The evolution of EBV DNA as a biomarker in nasopharyngeal carcinoma immunotherapy: a bibliometric and knowledge mapping analysis Authors: Yang Chen, Qicong Chen, Xiaosu Zou, Wenqian Nong, Taicheng Lu, Sifang Xie, Weiming Deng, Min Li, Guiping Lan, Gang Wang, Shenhong Qu Published: 2026 ABSTRACT BackgroundThe treatment response to immunotherapy for nasopharyngeal carcinoma (NPC) exhibits significant interindividual variability. As a biomarker closely associated with NPC, Epstein-Barr virus (EBV) DNA holds substantial importance in this research context. However, there is still an insufficient comprehensive assessment of the current research status and development trends of EBV DNA as a biomarker for NPC immunotherapy in the existing literature.MethodsIn this study, we conducted a bibliometric analysis of 164 original English articles and reviews on EBV DNA as a biomarker in NPC immunotherapy, utilizing the WoSCC database along with CiteSpace and Bibliometrix software tools. The analysis and visualization were performed from the following aspects: publication trends, country/region distribution, keyword co-occurrence networks, patterns of co-authorship and co-citation, and the evolution of cited journals and themes.ResultsSince 2017, research in this area has developed rapidly, with China and the United States being the two main contributors. The primary research hotspots include PD-1 inhibitors, antitumor activity, and exploration of molecular mechanisms. The field showed a clear shift from traditional themes centered on tumor burden, recurrence, radiotherapy, and prognosis toward immunotherapy-related themes, including PD-1, clinical trials, and dynamic monitoring.ConclusionsEBV DNA is increasingly viewed as a dynamic biomarker that integrates viral biology, tumor burden, and treatment response in NPC. Plasma EBV DNA is mainly derived from apoptotic or necrotic EBV-positive tumor cells, making its baseline level a surrogate of systemic tumor burden. Its kinetic changes during immunotherapy indicate tumor control, while fluctuations or elevations reflect combined effects of tumor destruction, viral reactivation and host immunity. Its value in prognostic stratification and dynamic monitoring has been relatively well established. Broader clinical use will require assay standardization and prospective validation.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12985-026-03197-z. Introduction Nasopharyngeal carcinoma (NPC) is one of the most prevalent malignancies of the head and neck, particularly endemic in southern China and Southeast Asia. It is closely associated with Epstein-Barr virus (EBV) infection [1]. In recent years, immune checkpoint inhibitors (ICIs)—particularly PD-1/PD-L1 inhibitors—have substantially reshaped the therapeutic paradigm for recurrent or metastatic NPC (R/M NPC). Several high-quality phase III clinical trials have established PD-1 inhibitor–based regimens as the new standard first-line treatment, superseding conventional chemotherapy alone in R/M NPC [2]. A representative example is the multicenter CAPTAIN-1st trial, conducted across 28 hospitals, which suggested that camrelizumab combined with gemcitabine–cisplatin chemotherapy significantly improved long-term survival, with a 37.8% 2-year overall survival rate among patients with R/M NPC [3]. Despite clear advances in clinical treatment, responses to ICIs remain heterogeneous, and only a subset of patients achieve durable clinical benefit. Meta-analytic evidence indicates that combining PD-1 inhibitors with anti-angiogenic agents yields an objective response rate of merely 49% in R/M NPC, and the majority of patients ultimately develop resistance to immunotherapy, resulting in suboptimal outcomes [4]. A major unmet need in immuno-oncology is to identify patients who are most likely to benefit from ICI therapy. Responses to immune checkpoint blockade are often delayed and may not conform to the radiographic patterns typically seen with cytotoxic therapy. Clinical interpretation is therefore challenging. This difficulty is further increased by atypical response patterns, including pseudoprogression and early clinical progression, which can be difficult to distinguish from true disease progression. Furthermore, reliable real-time biomarkers for monitoring treatment sensitivity remain limited. This makes it difficult to distinguish true disease progression from transient immune-mediated changes in tumor burden. As a result, timely adaptive therapeutic interventions may be difficult to implement. Furthermore, the lack of reliable real-time biomarkers to monitor treatment sensitivity hinders the timely differentiation between genuine disease progression and transient, immune-mediated fluctuations in tumor burden, thereby limiting opportunities for adaptive therapeutic interventions. Current biomarker strategies remain insufficient to address these challenges. PD-L1 expression, although widely studied, shows inconsistent predictive performance and is influenced by spatial and temporal heterogeneity. Tumor mutational burden appears to be less informative in NPC, which is typically characterized by a relatively low mutational load. Other immune-related signatures, including tertiary lymphoid structures [5] and miRNA signatures [6], provide biological insights but are not readily applicable in routine clinical practice. Collectively, these limitations highlight a unmet need for robust, dynamic, and clinically