EBV-Positive versus EBV-Negative Nasopharyngeal Carcinoma: Molecular Subtypes, Genomic Landscapes, Immune Microenvironments, and Therapeutic Implications

Nasopharyngeal carcinoma (NPC) is a geographically heterogeneous epithelial malignancy whose biology, prognosis, and therapeutic responsiveness are fundamentally stratified by Epstein–Barr virus (EBV) status. While EBV-positive disease predominates in endemic regions of Southern China and Southeast Asia, EBV-negative NPC is more prevalent in Western non-endemic cohorts, where human papillomavirus (HPV)-associated and HPV/EBV double-negative tumors emerge with distinct genomic and clinical features. Recent advances in whole-genome sequencing, epigenomic profiling, single-cell transcriptomics, and immunotherapy trials have revealed that EBV status is not merely a diagnostic marker but a determinant of molecular architecture, immune ecosystem composition, and therapeutic vulnerability. This review synthesizes current evidence to offer clinically actionable insights for oncologists, pathologists, and translational researchers.

EBV Status and the Molecular Taxonomy of NPC

The World Health Organization (WHO) recognizes three NPC histologic subtypes: keratinizing squamous cell carcinoma (typically EBV-negative), non-keratinizing carcinoma (typically EBV-positive), and basaloid squamous cell carcinoma. EBV-positive NPC overwhelmingly presents as non-keratinizing undifferentiated carcinoma in endemic regions and accounts for over 90% of undifferentiated NPC in high-risk populations 3. In a United Kingdom non-endemic cohort, 67% of tumors were EBV-positive, 18% HPV-positive, and 14% EBV/HPV-negative, with EBV and HPV showing mutual exclusivity within individual tumors 6.

EBV-positive NPC is not a monolithic entity. Multi-omics profiling integrating whole-genome bisulfite sequencing, ATAC-seq, and single-cell RNA sequencing has revealed two distinct epigenetic subtypes: HyperNPC, characterized by global genome-wide hypermethylation (approximately 80% of EBV-positive cases), and HypoNPC, featuring global hypomethylation (approximately 20%) 11. Both subtypes share EBV-specific differentially methylated regions (EBV-DMRs) enriched at CTCF and SMARCA4 regulatory elements, linking EBV latent infection to coordinated chromatin architectural reprogramming. Crucially, EBV latent programs drive a transcription factor binding switch from differentiation-associated KLF/SP family members to innate and adaptive immunity-related NF-κB and IRF families—actively reshaping tumor–immune crosstalk at the regulatory level 11.

A clinically important nuance is that approximately 15% of endemic NPC patients present with undetectable or very low plasma EBV DNA (≤20 copies/mL), often correlating with lower tumor volume and earlier-stage disease, while tissue EBV-encoded RNA in situ hybridization (EBER-ISH) may still confirm viral positivity. Stage-matched survival in this subgroup is comparable to EBV DNA-positive counterparts, cautioning against reliance on plasma EBV DNA alone for diagnosis or staging 1.

Comparative Mutational and Genomic Landscapes

EBV-positive NPC exhibits a genetically cohesive landscape centered on constitutive NF-κB activation. Whole-genome sequencing studies demonstrate that 90% of EBV-positive tumors harbor co-selected NF-κB-activating events via two mutually exclusive mechanisms: (1) overexpression of the EBV latent membrane protein 1 (LMP1) in approximately 33% of cases, or (2) somatic inactivation of NF-κB negative regulators (TRAF3, CYLD, NFKBIA, NLRC5, TNFAIP3) 2. The most significantly mutated gene is TP53, and the copy-number landscape is characterized by arm-level losses at 3p, 9p, 11q, 14q, and 16q, with gains at 1q, 3q, 7q, 8q, 12p, and 12q 23.

A therapeutically actionable finding is homozygous MTAP deletion co-occurring with CDKN2A loss in approximately 34% of EBV-positive tumors. In preclinical models, MTAP-deficient tumors demonstrate increased sensitivity to MAT2A inhibition, supporting MAT2A as a potential therapeutic target. The specific activity of individual MAT2A inhibitors in NPC should be interpreted based on available preclinical and clinical evidence 2. PI3K/AKT pathway aberrations represent an additional recurrent co-occurring feature 3.

