Extranodal NK/T-cell lymphoma (ENKTL) is a rare extranodal non-Hodgkin lymphoma (NHL) that primarily occurs in the upper aerodigestive tract and has an aggressive presentation, with locoregional invasion in the nasopharynx causing necrosis, hemorrhage, and impingement on anatomic structures including the orbits [1,2,3]. The incidence of ENKTL is much higher in East Asia and Latin America than in Europe and North America, representing up to 15% of all NHL diagnoses, likely due to underlying geodemographic differences in human leukocyte antigen (HLA) genes and genetic susceptibility [4,5,6,7]. Outcomes for ENKTL are generally poor, with 5-year overall survival (OS) rates of approximately 50% when treated with asparaginase-based multi-agent combination chemotherapy regimens such as SMILE [8,9] but as low as 25–28% in high-risk patients as assessed by various prognostic indices of NK lymphoma [10,11]. The majority of patients with ENKTL present with limited-stage, localized disease, in which outcomes are better, with 5-year and 10-year OS rates of 64–89% and 57%, respectively, utilizing combination chemotherapy and radiation [12,13,14]. However, patients with relapsed or refractory disease have dismal outcomes, with median survival of less than 12 months [15,16]. Despite advances in the frontline management of ENKTL with combination chemoradiotherapy approaches [14], there is a pressing need for novel and rational therapies, informed by the molecular biology of ENKTL, particularly for patients with relapsed/refractory disease, for patients who cannot tolerate intensive frontline induction therapy, and for patients with high-risk disease features and advanced-stage disease.

Historically, the pathobiology of ENKTL and its molecular drivers have been poorly understood. Gene rearrangement studies suggest that the majority of ENKTL cases are of NK cell rather than T cell lineage [17,18,19]. Although there are no apparent differences in clinical presentation or survival between NK or T cell lineages [20], there is emerging evidence that different signaling pathways are constitutively activated in each cell lineage, reflecting a complex and heterogeneous molecular driver landscape of ENKTL [19]. Classical descriptions of ENKTL oncogenesis by canonical T and NK cell signaling have focused on the ubiquitous presence of EBV in ENKTL tumor cells, and it is posited that EBV infection induces the overexpression of the transmembrane oncoprotein LMP1 in infected NK and T cells, resulting in ligand-independent activation of the NF-κB and MAPK signaling pathways [18,19,21]. These signaling pathways, with loss of tumor suppressor genes on chromosome 6q21, as is commonly observed in ENKTL, cause tumor cell proliferation and avoidance of apoptosis by regulation of c-MYC and survivin and also upregulate cell surface expression of PD-L1 to promote escape from immune surveillance [18,22]. However, ENKTL cells do not experience immortalization in response to EBV infection, as is seen in B-cell lymphomas, but rather become sensitized to proliferative cytokines such as IL-2, suggesting that other molecular alterations must occur in a multistep process for EBV-infected NK or T cells to undergo malignant transformation [23,24]. Recent advances in somatic next-generation and whole-genome sequencing have provided emerging evidence that ENKTL is characterized by genomic aberrations in multiple signaling pathways, including broad epigenetic and gene methylation changes [17,19,23,25].

It is increasingly observed that epigenetic modifiers are the largest group of mutated genes in ENKTL, with one meta-analysis of nine next-generation sequencing (NGS) studies finding that 25% of identified mutations in ENKTL affected epigenetic regulators [19]. Epigenetic aberrations play a critical role in tumorigenesis by silencing tumor suppressor genes and altering the regulation of oncogenes. A cohesive understanding of how epigenetic changes drive the ENKTL clone is starting to emerge [32].