accessible biomarkers in NPC immunotherapy. Against this background, EBV DNA has increasingly emerged as a major focus in NPC immunotherapy research. Plasma EBV DNA, as a circulating free viral nucleic acid released by tumor cells, has the advantages of being non-invasive, sensitive and dynamically monitorable. It has gradually entered the clinical practice of NPC from basic research and has become an important indicator for disease screening [7], staging assistance [8], therapeutic effect evaluation [9] and recurrence prediction [10]. With the widespread application of immunotherapy in recurrent/metastatic nasopharyngeal carcinoma (R/M NPC), the value of EBV DNA is no longer limited to traditional prognostic prediction. Evidence has increasingly linked baseline EBV DNA levels and early on-treatment changes with immunotherapy response and survival benefits. These findings provide a basis for identifying patients who may benefit from immunotherapy. They may also help guide the optimization of treatment strategies. One prospective study showed that, among patients with NPC receiving anti-PD-1 monotherapy, higher baseline EBV DNA levels were associated with shorter progression-free survival (PFS) and OS, whereas a limited early decline in EBV DNA during treatment was similarly associated with reduced OS and a lower likelihood of clinical benefit [11]. Nevertheless, the significance of EBV DNA in NPC immunotherapy has not yet been fully elucidated. On the one hand, changes in EBV DNA during treatment may reflect tumor cell clearance and a reduction in disease burden. On the other hand, its dynamic fluctuation may also be influenced by multiple factors, including alterations in host immune status [12], non-tumor-derived viral signals [13], and variability in assay platforms. In addition, substantial heterogeneity persists across studies with respect to specimen type, detection methodology, and threshold definition, all of which directly limit cross-study comparability and hinder clinical translation. Under these circumstances, conventional narrative reviews are insufficient to reveal the developmental trajectory of this interdisciplinary field. Bibliometric analysis can delineate the overall landscape of research output. It can also identify the intellectual base through co-citation analysis. In addition, keyword co-occurrence and thematic mapping can help trace shifts in research priorities. In this way, it becomes possible to clarify how a field gradually evolves from a collection of fragmented questions into a coherent research direction [14]. EBV DNA and NPC immunotherapy together constitute an interdisciplinary field spanning virology, oncology, liquid biopsy, and immunotherapy. The importance of undertaking a bibliometric analysis in this domain rests not only in exposing patterns in publication output, but also in providing a systematic perspective for clarifying the research trajectory, capturing shifts in research focus, and identifying barriers to clinical translation. Although Wang et al. [15] have previously reviewed the research hotspots and developmental trends in NPC immunotherapy. Building on this foundation, the present study further focused on the research progress of EBV DNA as a biomarker for NPC immunotherapy. By retrieving relevant publications published between 2000 and 2025 and applying an integrated approach combining bibliometric analysis with knowledge mapping, the evolutionary trajectory of research in this field was systematically characterized. In addition, from the perspective of translational medicine, the clinical value, biological significance, and practical challenges of EBV DNA in the context of NPC immunotherapy were further discussed. Materials and methods Data sources The dataset used in this study was from Web of Science Core Collection (WoSCC) database, the literature search was conducted in 27th June 2025. The literature search strategy was as follows: [(((TS = ("nasopharyngeal cancer" OR "nasopharyngeal carcinoma" OR "NPC" OR "nasopharynx cancer")) AND TS = ("immunotherap*" OR "immune checkpoint inhibitor*" OR "PD-1 inhibitor" OR "PD-L1 inhibitor" OR "CTLA-4 inhibitor" OR "cancer vaccine*")) AND TS = ("EBV DNA" OR "Epstein-Barr virus DNA" OR "plasma EBV DNA" OR "EBV" OR "Epstein-Barr virus")) AND TS = ("biomarker" OR "diagnos*" OR "prognos*" OR "predict*" OR "monitor*" OR "response prediction")]. Inclusion and exclusion criteria To ensure the transparency and reproducibility of the research process, a two-stage literature screening process was used. Firstly, the original text data exported from the WoSCC database was imported into CiteSpace software, and the data cleaning algorithm built into the software was used to automatically deduplicate the data. The system identified and removed the redundant records that were repeatedly included, and then manually checked. Inclusion criteria were as follows: 1. The research topic must be clearly related to NPC, EBV DNA and immunotherapy. 2. The literature type is limited to Article or Review. 3. Articles published in English only. Exclusion criteria were as follows: 1. Trials reporting EBV-DNA only in the background but not as a core analysis variable were excluded. 