identifying diagnostic biomarkers, and developing targeted therapeutics. In vitro studies using EBV-infected cell lines have shown that the virus directly modifies both the host cellular epigenome and its own genome through DNA methylation [10,11]. DNA methylation allows for dynamic changes in gene expression via hypermethylation (silencing) or hypomethylation (activation) of functional promoter regions. These alterations serve as epigenetic “switches” that can respond to external environmental cues, such as those present in the immunosuppressed tumor microenvironment. In vitro evidence has demonstrated the capacity of EBV to modulate the host epigenome by upregulating oncogene expression and downregulating tumor suppressor genes, thereby promoting B-cell malignant transformation [10,11,12]. Our study takes this investigation to in vivo clinical tissue samples of EBV(+)/(-) DLBCL to better define the tumor cell methylome and clarify the epigenetic role of EBV.

We characterized the DLBCL methylome in EBV-mediated disease using the high-throughput DNA methylation microarray, Illumina MethylationEPIC v2.0, which interrogates over 935,000 CpG sites across the human genome. This approach has been successfully employed across numerous in vivo cancer studies, including The Cancer Genome Atlas (TCGA), which profiled thousands of samples across diverse tumor types [13,14,15,16]. The Illumina EPIC methylation array is particularly well-suited for formalin-fixed tissue, allowing for a comprehensive genome-wide methylation profiling. Through this technology, the TCGA project has systematically characterized epigenomic alterations, yielding insight into oncogene activation, epigenetic regulation of gene expression, and tumor microenvironment. These insights have informed advances in diagnostics and targeted immunotherapies [14].

We applied the Illumina EPIC methylation array to a clinical cohort of EBV(+)/(-) DLBCL to identify distinct tumor methylomes associated with EBV-driven oncogenesis. This pilot study demonstrates that EBV(+) DLBCL harbors a unique signature compared to EBV(-) disease, underscoring the need for future investigation in larger cohorts.

2. Materials and Methods

2.1. Clinical Study Population and Tissue Sample Collection

A total of 31 clinical tissue cases of DLBCL from our institution, including 9 EBV(+) and 22 EBV(-), from 2000-2023 were selected for the study. In addition, 12 control tissue cases were selected over this period for comparison, including: 2 EBV(+) polymorphic lymphoproliferative disorders, 2 EBV(+) plasmacytic hyperplasia post-transplant lymphoproliferative disorders, 4 EBV(+) infectious mononucleosis tonsil/lymph nodes, and 4 EBV(-) reactive follicular hyperplasia lymph nodes. The cases originated from a wide range of immunosuppressive states including post solid organ/stem cell transplantation, systemic chemoradiation therapy-related (iatrogenic), autoimmune disease treated with systemic immunosuppressive pharmacotherapy (e.g. methotrexate, long-term steroid use, etc.), and primary immunodeficiency. The contribution of immunosenescence, as defined by age >65 years old, was likely in most tumor cases.

Representative hematoxylin and eosin-stained (H&E) slides for each case were reviewed by the study hematopathologist (Volaric) to confirm the diagnosis. Standardized automatic ISH staining platform (Leica Biosystems) for EBER-ISH was used in the evaluation. All cases were categorized as EBV(+) using EBER-ISH (ISH5687-A). EBV-positivity was defined in DLBCL or control cases as exhibiting at least 50% cellular staining in either tumor cells or lymphocytes for EBER-ISH.⁵ EBV-negative cases showed no cellular staining. Only cases in which diagnostic material was obtainable (formalin-fixed paraffin-embedded (FFPE) tissue blocks or unstained sections) were included. The study was approved by the UVM Institutional Review Board.