In many types of cancer, promoter regions of tumor suppressor genes are frequently hypermethylated during carcinogenesis, resulting in transcriptional silencing via the recruitment of histone deacetylases (HDACs) and repressive chromatin formation [32]. Comparison of ENKTL tumor cells and normal NK cells with methylation assays have demonstrated global promoter hypermethylation and gene silencing in ENKTL, with decreased mRNA transcription of 95 putative tumor suppressor genes, including BCL2L11 (BIM), DAPK1, PTPN6 (SHP1), TET2, SOCS6, and ASNS with known functions [26]. Specifically, BIM sensitizes NK cell lines to chemotherapy-induced apoptosis, DAPK1 mediates p53-dependent apoptosis, SHP1, and SOCS6 inactivate the JAK/STAT signaling pathway, and TET2 silencing may contribute to early global hypermethylation in ENKTL, so it is easily understood how hypermethylation of multiple tumor suppressor promoters may create additive effects leading to tumor growth and malignant transformation [26]. These findings have been confirmed in other studies, with the identification of additional tumor suppressor genes, including DLC1 involved in RAS signaling [27], PRDM1 involved in apoptosis and promotion of NK cell growth in cooperation with activating STAT3 mutations [28,29], PTPRK involved in JAK/STAT signaling and apoptosis [30], HACE1 involved in apoptosis [31], as well as several other genes (P73, hMLH1, CDKN2A, CDKN2B, RARβ, PCDH10, DLEC1, CADM, DAL1) with currently unknown functions [32].

A recent genome-wide DNA methylation and transcriptomic study by Mundy-Bosse et al. (2022) demonstrated that ENKTL cells likely represent a malignant transformation from NK cells normally present in mucosal tissues, with ubiquitous and profound DNA hypermethylation driven by EBV infection causing the arrest of normal NK cell differentiation [17]. In this study, patient-derived xenograft (PDX) mouse models of ENKTL were treated with the DNA hypomethylating agent 5-azacytidine, which resulted in a significant decrease in tumor burden and an increase in OS compared to control mice. In addition, 5-azacytidine treatment resulted in a global reduction in DNA methylation and the emergence of mature NK cell markers, suggesting that treatment with hypomethylating agents may induce terminal differentiation of ENKTL cells, as is seen in patients with acute promyelocytic leukemia treated with all-trans retinoic acid [17]. These findings confirm the results of other studies, which have found that exposure of ENKL cell lines to hypomethylating agents such as 5-azacytidine or decitabine restores the expression of these putative tumor suppressors [26,27,30,146]. In a phase I study, one patient with ENKTL responded to a combination of 5-azacytidine and romidepsin with profound demethylation of their tumor after treatment [147]. There is a strong pre-clinical rationale for the treatment of ENKTL with hypomethylating agents, although there may be discordant effects from other therapeutic agents. For example, treatment with 5-azacytidine upregulates genes associated with immunoregulatory functions, suggesting possible synergy with immunotherapy agents [17], while there is concern that hypomethylating agents may desensitize ENKTL cells to asparaginase by upregulating ASNS and also lead to lytic reactivation of EBV [26,148].

Previous studies of ENKTL have revealed mutations in genes that regulate epigenetic changes to the genome, including genes responsible for chromatin remodeling, histone acetyltransferases and methyltransferases, and DNA demethylases [32,149]. One of the most frequently mutated genes in 17–32% of ENKTL cases is BCOR [28,40,41,42], a BCL-6 interacting corepressor which is involved in the epigenetic modification of histones via HDACs and which is suggested to be a tumor suppressor given frequent loss-of-function mutations seen in ENKTL [19,41] and enhanced cell proliferation and IL-2 production when BCOR is silenced [42]. MLL genes are also commonly mutated in up to 19% of ENKTL cases, specifically MLL2 (KMT2D) and MLL3 (KMT2C), which are involved in histone methylation and may be tumor suppressors in B-cell lymphomas, although the role of these epigenetic regulators specifically in the tumorigenesis of ENKTL is unknown [19,28,32,43]. Other commonly mutated epigenetic regulatory genes in ENKTL include TET1/TET2, EP200, ASXL3, CREBBP, and ARID1A, with yet unknown functions in ENKTL biology [19,32,44].