2. Meeting Abstracts, Editorials, Letters to the editor, Case Reports, and book chapters that could not be used for analysis due to lack of a complete citation network were excluded. The retrieved records were initially screened on the basis of titles and abstracts, followed by full-text review. Any discrepancies were resolved through discussion until consensus was reached. A total of 164 articles were included in the formal analysis. Bibliometric analysis CiteSpace (v6.2.R4) and Bibliometrix (v4.2.1) were used for bibliometric analysis and visualization. Citation networks, dual-map overlays, keyword co-occurrence, thematic evolution, and multiple correspondence analysis were performed to identify research hotspots and knowledge gaps in NPC immunotherapy and EBV DNA research. Parameter definition The clustering results are evaluated based on the module value Q and the average silhouette value S. Here, the Q value is used to measure the significance of the network clustering structure. The larger the Q value, the clearer the boundaries between different clusters; when Q > 0.3, it is generally considered that the clustering structure is significant. The S value is used to assess the homogeneity within the same cluster. The closer the S value is to 1, the higher the consistency within the cluster; generally, when S > 0.5, the clustering results are considered reliable, and when S > 0.7, it indicates that the cluster has a high internal consistency. The specific calculation formulas for Q and S values are as follows [16, 17]:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Q=\frac{1}{2m}\sum_{ij}\left({A}_{ij}-\frac{{k}_{i}{k}_{j}}{2m}\right)\delta \left({c}_{i},{c}_{j}\right)$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$s(i)=\frac{b(i)-a(i)}{max\{a(i),b(i)\}},S=\frac{1}{n}\sum_{i=1}^{n}s(i)$$\end{document} In the dual map overlay of journals, the f value represents the frequency of citation paths, and the larger the f value, the greater the scale of knowledge flow carried by that path. A larger f value is visually represented by a thicker spline curve. The Z value represents the relative strength after standardization of this frequency. The higher the Z value, the more prominent the path is in the overall citation network. The calculation of the Z value is generally expressed as follows:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Z=\frac{x-\mu }{\sigma }$$\end{document} Results Publication trends A total of 164 publications were included in this study. As shown in Fig. 1A, the annual number of studies on EBV DNA and NPC immunotherapy increased steadily over time. Trend analysis revealed two distinct phases (Fig. 1B). From 2000 to 2016, publication output was low and slowly increasing, with an average of 2.23 articles per year. In contrast, from 2017 to 2025, publications rose markedly to an average of 14 articles per year, representing a 6.27-fold increase. Since 2021, publication numbers have remained consistently high, this trend fully reflects that EBV DNA has become the focus of research in the NPC immunotherapy.Fig. 1Annual literature trends on EBV DNA as a biomarker associated with immunotherapy for NPC. (A) Annual number of publications. (B) Annual growth trend of publications. The red curve represents the fitted trend line Annual literature trends on EBV DNA as a biomarker associated with immunotherapy for NPC. (A) Annual number of publications. (B) Annual growth trend of publications. The red curve represents the fitted trend line Country/region distribution analysis of publication output An analysis of publication output and proportional contribution across the 164 articles by country of research origin showed that 24 countries have contributed to this field, with the literature being concentrated mainly in Asia. Among them, China produced substantially more publications than any other country (Fig. 2A). The annual publication trends of the top five countries are shown in Supplementary Figure S1A. As illustrated in Fig. 2B, China and the United States occupied dominant positions in this field and maintained the closest collaborative relationship. The local collaboration network further showed close collaborative ties among China, the United States, and Japan. (Supplementary Figure S1B).Fig. 2Geographical Distribution and International Collaboration Map of EBV DNA as a Biomarker in NPC Immunotherapy Literature. (A) Literature Distribution Map. The number of publications from each country/region and their corresponding areas are labeled in designated zones, with color intensity reflecting publication volume. Gray areas indicate no publications. (B) Visualization of international collaboration. Red lines represent collaborative relationships between regions. Line thickness indicates collaboration intensity, while color intensity corresponds to publication volume. Gray areas denote countries/regions with no publications Geographical Distribution and International Collaboration Map of EBV DNA as a Biomarker in NPC Immunotherapy Literature. (A) Literature Distribution Map. The number of publications from each country/region and their corresponding areas are labeled in designated zones, with color intensity reflecting publication volume. Gray areas indicate no publications. (B) Visualization of international collaboration. Red lines represent collaborative relationships between regions. Line thickness indicates collaboration intensity, while color intensity corresponds to publication volume. Gray areas denote countries/regions with no publications Keyword co-occurrence analysis To delineate the core knowledge structure surrounding EBV DNA as a biomarker in NPC immunotherapy, keyword-based analyses were