2.2. DNA Methylation Analysis

2.2.1. Global Methylation Profiling Using Illumina EPIC Methylation Array

FFPE tumor and control tissue sections were macro-dissected and sent to CD Genomics (NY, USA) for Illumina Infinium MethylationEPIC BeadChip array (935K). Samples were subjected to DNA extraction using protocols optimized for archival FFPE material. DNA quantity and quality were assessed prior to array hybridization. The array includes >935,000 CpG probes that span gene promoters, enhancers, gene bodies, and intergenic regions, and includes known cancer-associated loci and open chromatin regions. After hybridization and scanning, raw data were delivered in the form of IDAT files containing methylated and unmethylated signal intensities.

2.2.2. Bioinformatic Analysis

Illumina EPIC methylation analysis was performed using R (version 4.4.1). Raw IDAT files were processed using the SeSAMe package (v1.22.2) [17]. Initial quality filtering was conducted using the SeSAMe pOOBAH (p-value with out-of-band [OOB] array hybridization) method, which performs dye bias correction and calculates detection p-values. The background subtraction was done using the QCDPB criteria: Q, qualityMask Masking probes of poor design; C, inferInfiniumIChannel Infer channel for Infinium-I probes (needed for dye bias correction step); D, dyeBiasNL actual dye bias correction (non-linear); P, pOOBAH Detection p-value masking using oob; B noob background subtraction using oob. Only samples with median intensity >10.5 and mean detection p-values <0.05 were retained (n=43), resulting in 889,341 high-confidence probes for downstream analysis. Normalized beta values (ranging from 0 to 1) were used for downstream analysis. Principal component analysis (PCA) was performed using `prcomp()` to assess global methylome clustering. Differential methylation analysis was performed using linear models implemented in the ‘DML’ (differential methylation locus) function of the SeSAMe package, comparing EBV(+) vs. EBV(–) DLBCL and DLBCL vs. control tissues. Probe-level significance was adjusted using p-values of 0.05. Significant probes were annotated to genes using sesameAnno_attachManifest for EPICv2. Significant probes were aggregated at the gene-level to identify differentially methylated regions (DMRs) based on a log2 fold change cutoff and multiple testing thresholds. For enrichment analysis of gene sets, we used the testEnrichment function of the SeSAMe package, which performs a Fisher's exact test to determine if the number of significant probes associated with a gene is greater than expected by chance. We further used oncoEnrichR (v1.5.2), to map differentially methylated genes to curated cancer pathways. Immune cell composition inference employed IDOL Optimized CpGs from the Illumina Human Methylation Dataset package FlowSorted.Blood.EPIC (v2.8.0). This estimated the relative abundance of immune cell populations across samples. Differences in immune composition between EBV(+) and EBV(–) DLBCL were assessed using the Wilcoxon rank-sum test (p<0.05). Commercially available antibodies were used for immunohistochemical analysis of CD20 (Ventana Ultra, Thermo Fisher L26), CD3 (Leica, Biocare LN10), and CD56 (Ventana Ultra, Roche MRQ-42) on a representative case of EBV(+) and EBV(-) DLBCL.

3. Results

3.1. Clinicopathologic Characteristics of Study Cohort

To elucidate the DNA methylation profiles of EBV-mediated DLBCL, we established a clinical cohort of DLBCL (n=43) from our institution, consisting of EBV(+) DLBCL (n=9), EBV(-) DLBCL (n=22), and control cases [EBV(+) n=8; EBV(-) n=4]. EBV status for each case was determined by EBER-ISH chromogenic staining, whereby EBV(+) cases exhibited ≥50% positive nuclear staining in lesional cells, and EBV(-) cases showed no cellular staining. Representative EBV(+) and EBV(-) DLBCL cases, demonstrating EBER-ISH staining patterns, are shown in Figure 1.

The clinicopathologic characteristics of the EBV(+)/(-) DLBCL cohort (n=31) are summarized in Table 1. Most patients (58%) were aged >65 years old, and roughly one-third had a defined immunodeficiency, with post-transplantation (22%) and therapy-related/iatrogenic (22%) immunodeficiency states particularly enriched in EBV(+) DLBCL cases. The majority of cases (55%) were treated with standard R-CHOP chemotherapy. Less than half of cases responded to therapy, with disease-related mortality occurring in 67% of EBV(+) DLBCL cases compared to 32% of EBV(-) DLBCL cases.