Immune evasion is genomically encoded. Over 78% of EBV-positive tumors harbor somatic or EBV-driven immune-evading alterations: loss of type I interferon genes (IFNA1, IFNA2, IFNA8, IFNE) in ~16% of cases; alterations in antigen-presentation machinery (HLA class I/II, NLRC5, CIITA) in ~31%; and expression of EBV-derived BNLF2a, a transporter associated with antigen processing (TAP) inhibitor, detected in at least 13 of 70 sequenced tumors 2. Tumor mutational burden (TMB) is low in EBV-positive NPC (median ~0.95 mutations/Mb), reflecting viral-driven rather than carcinogen-driven oncogenesis, and TMB is not a reliable predictor of immune checkpoint inhibitor (ICI) response in this disease 327.

In contrast, EBV-negative NPC—particularly keratinizing subtypes in non-endemic cohorts—exhibits higher TMB, more frequent TP53 dysfunction, and HNSCC-like genomic features with conventional carcinogen-associated mutational signatures. HPV-positive subtypes introduce E6/E7-driven biology with distinct cell-cycle dysregulation. However, comparative whole-genome datasets directly contrasting EBV-positive and EBV-negative disease remain limited, and HypoNPC shows enrichment of APOBEC mutational signatures (COSMIC signature 2) that are distinct from HyperNPC, illustrating how methylation state and mutagenic process co-vary within EBV-positive disease itself 11.

Immune Microenvironment Differences

EBV-positive NPC presents a paradoxical immune ecosystem: highly inflamed yet functionally suppressed. Single-cell RNA sequencing of over 104,000 cells from 19 EBV-positive tumors identified global upregulation of interferon response pathways across tumor and stromal compartments, alongside malignant cells with an epithelial–immune "dual feature" state associated with poor prognosis. These dual-feature cells actively repress T-cell IFN-γ production and co-segregate with exhausted CD8+ tumor-infiltrating lymphocyte (TIL) phenotypes, illustrating that tumor-intrinsic immune suppression operates within an ostensibly pro-inflammatory niche 7.

A mechanistic link between EBV epigenomic reprogramming and T-cell dysfunction is provided by CD74 (the MHC class II invariant chain). In EBV-positive models, IRF1 binding at the CD74 promoter is confirmed, and CD74 expression correlates strongly with exhaustion marker signatures including HAVCR2 (TIM-3), TIGIT, CTLA-4, and LAG-3 (r² = 0.55 in bulk RNA-seq cohorts) 11. CD74–MIF/COPA signaling mediates tumor–immune communication and may underpin persistent T-cell dysfunction even in antigen-rich microenvironments.

Immunohistochemical analyses in endemic cohorts reveal that PD-L1 is expressed on tumor cells in approximately 69% of EBV-positive NPC (at ≥1% cutoff), and CD8+ TILs are universally present; however, PD-1 is expressed on infiltrating lymphocytes in only approximately 11% of tumors, suggesting that exhaustion biology in EBV-positive NPC extends beyond the canonical PD-1/PD-L1 axis 13. Importantly, low PD-L1 expression on immune cells (not tumor cells) independently predicts local recurrence and inferior disease-free survival after radiation-based therapy, while high PD-L1 on both immune and tumor cells is an independent favorable prognostic factor for overall survival—highlighting compartment-specific prognostic effects 14.

EBV-specific CD8+ TILs can further be phenotypically discordant with PD-1: in an EBV-driven lymphoepithelioma-like carcinoma model, most EBV-specific TILs lacked PD-1 despite abundant tumor PD-L1, raising concern that anti-PD-1/PD-L1 blockade may be insufficient to reinvigorate EBV-specific responses in this setting 8. In peripheral blood, EBV-positive advanced disease is associated with lower CD8% and higher NK cell frequency, with NK counts inversely correlating with viral load—suggesting complex innate–viral regulatory dynamics 9. EBV-encoded miRNA BART2-5p further promotes metastatic progression by silencing RND3 (a negative regulator of Rho/ROCK signaling), directly linking EBV-encoded non-coding RNAs to the aggressive phenotype 5.

Therapeutic Implications and Biomarker Strategy

Radiotherapy and platinum-based chemotherapy remain the standard of care for locoregionally advanced NPC regardless of EBV status. Intensity-modulated radiotherapy (IMRT) in EBV-positive NPC reduces plasma EBV DNA, CD8+PD-1+ T cells, and regulatory T cells (Tregs) concurrently, suggesting that radiation modulates both viral burden and the immunosuppressive milieu 12.