performed. The 20 most frequently used keywords are summarized in Supplementary Table S1, while Supplementary Figure S2 presents the frequencies and clustering patterns of the top 10 keywords in the form of a heatmap. In addition, keyword burst analysis (Table 1) revealed the temporal evolution of research priorities and may provide a useful reference for future investigations. Keyword co-occurrence analysis showed that nasopharyngeal carcinoma, Epstein-Barr virus, and immunotherapy occupied the center of the network and were closely associated with terms such as recurrent and chemotherapy (Fig. 3A). In the clustering analysis, a Q value of 0.7065 and an S value of 0.788 were consistent with good clustering validity and reliability (Fig. 3B). The clustering results further indicated that, in addition to clinically oriented themes such as PD-1, clinical trials, and therapy, mechanistic terms such as epitopes were also represented, suggesting that this field places parallel emphasis on mechanistic investigation and clinical translation.Table 1.Top 20 keywords with the strongest citation burstsFig. 3Visualization of Keyword Co-occurrence Analysis. (A) Keyword co-occurrence analysis. Each colored node represents an author keyword, with node size proportional to publication output. Discrete regions formed by nodes and links indicate associations between different keywords. (B) Keyword clustering analysis. Keywords clustered using the LLR algorithm, with differently colored nodes and edges representing distinct clusters. (C) Timeline visualization of EBV DNA biomarker-related keyword clusters for NPC immunotherapy. Horizontal lines of varying colors and their labels denote keyword clusters. Nodes on each line represent keywords, with their positions indicating the year of first publication for that keyword, forming a timeline of keyword cluster evolution Table 1.Top 20 keywords with the strongest citation bursts Top 20 keywords with the strongest citation bursts Visualization of Keyword Co-occurrence Analysis. (A) Keyword co-occurrence analysis. Each colored node represents an author keyword, with node size proportional to publication output. Discrete regions formed by nodes and links indicate associations between different keywords. (B) Keyword clustering analysis. Keywords clustered using the LLR algorithm, with differently colored nodes and edges representing distinct clusters. (C) Timeline visualization of EBV DNA biomarker-related keyword clusters for NPC immunotherapy. Horizontal lines of varying colors and their labels denote keyword clusters. Nodes on each line represent keywords, with their positions indicating the year of first publication for that keyword, forming a timeline of keyword cluster evolution The temporal evolution of keyword hotspots was evident. Early research focused primarily on traditional themes such as radiotherapy, chemotherapy, and recurrence, whereas more recent studies have shifted toward directions such as PD-1 and clinical trials (Fig. 3C). Keyword frequency analysis further showed that, beyond fundamental virological themes and disease entities, the literature has increasingly concentrated on clinical efficacy and trial methodology (Fig. 4A). As illustrated in Fig. 4B, major research-producing institutions such as Sun Yat-sen University have focused not only on EBV, but also on clinical application. Research has mainly focused on the antitumor activity of recurrent NPC immunotherapy. Multicenter clinical trials have also become increasingly prominent. This pattern suggests that the field may be moving from biomarker discovery toward biomarker-informed therapeutic investigation.Fig. 4Word cloud and Sankey diagram of keywords related to EBV DNA as a biomarker in NPC immunotherapy research, extracted using the Bibliometrix R package. (A) Keywords associated with EBV DNA as a biomarker in NPC immunotherapy research. Colors represent different keywords, with word size reflecting frequency of occurrence. High-frequency keywords include Epstein-Barr virus, antitumor-activity, cancer, and recurrent. (B) Sankey diagram showing the association between research institutions, countries, and keywords Word cloud and Sankey diagram of keywords related to EBV DNA as a biomarker in NPC immunotherapy research, extracted using the Bibliometrix R package. (A) Keywords associated with EBV DNA as a biomarker in NPC immunotherapy research. Colors represent different keywords, with word size reflecting frequency of occurrence. High-frequency keywords include Epstein-Barr virus, antitumor-activity, cancer, and recurrent. (B) Sankey diagram showing the association between research institutions, countries, and keywords Co-author and citation author analysis Chen, Qiu-Yan exhibited the highest publication output, while Chen YP was identified as the most cited author in this field (Supplementary Table S2). The author collaboration network revealed several core research teams with relatively concentrated collaboration (Fig. 5A). Institutional analysis indicated extensive international cooperation, spearheaded by Sun Yat-sen University and the National Cancer Centre Singapore (Fig. 5B). Sun Yat-sen University also maintained close connections with institutions such as the Chinese University of Hong Kong, highlighting its prominent role in this field (Fig. 5C). LLR-based keyword clustering further indicated that high-output institutions, represented by Sun Yat-sen University, primarily