3.2. Global DNA Methylation Analysis of Clinical Cohort

Global DNA methylation analysis was performed on the EBV(+)/(-) DLBCL cohort (n=31) and controls (n=12) using the Ilumina EPIC methylation array, which interrogates approximately 950,000 CpG loci across the human genome. Quality control was carried out using the SeSAMe package to ensure data reliability and minimize technical artifacts.¹⁷ Based on signal intensity and detection p-value assessments, samples and probes failing established thresholds were removed (Table S1). Probes were filtered out if they had detection p-values >0.05, or were poorly designed (Figure S1). Following quality control and filtering, 889,341 probes were retained for analysis.

3.3. Methylome Patterns: All DLBCL versus Control Cases

All EBV(+)/(-) DLBCL (n=31) were directly compared to control cases (n=12), which included: 2 EBV(+) polymorphic lymphoproliferative disorders, 2 EBV(+) plasmacytic hyperplasia post-transplant lymphoproliferative disorders, 4 EBV(+) infectious mononucleosis tonsil/lymph nodes, and 4 EBV(-) reactive follicular hyperplasia lymph nodes. PCA revealed a clear separation between DLBCL and controls (Figure 2A). A comparison of the Beta values representing methylation levels of all probes, overall showed a trend of hypomethylation in DLBCL compared to control cases (Figure 2B), suggesting global methylation associated with DLBCL pathogenesis.

Differential methylation analysis, performed using the statistical probe filtering and normalization pipeline of the SeSAMe package [17], identified 330,872 differentially methylated CpG sites (p <0.05) between DLBCL and controls (Table S1; Figure S2A). Gene-level aggregation of probes identified 4,110 hypermethylated genes and 120 hypomethylated genes in DLBCL (FDR <0.05 and log₂ fold change of 2; Table S2A), consistent with prior reports of global hypomethylation in cancer (Figure 2C) [18,19,20,21]. Many of these genes, including tumor suppressors such as CDKN2A, HLA-A, and FAT4, have established roles in hematolymphoid malignancies (Figure 2D). Pathway enrichment analysis of hypermethylated genes revealed significant involvement of cancer-related pathways, notably Wnt/β-catenin and Cadherin signaling, which have complex roles in lymphomagenesis (Figure 2E) [22,23,24,25,26,27]. No significant enrichment pertaining to cancer-associated pathways was discovered in the hypo-methylated gene set. The tumor suppressor gene CDKN2A, known to contribute to aggressive disease behavior when inactivated [28,29,30,31,32,33], was significantly hypermethylated in DLBCL compared to controls (Figure 2F).

3.4. Methylome Patterns: EBV(+) DLBCL versus EBV(-) DLBCL

We next sought to identify methylation differences between EBV(+) and EBV(-) DLBCL. PCA analysis revealed clear separation between EBV(+) (n=9) and EBV(-) cases (n=22) (Figure 3A). Differential methylation analysis using SeSAMe identified 117,334 differentially methylated CpG sites (p<0.05) between EBV(+) DLBCL and EBV(-) DLBCL (Table S1; Figure S2B). Comparison of beta values across all probes indicated an overall hypermethylation pattern in EBV(+) DLBCL relative to EBV(-) DLBCL (Figure 3B). Gene-level methylation analysis identified 2,114 differentially methylated genes between EBV(+) DLBCL and EBV(-) DLBCL (FDR <0.05 and log₂ fold change of 2; Table S2B; Figure 3C). The distribution was skewed towards hypomethylation in EBV(+) cases (630 hypermethylated vs. 1,484 hypomethylated genes). Of these, 1,557 genes had cancer-related functions, with a similar skew towards hypomethylation (367 hypermethylated and 1,190 hypomethylated).