An epigenetic regulator gene with documented oncogenic function in ENKTL is EZH2, a histone methyltransferase that is not mutated in ENKTL but rather is aberrantly overexpressed and promotes NK tumor cell growth independently of its histone methyltransferase activity [33]. This non-canonical activity of EZH2 is likely due to its direct phosphorylation by JAK3, which converts EZH2 from a gene repressor to a transcriptional activator of genes involved in cell proliferation [34]. The JAK/STAT pathway is likely an upstream regulator of EZH2, as evidenced by studies demonstrating that treatment of ENKTL cells with the JAK3 inhibitor tofacitinib reduced EZH2 expression [38] and decreased ENKTL cell growth [34]. EZH2 is a known druggable target, with an FDA-approved EZH2 inhibitor (tazemetostat) for EZH2-wild type and EZH2-mutated follicular lymphoma [35]. However, treatment of ENKTL PDX mice with tazemetostat alone as well as in combination with 5-azacytidine did not improve survival in one study [17] and was ineffective in ENKTL cell lines [34], suggesting that inhibition of the EZH2 methyltransferase catalytic site does not affect its non-canonical and enzyme-independent oncogenic functions. EZH2 inhibitors with novel mechanisms of action, such as 3-deazaneplanocin-A (DZNep), which disrupts the metabolism of methyl donors required for histone methylation by EZH2 and also accelerates proteasomal degradation of EZH2, have demonstrated enhanced lymphoma cell apoptosis compared to tazemetostat [36,37,38].

EZH2 inhibitors are a rational drug target in ENKTL, although there is pre-clinical evidence that combination with agents targeted at its upstream regulators, such as JAK/STAT inhibitors, may also cause ENKTL cell death in patients with high EZH2 tumor expression [34]. Further, targeting downstream target genes of EZH2, such as cyclin D1, may also be efficacious, with one study demonstrating that combined inhibition of upstream JAK with ruxolitinib and CDK4/6 with ribociclib produced synergistic inhibition of ENKTL cell growth [39].

Large-scale somatic sequencing of ENKTL has enabled the description of new biomarkers which are both prognostic and predictive, although large-scale studies will be required for clinical validation. Mutations in MLL2 (KMT2D) and TET2 were associated with inferior prognosis in ENKTL [43], and overexpression of EZH2 was associated with higher tumor cell proliferation, advanced stage, and inferior survival [38]. Further, the presence of KMT2D mutations in circulating tumor DNA (ctDNA) was prognostic and correlated with total metabolic tumor volume, suggesting that serial measurements of ctDNA could be used to monitor treatment response and the presence of residual disease [128]. MicroRNAs (miRNAs) are short noncoding RNA sequences that inhibit the expression of target genes by suppressing translation or promoting mRNA degradation, and many have been found to be overexpressed in ENKTL as putative oncogenes as well as prognostic biomarkers [22]. For example, elevated plasma levels of miR-221 are associated with inferior overall survival [129], and miR-155 is associated with disease response [130].

Predictive biomarkers are particularly important for ENKTL to promote maximal response to therapy for patients with an aggressive and difficult-to-treat malignancy. In a study of 17 patients with ENKTL, patients with methylated PTPRK tumors who were treated with the SMILE protocol had significantly worse overall survival and a trend towards inferior disease-free survival [30]. In addition, ASNS expression is strongly correlated with cell survival in response to asparaginase treatment, which may serve as a clinically useful biomarker to determine which patients will respond to asparaginase-containing chemotherapy or which patients with low ASNS expression may be treated with lower doses of asparaginase to avoid toxicity [26]. miRNAs have been identified as predictive biomarkers with putative therapeutic targets; for example, a novel inhibitor of miR-155 induced apoptosis in ENKTL cell lines as well as xenografts by downregulation of STAT3 and VEGF signaling pathways and upregulation of the pro-apoptotic BRG1 [45]. High expression of SNHG12, a long noncoding RNA (lncRNAs) sequence which is overexpressed in ENKTL, conferred cisplatin resistance in ENKTL cells [131] and also may be responsible for multidrug resistance in ENKTL owing to its contribution to P-glycoprotein overexpression, which is a known mechanism for anthracycline resistance in ENTKL [150]. Another lncRNA, brain cytoplasmic RNA 1 (BCYRN1), is also overexpressed in ENKTL and is associated with inferior PFS as well as resistance to asparaginase [132].