focused on biomarker validation and clinical translation, whereas the NCI emphasized mechanistic research (Fig. 5D).Fig. 5Visualization of author collaboration analysis. (A) Each colored node represents an author, with node size proportional to the number of published papers. Independent clusters formed by nodes and links reveal collaboration patterns among authors. Authors with over six published papers are shown. (B) Co-occurrence network analysis of institutions. The institutional collaboration network comprises 238 nodes and 496 links, with a network density of 0.0176. Sun Yat-sen University and the National Cancer Center Singapore were identified as core institutions. (C) Visualization of institutional collaboration reveals multiple research centers maintaining active cooperation within a highly interconnected academic network centered on Sun Yat-sen University. (D) Cluster analysis of institutional research themes. To explore key research areas, the LLR algorithm was applied for keyword clustering. Three major core clusters emerged: nasopharyngeal carcinoma patients, era, and metastatic pulmonary lymphoepithelioma-like carcinoma Visualization of author collaboration analysis. (A) Each colored node represents an author, with node size proportional to the number of published papers. Independent clusters formed by nodes and links reveal collaboration patterns among authors. Authors with over six published papers are shown. (B) Co-occurrence network analysis of institutions. The institutional collaboration network comprises 238 nodes and 496 links, with a network density of 0.0176. Sun Yat-sen University and the National Cancer Center Singapore were identified as core institutions. (C) Visualization of institutional collaboration reveals multiple research centers maintaining active cooperation within a highly interconnected academic network centered on Sun Yat-sen University. (D) Cluster analysis of institutional research themes. To explore key research areas, the LLR algorithm was applied for keyword clustering. Three major core clusters emerged: nasopharyngeal carcinoma patients, era, and metastatic pulmonary lymphoepithelioma-like carcinoma Citation journal analysis A bibliometric analysis of cited journals reveals research trends and academic influence within the field of EBV DNA in NPC immunotherapy. The most frequently cited journals are detailed in Supplementary Table S3, with the top ten journals recognized as leaders in their respective fields. The Journal of Clinical Oncology ranks first in citation frequency, followed by Clinical Cancer Research, The Lancet, and Cancer Research, all classified as Q1 journals. In terms of impact factor, The Lancet and The New England Journal of Medicine are the highest ranked. The citation network among these journals is depicted in Supplementary Figure S4. Furthermore, the dual-map overlay of journals (Fig. 6) illustrates the major knowledge flow and cross-disciplinary citation patterns in this field. Two primary citation paths were identified: journals in molecular biology and immunology are predominantly cited by journals in molecular biology and genetics, with Z = 2.53 and f = 636. Additionally, journals in medicine and clinical medicine are frequently cited by journals in molecular biology and genetics, with Z = 4.22 and f = 1004.Fig. 6Journal dual-mapping overlay diagram of EBV DNA as a biomarker in immunotherapy literature. Left nodes represent contained publications, while right nodes represent references cited within those publications. Labels denote disciplinary fields, and links represent citation pathways. Molecular biology and immunology journals consistently cite molecular biology and genetics journals, with a Z-value of 2.53 and f-value of 636; Medical and clinical medicine journals consistently cite molecular biology and genetics journals, with a Z-value of 4.22 and an f-value of 1004 Journal dual-mapping overlay diagram of EBV DNA as a biomarker in immunotherapy literature. Left nodes represent contained publications, while right nodes represent references cited within those publications. Labels denote disciplinary fields, and links represent citation pathways. Molecular biology and immunology journals consistently cite molecular biology and genetics journals, with a Z-value of 2.53 and f-value of 636; Medical and clinical medicine journals consistently cite molecular biology and genetics journals, with a Z-value of 4.22 and an f-value of 1004 Topic evolution and thematic map The evolution of research hotspots was examined through thematic evolution analysis, and potential future directions were inferred accordingly. From 2000 to 2020, the principal hotspots were concentrated on intensity-modulated radiotherapy, tumor-intrinsic characteristics, and EBV-related features of NPC, whereas from 2021 to 2025, the focus shifted toward antitumor activity, immunotherapy, gene expression, and biomarkers associated with poor prognosis (Fig. 7A). Thematic clustering analysis identified two main clusters. The blue cluster represented established themes related to radiotherapy and chemotherapy. The red cluster represented newer topics involving immunotherapy and virological mechanisms. This distribution suggests an evolution of the field from conventional treatment approaches to biomarker-informed precision immunotherapy (Fig. 7B).Fig. 7Thematic Evolution and Topic Map. (A) Keyword evolution analysis of EBV DNA in NPC immunotherapy from 