Among the hypermethylated genes were both lymphoid oncogenes and tumor suppressor genes, including 15 oncogenes, 4 tumor suppressor genes, and 6 with dual oncogenic and tumor suppressor roles (Figure 4A). Notably, hypermethylated tumor suppressor genes in EBV(+) DLBCL included CBFA2T3, CSNK1E, and HDAC10, all of which are associated with hematolymphoid malignancies [34,35,36,37,38]. Pathway enrichment analysis of the hypermethylated genes revealed involvement in tumor suppressor pathways such as P53 feedback loops and TGF-beta signaling [2,39,40,41,42], oncogenic pathways including P38 MAPK and Ras pathways [43], and immune signaling pathways like IFN-gamma signaling.[44,45]. The HDAC10 gene showed significant hypermethylation in EBV(+) DLBCL (Figure 4C). While HDAC10 gene can function as either an oncogene or tumor suppressor depending on cellular context, it has been reported to exert tumor-suppressive effects in B-cell lymphomagenesis [38,46,47,48].

We also characterized hypomethylated genes in EBV(+) DLBCL. The top 25 hypomethylated genes included 6 oncogenes, 8 tumor suppressor genes, and 11 genes with dual functions (Figure 4D). These included oncogenes FLT3 and HDAC6, as well as dual-function genes NPM1 and BTK, all implicated in lymphomagenesis [49,50,51,52,53,54,55]. Pathway analysis showed that hypomethylated genes were enriched in oncogenic cascades such as Insulin/IGF MAPK cascade, DNA replication, and RAS pathway (Figure 4E) [43,56,57]. The BTK gene, a well-established oncogene in B-cell lymphomagenesis, exhibited pronounced hypomethylation in EBV(+) DLBCL compared to EBV(-) DLBCL (Figure 4F) [58].

3.5. Immune Cell Composition in EBV(+) DLBCL

The immune cell composition of DLBCL plays an important role in the clinicopathologic progression of the disease. To assess differences between EBV(+) and EBV(-) DLBCL, we performed immune cell inference through deconvolution of the methylation data. This analysis revealed a distinct immune cell composition profile for EBV(+) DLBCL, with an enrichment of certain immune cell types and relative depletion of others. Specifically, EBV(+) DLBCL demonstrated a significant predominance of CD8(+) T-cells (p=0.01), with a relative predominance of monocytes (p=0.11), neutrophils (p=0.08), and NK cells (p=0.11). In contrast, B-cells (p=0.07) and CD4(+) T cells (p=0.63) were relatively depleted compared to EBV(-) DLBCL (Figure 5A). These findings were supported by immunohistochemical staining for B-cells (CD20), T cells (CD3), and NK cells (CD56) in representative cases of EBV(+) and EBV(-) DLBCL cases (Figure 5B and 5C, respectively). EBV(+) tumors showed an inflammatory background enriched in CD8(+) T cells and NK cells, while EBV(–) cases exhibited higher B-cell content.

4. Discussion

EBV(+) DLBCL is more clinically aggressive, resulting in worse overall survival and treatment response compared to EBV(-) disease [1,2,3]. However, there is no targeted therapy or prognostic biomarker for EBV(+) DLBCL. Few studies have investigated the epigenomic heterogeneity of EBV-mediated DLBCL, with most studies focused heavily on overall DLBCL pathogenesis and non-viral disease mechanisms [59,60,61,62,63]. The epigenetic landscape of B-cell lymphomas is increasingly being recognized as crucial in disease pathogenesis and therapeutic targeting [64,65,66,67]. EBV epigenetically modifies the host cellular genome through selective DNA methylation, potentially promoting lymphomagenesis [10,11]. To better understand the influence of EBV on the DNA methylome of DLBCL, this study provides a systematic epigenetic characterization of EBV(+) DLBCL in relation to EBV(-) DLBCL and controls. Our results indicate that EBV(+) DLBCL exhibits a distinct DNA methylome from EBV(-) DLBCL, possibly influencing the observed clinical heterogeneity in EBV(+) tumors while also revealing therapeutic and prognostic biomarkers of disease.