In recurrent or metastatic EBV-positive NPC, PD-1 inhibitors have established consistent but modest single-agent activity. The POLARIS-02 trial demonstrated an objective response rate (ORR) of approximately 20–24% with toripalimab monotherapy, with median overall survival of 15.1–17.4 months in heavily pretreated patients, comparable to pembrolizumab and nivolumab data 1527. Critically, PD-L1 expression status (PD-L1-positive 27.1% ORR versus PD-L1-negative 19.4%, P = 0.31) does not reliably distinguish responders, reinforcing that PD-L1 alone is an inadequate biomarker in EBV-positive disease 27.

Combination strategies have advanced significantly. The JUPITER-02 trial demonstrated that toripalimab combined with gemcitabine–cisplatin chemotherapy extended median progression-free survival (PFS) to 21.4 months versus 8.2 months with chemotherapy alone, with 3-year overall survival rates of 64% versus 49%—leading to FDA approval of toripalimab for NPC in January 2024 18. Penpulimab received FDA approval in April 2025 for non-keratinizing NPC based on the AK105-304 trial (median PFS 9.6 versus 7.0 months, HR 0.45, P < 0.0001) and the AK105-202 single-arm trial showing 28% ORR as monotherapy 19.

The modest monotherapy activity is mechanistically explained by multi-channel EBV-driven immune suppression: somatic antigen-presentation and interferon gene losses, BNLF2a-mediated TAP inhibition, and CD74-centered T-cell exhaustion wiring that engages LAG-3, TIM-3, and TIGIT beyond the PD-1 axis 211. These findings provide a strong rationale for multi-checkpoint combination strategies and for targeting CD74/MIF as a complementary suppressive axis.

EBV-targeted therapeutics represent a unique opportunity in EBV-positive disease. EBNA1 inhibitors, which disrupt EBV episome maintenance, show preclinical efficacy in EBV-positive NPC models, though cell-context variability in lytic switching remains a clinical translation challenge 10. Adoptive EBV-specific cytotoxic T-lymphocyte (CTL) therapy demonstrates durable benefit in EBV-positive disease (10-year overall survival of 44% in one cohort), particularly when deployed as consolidation after chemotherapy-induced cytoreduction in patients with oligometastatic disease and low baseline EBV DNA 27. These approaches have no applicable rationale in EBV-negative NPC, which lacks viral antigens as immunotherapy targets.

For MTAP-deleted EBV-positive NPC (approximately one-third of cases), MAT2A inhibition offers a precision-oncology strategy distinct from non-EBV-driven disease 2. EBV-negative NPC, aligning with conventional HNSCC genomic profiles, may be better served by HNSCC-derived targeted or immunotherapy paradigms.

Plasma EBV DNA is the most clinically actionable biomarker for EBV-positive NPC: it provides diagnostic confirmation when EBER-ISH is unavailable, guides prognostic stratification, and dynamically tracks treatment response and emerging progression. However, its false-negative rate of approximately 15% in endemic NPC requires tissue EBER-ISH as the gold standard for viral attribution 112.

Limitations and Future Directions

Current evidence carries important limitations. The vast majority of multi-omics, immunology, and clinical studies focus on endemic, EBV-positive NPC, with limited direct prospective comparisons to EBV-negative disease. Variable EBV testing methods (plasma DNA quantification, EBER-ISH, serology) and inconsistent PD-L1 antibody platforms and scoring criteria across studies complicate cross-cohort comparisons. Spatial immunology, single-cell atlases, and multi-omics epigenomic profiling have been conducted almost exclusively in EBV-positive disease, leaving the immune ecosystem of EBV-negative NPC largely uncharacterized.

Future priorities should include: (1) prospective multi-center cohorts spanning endemic and non-endemic geographies with standardized EBV testing and integrated multi-omics (genome, methylome, chromatin accessibility, transcriptome, proteome); (2) spatial immunology and single-cell transcriptomics in EBV-negative NPC to establish comparable immune atlases; (3) longitudinal liquid biopsy frameworks integrating plasma EBV DNA, circulating tumor DNA, and immune-cell profiling to enable adaptive treatment strategies; (4) biomarker-driven randomized trials testing multi-checkpoint regimens (anti-PD-1 plus anti-LAG-3/TIM-3/TIGIT), EBV-targeted agents, and MAT2A inhibitors in MTAP-deleted disease; and (5) development of multi-parameter predictive classifiers incorporating CD74 expression, exhaustion marker signatures, viral load, and genomic features for rational ICI and combination therapy selection.

Standardized Comparison Table: EBV-Positive versus EBV-Negative NPC