2000 to 2025. Adjacent time periods sharing keywords are connected by streamlines, with each streamline’s width proportional to the number of shared keywords. Between 2000 and 2020, research focused on cancer, EB virus, and NPC. From 2021 to 2025, new topics such as antitumor activity, gene expression, PD-L1, and poor prognosis gradually emerged. (B) Topic Evolution Analysis. Each point represents a keyword, with distances reflecting co-occurrence frequency. The red cluster encompasses rapidly growing research areas like antitumor activity, immunotherapy, and EBV; the blue cluster covers traditional treatment themes including radiotherapy, chemoradiotherapy, and induction chemotherapy. (C) Thematic map. The first quadrant, identified as "Motor themes," represents the primary research frontiers within the field. These topics exhibit both high density and strong influence, suggesting substantial progress and extensive academic engagement. The second quadrant, denoted as "Niche themes," reflects areas with a well-established foundation and intensive internal development. However, their overall impact remains limited, indicating that despite active exploration, major breakthroughs have yet to emerge. The third quadrant, labeled "Emerging or declining themes," encompasses subjects in the early or transitional stages of research. Characterized by low density and limited influence, these themes are in a phase of conceptual refinement and empirical accumulation. Finally, the fourth quadrant, referred to as "Basic themes," includes fundamental research directions with promising practical relevance that merit deeper investigation in future studies Thematic Evolution and Topic Map. (A) Keyword evolution analysis of EBV DNA in NPC immunotherapy from 2000 to 2025. Adjacent time periods sharing keywords are connected by streamlines, with each streamline’s width proportional to the number of shared keywords. Between 2000 and 2020, research focused on cancer, EB virus, and NPC. From 2021 to 2025, new topics such as antitumor activity, gene expression, PD-L1, and poor prognosis gradually emerged. (B) Topic Evolution Analysis. Each point represents a keyword, with distances reflecting co-occurrence frequency. The red cluster encompasses rapidly growing research areas like antitumor activity, immunotherapy, and EBV; the blue cluster covers traditional treatment themes including radiotherapy, chemoradiotherapy, and induction chemotherapy. (C) Thematic map. The first quadrant, identified as "Motor themes," represents the primary research frontiers within the field. These topics exhibit both high density and strong influence, suggesting substantial progress and extensive academic engagement. The second quadrant, denoted as "Niche themes," reflects areas with a well-established foundation and intensive internal development. However, their overall impact remains limited, indicating that despite active exploration, major breakthroughs have yet to emerge. The third quadrant, labeled "Emerging or declining themes," encompasses subjects in the early or transitional stages of research. Characterized by low density and limited influence, these themes are in a phase of conceptual refinement and empirical accumulation. Finally, the fourth quadrant, referred to as "Basic themes," includes fundamental research directions with promising practical relevance that merit deeper investigation in future studies The strategic thematic evolution matrix further revealed the current status and developmental dynamics of individual themes. The first quadrant, corresponding to motor themes, represented the core hotspots and future priorities of the field. In the present study, themes such as PD-L1 expression and poor prognosis were found among the relatively mature and influential directions. The second quadrant, representing niche themes, included topics with a solid research foundation that have not yet become mainstream. The third quadrant comprised emerging or declining themes, which were characterized by relatively low research density and limited influence. The fourth quadrant represented basic themes, which possess potential practical value and room for further exploration; at present, these include cancer, recurrent, Epstein-Barr virus, and antitumor activity. By contrast, traditional themes such as EBV DNA levels, DNA load, and complete remission were located at the periphery of the network (Fig. 7C). Overall, the field is evolving beyond traditional assessments of EBV-related diseases. There is a growing emphasis on stratified management and the optimization of therapeutic strategies within the context of immunotherapy. Discussion Since 2017, the number of publications concerning EBV DNA as a biomarker for NPC immunotherapy has increased substantially. This trend may reflect a growing interest in precision medicine. It may also be related to recent therapeutic advances in NPC immunotherapy [18]. China has ranked first worldwide in publication output in this field, a pattern that mirrors the geographic distribution of NPC incidence. Chinese investigators have access to the largest NPC patient cohorts and the most abundant clinical sample resources globally, thereby providing a strong foundation for related research. Although the incidence of NPC is relatively low in the United States, the country has maintained clear strengths in fundamental tumor immunology, technological innovation in high-throughput sequencing, and