A total of 2,114 differentially methylated genes, corresponding to 1,557 cancer-associated genes, were identified in EBV(+) DLBCL compared to EBV(-) DLBCL, defining a biologically distinct DNA methylome. This trend was also observed in the DLBCL-versus-controls analysis, where DLBCL cases exhibited global hypomethylation across all probe sites relative to controls. Global DNA hypomethylation is a well-documented feature of tumors compared to normal tissue and is considered a driver of tumorigenesis by enabling transcriptional activation of oncogene promoters, promoting genomic instability, and increasing tumor heterogeneity [19]. In DLBCL specifically, hypomethylation relative to normal B-cells has been reported and can serve as both a diagnostic and prognostic biomarker, capable of distinguishing molecular subtypes [21]. The study by Chambwe et al. demonstrated that increased DNA methylation variability in DLBCL correlates with worse clinical outcomes and overall survival. This variability reflects a loss of characteristic bimodal methylation distribution seen in normal tissue, a phenomenon observed across multiple tumor types [21,68] Such epigenetic variability is closely linked to clinical outcome and tumor heterogeneity as part of tumorigenesis [68]. Our findings confirm the presence of epigenetic variability in all DLBCL cases relative to control tissue, as well as within EBV-defined subgroups, supporting the utility of DNA methylation patterns as a global biomarker of tumor heterogeneity.

While the genes in EBV(+) DLBCL were skewed towards hypomethylation, the overall methylome across all probe sites exhibited a hypermethylated state in EBV(+) compared to EBV(-) tumors. Furthermore, although the global methylome in DLBCL showed a hypomethylation bias toward controls, the subset of genes of oncologic significance was predominantly hypermethylated, with no significant hypomethylated genes detected in this group. These findings are consistent with several studies demonstrating EBV’s role in promoting global hypermethylation of the host genome during carcinogenesis [10,69]. Both in vitro and clinical studies have shown that EBV actively modifies its own genome and the host cellular genome to establish viral latency, evade immune surveillance, and persist in the host B-cell [10,11,64,69]. EBV-mediated lymphoblastoid transformation is thought to result from accumulated pathogenic mutations and epigenetic modifications that drive oncogenesis. Hypermethylation of promoter regions in key tumor suppressor genes, coupled with hypomethylation of oncogenes, has been cited as a critical epigenetic mechanism in lymphomagenesis [11,64]. In DLBCL, such differential methylation patterns can define molecular subtypes and, in some cases, dictate prognosis [21,63,65,70]. Our findings demonstrate that EBV-mediated DLBCL harbors both hyper- and hypomethylated states within its tumor methylome, suggesting potential drivers of aggressive lymphomagenesis compared to EBV(-) disease.

We have characterized distinct methylomes defining DLBCL from control tissue and EBV(+) from EBV(-) tumors. The differential methylation of oncogenic genes in DLBCL, overall, skewed towards a hypermethylated state. However, we have also identified a significant set of hypomethylated genes in EBV(+) DLBCL as well. These patterns support distinct tumor heterogeneity and differences in prognosis in viral-mediated DLBCL. Hypermethylated genes of DLBCL included both oncogenic and tumor suppressor genes, with several exhibiting both functions. Overall, however, there were more tumor suppressor genes that were hypermethylated in DLBCL relative to controls. For EBV(+) DLBCL, we identified a similar mix of oncogenes, tumor suppressor genes, and genes with both functions that were hypermethylated. The hypomethylated genes also exhibited a mix of oncologic function with a predominance of genes with both oncogenic and tumor suppressor functions. All genes identified across the compared groups correlated to significant oncologic function in hematolymphoid malignancy and other human cancers. These defining genetic patterns correlated with unique gene pathways. DLBCL showed enrichment for hypomethylated genes involved in Wnt/β-catenin and Cadherin signal transduction pathways, which are known to play complex roles in lymphomagenesis [22,23,24,25,26,27]. Upregulation of Wnt/β-catenin signaling can cause more aggressive clinical behavior in DLBCL by promoting the endothelial to mesenchymal transition [25,26]. The tumor suppressor gene, CDKN2A, was significantly hypermethylated in DLBCL relative to controls, suggesting downregulation and loss of function. The functional loss of CDKN2A is linked to dysregulated P53 expression leading to more aggressive lymphomagenesis and worse clinical outcome in DLBCL despite standard chemotherapy [28,29,30,31,32,33].

