In this issue of Blood, Akazawa et al1 used an integrated approach based on multiomics, including methylation profiling, to identify 2 subsets of natural killer (NK)-cell type chronic active Epstein-Barr virus (CAEBV). These findings should lead to better prognostic determinations and potential treatment opportunities.
Systemic T/NK-cell type CAEBV is a rare but severe disease is included in the 5th edition of the World Health Organization classification of hematolymphoid tumors.2 It is characterized by chronic infection of T or NK cells by EBV. The prognosis is poor and unpredictable with some patients remaining clinically stable for years before suddenly developing fulminant hemophagocytic lymphohistiocytosis (HLH) or overt T/NK-cell lymphoma. Potentially because of the heterogeneous clinical presentation and clinical course, no tailored therapy has emerged.3,4 The only curative treatment remains allogeneic hematopoietic stem cell transplantation (allo-HSCT), which carries significant risks.5
In this study, the authors employed DNA methylation microarray, transcriptome, methylome, whole exome sequencing, whole genome sequencing, and single-cell RNA sequencing (scRNAseq) analyses to study the peripheral blood mononuclear cells of patients with NK-cell type CAEBV. They found distinct DNA methylation profiles in circulating EBV-infected cells that involved CpG island promoter methylation (CIMP) as previously described in other malignancies. They showed that patients with a CpG island hypermethylation phenotype (CIMP positive) in their circulating EBV-infected cells exhibited strong similarities to NK-cell lymphoma (ENKTL), including comparable methylation patterns, a higher mutational burden, and notably copy number alterations in the cancer-associated genes associated with a significantly poorer prognosis than patients with a negative CIMP. This suggests a prelymphomagenesis state in the subset of patients with positive CIMP, which correlated with a decreased overall survival (or increased progression to death) of these patients.
The authors further showed that the increased methylation status of tumor suppressor genes (such as TP53, CDKN1C, and LDOC1) led to their silencing, thus representing an overt neoplastic factor in EBV-infected cells. Mechanistically, the authors found that DNA hypermethylation occurred in regions associated with enhancer of zeste homolog 2 (EZH2) binding sites and trimethylation of histone H3 lysine 27 (H3K27me3), an inhibitory reversible histone modification. Interestingly, EZH2 expression is increased by an EBV infection. A comparison of CIMP-positive and CIMP-negative patients suggested that EZH2 upregulation by EBV caused reversible silencing of H3K27me3 in CIMP-negative patients that was switched in CIMP-positive patients to stable DNA hypermethylation and the silencing of tumor suppressor and antiviral genes.
Interestingly, somatic mutations in methylation-modifier genes, such as ARID1A and KMT2D, were detected in the EBV-infected cells of patients with a poorer prognosis and may be linked to or participate in the CIMP methylation profile or may be linked to the switch to stable gene silencing (in CIMP-positive patients). However, the relative contributions of methylation patterns vs somatic mutations to the prognosis remain unclear. Interestingly, these high-risk characteristics were observed even in patients under 10 years of age. This methylation profiling delineates a particular subset of patients with CAEBV with more prominent neoplastic features, thereby positioning them closer to ENKTL on the spectrum of EBV+ NK-cell lymphoproliferative diseases.
Several additive or synergic mechanisms could explain the varied features of CAEBV, including abnormal infection of T/NK cells, chronic persistence of infected cells that evade the immune system, and subsequent acute transformation with neoplastic features. In addition to neoplastic features related to the methylation profile, the authors additionally link abnormal methylation status to defective anti-EBV immunity; hypermethylation of antiviral genes, such as MX1 and MX2, correlates with decreased type-1 interferon response restricted to infected NK cells, as previously suggested by Luo et al.6
The effects of CpG hypermethylation on antiviral gene silencing and infected cell proliferation were reproduced in vitro and in vivo in patient-derived xenograft mice models by 5-azacytidine hypomethylating agents. This finding further connects methylation to intrinsic anti-EBV immunity and infected cell survival/proliferation, potentially opening new therapeutic avenues. Methylation status, regardless of origin, could thus be considered an additional pathophysiological factor in this disease that likely evolves through multiple steps, affecting both immune surveillance and cellular transformation. Differentiating such subsets of CAEBV could help stratify patients and identify which individuals should receive more aggressive treatment. This stratification may also guide therapy toward methylation-modifying agents, such as 5-azacytidine, as tested by the authors.
Finally, the authors performed single-cell RNA sequencing on a limited number of patient samples. To detect EBV-infected cells, the authors introduced a significant technical improvement. Traditional scRNAseq methods enable only the detection of polyA RNA, missing the polyU EBV-encoded small RNAs, which is the highest EBV viral RNA expressed in infected cells. Using this robust method of EBV detection, the number of different lymphocyte subsets that harbor EBV was much lower than previously described and was restricted to the natural targets of EBV (memory B cells) and tumoral NK cells with no evidence of infection in myeloid cells or other T-cell subsets, contrary to what was recently shown.7 Interestingly, the authors showed that EBV-infected NK cells exhibit a phenotype of tissue resident cells, supporting that infection of NK cells occurs in tissues like tonsils in which primary infection takes place.
This study primarily analyzed EBV+ NK-cell CAEBV. Because no obvious phenotypic differences have been established between T- and NK-cell systemic CAEBV, it remains to be evaluated whether these conclusions apply to other CAEBV subsets. For instance, hydroa vacciniforme has a poorer prognosis in Asian patients,8,9 and one could hypothesize that these cases are more closely related to γ-δ T-cell lymphoma based on methylation profiling.
Conflict-of-interest disclosure: The authors declare no competing financial interests.