the early clinical development of multiple ICIs [19]. The interplay of academic competition and collaboration between China and the United States has jointly accelerated research output in this area. The Journal of Clinical Oncology (J Clin Oncol) has emerged as one of the most influential journals in this field, and the studies published therein may serve as important points of reference. This observation indicates that research on EBV DNA in the context of NPC immunotherapy has progressed beyond a purely mechanistic focus. Its potential relevance to clinical treatment and prognostic evaluation is receiving increasing attention. Keyword and thematic analyses further corroborate this trend. Immunotherapy, response assessment, and clinical trials have gradually emerged as major research hotspots, with EBV DNA being increasingly examined in new clinical treatment settings. However, broad disease-level terms still dominate the keyword network. More specific issues, such as testing time points, dynamic thresholds, and intervention strategies, have yet to form stable research clusters. These findings suggest that, although the clinical need is evident, a systematic framework for interpreting and applying EBV DNA in this context remains underdeveloped. Therefore, future research should concentrate on defining the expected patterns of dynamics, interpretive logic, and decision-making value of EBV DNA in NPC immunotherapy. This approach would yield more informative insights than merely accumulating similar descriptive findings. After manually reviewing the key retrieved studies (Supplementary Table S4), it was found that the role of EBV DNA in NPC immunotherapy can generally be delineated across three dimensions: prognostic stratification, treatment-response prediction, and dynamic monitoring. At present, the most mature body of evidence supports its prognostic value, with numerous studies showing that a high baseline EBV DNA level is generally associated with poorer survival outcomes [20, 21]. Nonetheless, the most promising avenue seems to be its function in dynamic monitoring during treatment. Prior research indicates that an early reduction in EBV DNA during treatment correlates with an elevated overall response rate (ORR), extended progression-free survival (PFS), and enhanced overall survival, implying that EBV DNA may offer clinically significant insights into therapeutic trends [22, 23]. Xu et al. [11] further reported, in patients with recurrent or metastatic NPC treated with anti–PD-1 monotherapy, that higher baseline plasma EBV DNA was associated with shorter OS and a lower rate of durable clinical benefit, while the ratio of week-4 to baseline EBV DNA was also correlated with outcome. These findings further support the prognostic and dynamic monitoring value of EBV DNA. By contrast, robust evidence supporting a true predictive role remains limited. A multicenter phase III trial showed that tislelizumab plus chemotherapy significantly improved PFS, this benefit was observed in both baseline EBV DNA-positive and EBV DNA-negative patients. In both groups, efficacy was superior to that of placebo plus chemotherapy [24]. This finding suggests that the decision to use ICIs cannot be determined solely on the basis of baseline EBV DNA status. In another study evaluating immunotherapy combined with chemotherapy [25], among patients who remained EBV DNA-positive after 2–6 treatment cycles, those who received Immu/Cape maintenance therapy achieved significantly longer PFS than those treated with Immu alone, whereas no such difference was observed in patients with undetectable EBV DNA. These findings suggest that EBV DNA may still hold potential for guiding individualized treatment strategies. However, no prospective trial has yet demonstrated that EBV DNA can be used directly to adjust NPC immunotherapy regimens. On this basis, a conceptual framework was proposed to integrate the potential roles of EBV DNA across different treatment time points (Fig. 8). Given that current evidence remains insufficient with respect to its predictive value for treatment response, standardized dynamic thresholds, and adaptive intervention strategies, this framework should be interpreted cautiously. It is more appropriately viewed as an evidence-integrating and hypothesis-generating model. It does not yet represent a validated clinical decision-making process.Fig. 8Clinical framework for the use of plasma EBV DNA in NPC immunotherapy. The figure illustrates the potential roles of plasma EBV DNA in pretreatment risk stratification, early on-treatment dynamic monitoring, and post-treatment longitudinal surveillance. Baseline EBV DNA may assist initial risk assessment, while early changes during treatment, including decline, persistent positivity, or rebound, may help inform response evaluation and subsequent management. Continued monitoring after treatment may further support disease surveillance and the identification of recurrence or metastatic progression. This framework is intended to integrate current evidence and provide clinical reference rather than serve as a validated treatment process. [a] https://doi.org/10.1016/j.oraloncology.2025.107743IF: 3.9 Q1 B2. [b] https://doi.org/10.1001/jamanetworkopen.2022.0587IF: 9.7 Q1 B1. [c] https://doi.org/10.1016/j.esmoop.2023.101629IF: 8.3 Q1 B1. [d] https://doi.org/10.3390/v15010066IF: 3.5 Q2 B3 Clinical framework for the use of plasma EBV DNA in NPC immunotherapy. The