Both hypermethylated and hypomethylated oncologic genes were detected in EBV(+) DLBCL relative to EBV(-) tumors, reflecting the tumor heterogeneity of viral-mediated disease. Hypermethylated oncologic genes correlated to tumor suppressive genetic pathways including P53 feedback loops and TGF-beta signaling. Loss of normal P53 function is associated with DLBCL occurrence and worse overall prognosis even in the setting of standard R-CHOP chemotherapy [28,39]. Transforming growth factor beta (TGF-beta) signaling is highly conserved and important for organismal development, but is shown to be exploited by various human cancer types to promote tumorigenesis, immune system evasion, and microenvironment modification [71]. In its normal state, TGF-beta functions as a tumor suppressor in B-cells, promoting cryostasis and modulating cellular differentiation and apoptotic pathways [42,71]. However, in vitro studies show that EBV-infected B-cell lymphoma cell lines dysregulate TGF-beta receptor signaling making the cells impervious to the anti-tumor effects of endogenous TGF-beta [40,72]. Further, these studies link the dysregulation of TGF-beta signaling with constitutive activation of the P38 MAPK oncogenic pathway, which we also found to be enriched in the hypermethylated gene set of EBV(+) DLBCL [40]. Interestingly, the IFN-gamma pathway was also enriched, suggesting functional suppression of its immunogenic anti-viral effects in EBV-mediated lymphomagenesis [45].

Finally, we highlighted the HDAC10 gene as significantly hypermethylated and potentially suppressed in EBV(+) DLBCL relative to EBV(-) disease. The HDAC10 gene encodes histone deacetylase 10, a class II family, which has crucial roles in epigenetic regulation. Depending on cellular context, HDAC10 can induce both oncogenic and tumor suppressive effects across an array of human cancer types. In B-cell lymphoma, it can exert tumor suppressive effects in concert with the dysregulation of the TGF-beta pathway [38,73]. The hypermethylation of this gene in EBV(+) DLBCL implies suppression of the HDAC10 pathway and loss of inherent tumor suppressive function.

Several pathways were identified significantly enriched in the hypomethylated gene set in EBV(+) DLBCL compared to EBV(-) DLBCL. These pathways included prominent oncogenic signaling like the MAPK cascade, DNA replication, and RAS pathway, which are highly represented across human cancer types [43,56,57]. In addition, the known oncogenes FLT3, HDAC6, NPM1, and BTK were significantly hypomethylated in EBV(+) DLBCL, suggesting activation [49,50,51,52,53,54,55] In particular, Bruton’s tyrosine kinase (BTK) is highlighted as significantly hypomethylated in EBV(+) DLBCL. BTK is a well-characterized and important non-receptor tyrosine kinase that functions in B-cell signaling pathway. Abnormal upregulation of BTK can turn on its oncogenic function in promoting B-cell proliferation, activation, survival, and differentiation [54,58]. BTK inhibitors are powerful targeted therapies used in an array of B-cell malignancies, including DLBCL [53,55]. BTK activates a signaling cascade which involves the RAS pathway and downstream transcription initiation of MYC oncogene [54].

The immune cell microenvironment of DLBCL plays a critical role in tumor aggression and response to therapy. Inference of immune cell composition in EBV(+) and EBV(-) DLBCL was made using the global DNA methylation data. EBV(+) DLBCL exhibited a significant predominance of cytotoxic CD8(+) T-cells relative to EBV(-) DLBCL (p=0.01), as well as a relative predominance in neutrophils, monocytes, and NK cells comprising the background inflammatory microenvironment. In contrast, EBV(-) DLBCL showed a relative predominance in B-cells and CD4(+) T-cells. These findings were confirmed using immunohistochemical stains for B-cells (CD20), T-cells (CD3), and NK cells (CD56). This unique tumor microenvironment composition of EBV(+) DLBCL with a relative increase in inflammatory cells, CD8(+) T-cells, and reduced B-cells is confirmed by multiple studies [74,75,76]. The “immune-rich” environment of EBV(+) DLBCL is observed across immunodeficiency states and is associated with the activated B-cell (ABC) phenotype. The cellular composition includes immunologically exhausted cytotoxic T-cells and NK cells and the upregulation of immune checkpoint proteins (PD-L1) and associated monocytes/macrophages [74,75]. These constellation of factors all lead to increased tumor heterogeneity, an immunosuppressive profile, and worse clinical prognosis in EBV(+) DLBCL across various states of immunodeficiency [74,76].

While this study represents a novel characterization of EBV-mediated DLBCL using global methylation analysis, there are several limitations that should be noted. First, while this is a single-institutional pilot study, the cohort size of EBV(+) DLBCL is small (n=9), limiting comparability of the findings with EBV(-) tumors. Second, the nature of DLBCL as a systemic disease with limited clinical sampling of diagnostic tissue specimens makes obtainment of control tissue from the same patient very difficult in this retrospective approach. In turn, we obtained EBV(+)/(-) normal lymph node/tonsillar tissue as well as comparison cases of other EBV(+) B-cell lymphoproliferative disorders. We acknowledge that the control tissue from different patients limits comparability of the DLBCL findings. Lastly, the global EPIC methylation array employed by this study, although comprehensive, has its limitations in understanding more granular nuances of epigenetic change between compared datasets. Our study is a pilot initiative with the intent to first characterize the overall methylome of EBV-mediated DLBCL. Future, more extensive studies are necessary on a larger clinical cohort utilizing an integrative analysis of epigenetic and whole transcriptomic approaches. This will help elucidate specific genetic pathways and epigenetic alterations that may lead to diagnostic and prognostic biomarker identification in EBV(+) DLBCL.

5. Conclusions

Using a global DNA methylation array, we found that the methylome of EBV(+) DLBCL across immunodeficiency states is epigenetically distinct from EBV(-) disease, and that DLBCL as a whole, is epigenetically distinct from control tissue cases. Both EBV(+) DLBCL and all DLBCL cases exhibited distinct patterns of hyper- and hypomethylation, which corresponded to specific gene pathways with complex roles in lymphomagenesis. EBV(+) DLBCL demonstrated greater hypermethylation across all probe sets compared to EBV(-) tumors, with enrichment of hypermethylated genes involved in tumor-suppressive pathways such as P53 feedback loops and TGF-beta signaling. The immune cell microenvironment of EBV(+) DLBCL, inferred from methylation data, displayed an immune-rich yet immunosuppressive profile, characterized by increased cytotoxic T cells and NK cells relative to the neoplastic population. These findings support the conclusion that EBV(+) DLBCL, across various immunodeficiency settings, harbors an epigenetically distinct methylome that may contribute to the observed tumor heterogeneity and clinical aggressiveness of this disease. Larger studies are warranted to more comprehensively characterize the methylation landscape of EBV-mediated DLBCL, with the goal to identify targetable diagnostic and prognostic biomarkers.