figure illustrates the potential roles of plasma EBV DNA in pretreatment risk stratification, early on-treatment dynamic monitoring, and post-treatment longitudinal surveillance. Baseline EBV DNA may assist initial risk assessment, while early changes during treatment, including decline, persistent positivity, or rebound, may help inform response evaluation and subsequent management. Continued monitoring after treatment may further support disease surveillance and the identification of recurrence or metastatic progression. This framework is intended to integrate current evidence and provide clinical reference rather than serve as a validated treatment process. [a] https://doi.org/10.1016/j.oraloncology.2025.107743IF: 3.9 Q1 B2. [b] https://doi.org/10.1001/jamanetworkopen.2022.0587IF: 9.7 Q1 B1. [c] https://doi.org/10.1016/j.esmoop.2023.101629IF: 8.3 Q1 B1. [d] https://doi.org/10.3390/v15010066IF: 3.5 Q2 B3 Virological and biological significance of EBV DNA In NPC, EBV is primarily maintained as a latent infection. This condition is typically characterized by a type II latency program, wherein the viral genome persists as nuclear episomes, and latent products such as EBNA1, LMP1, and LMP2 are expressed [26]. In this state, the viral nucleic acids detected in peripheral blood are largely considered to be derived from genetic material that is passively released during tumor-cell turnover and apoptosis. These DNA fragments are short, readily degraded, and not protected by viral capsid structures, providing the biological basis for using peripheral EBV DNA levels as a surrogate measure of overall tumor burden [27]. However, this release pattern may be altered during immunotherapy. Treatment-induced tumor-cell apoptosis or necrosis can increase the release of EBV DNA. Concurrently, systemic immune homeostasis may be reshaped, leading to the reactivation of latent EBV in infected immune cells under specific immune or inflammatory stimuli. Once the lytic cycle is initiated, viral genetic material can be amplified, viral particles can be assembled, and viral components may be released into the circulation through cell lysis or non-lytic pathways, such as extracellular vesicle secretion [28]. Therefore, plasma EBV DNA is predominantly derived from spontaneous tumor-cell death and treatment-related tumor-cell injury, with a possible minor contribution from non-tumor viral reactivation. This complexity reflects the continuous interaction among EBV, NPC cells, and host immunity. EBV-associated tumor cells can impair host antitumor responses through immune evasion [29], whereas NPC immunotherapies, including ICIs, may restore T-cell-mediated tumor clearance [30]. These processes may reshape the dynamics of EBV DNA release and clearance in circulation (Fig. 9). Thus, fluctuations in plasma EBV DNA should be interpreted as a peripheral signal shaped by changes in tumor burden, treatment-related tumor clearance, and potential virological interference.Fig. 9Schematic diagram illustrating tumor cell-virus-host immune interactions and the mechanisms underlying the release of circulating EBV DNA in NPC. In EBV-positive NPC cells, the viral genome is predominantly maintained as episomes and expresses latency-associated proteins, including LMP1, LMP2, and EBNA1, which contribute to tumor biology and immune modulation. NPC cells present antigens to CD8⁺ T cells but often suppress host immune responses through the PD-L1/PD-1 inhibitory axis. Immune checkpoint blockade can disrupt this pathway and restore T cell cytotoxic activity. EBV DNA is released into the circulation through three main routes: (1) The primary source is spontaneous apoptosis and necrosis of tumor cells. (2) Effective therapeutic interventions or other stimuli can induce tumor cell lysis, leading to a substantial release of EBV DNA. (3) A minor fraction may arise from occasionally lytic reactivation of the virus. In plasma, EBV DNA predominantly circulates as highly fragmented double-stranded DNA and serves as an important biomarker for dynamic clinical monitoring Schematic diagram illustrating tumor cell-virus-host immune interactions and the mechanisms underlying the release of circulating EBV DNA in NPC. In EBV-positive NPC cells, the viral genome is predominantly maintained as episomes and expresses latency-associated proteins, including LMP1, LMP2, and EBNA1, which contribute to tumor biology and immune modulation. NPC cells present antigens to CD8⁺ T cells but often suppress host immune responses through the PD-L1/PD-1 inhibitory axis. Immune checkpoint blockade can disrupt this pathway and restore T cell cytotoxic activity. EBV DNA is released into the circulation through three main routes: (1) The primary source is spontaneous apoptosis and necrosis of tumor cells. (2) Effective therapeutic interventions or other stimuli can induce tumor cell lysis, leading to a substantial release of EBV DNA. (3) A minor fraction may arise from occasionally lytic reactivation of the virus. In plasma, EBV DNA predominantly circulates as highly fragmented double-stranded DNA and serves as an important biomarker for dynamic clinical monitoring This virological background provides a theoretical basis for interpreting the levels of EBV DNA in the NPC immunotherapy. Baseline plasma EBV DNA is generally regarded as a circulating biomarker reflecting the overall tumor burden of NPC. However, its biological significance extends beyond merely indicating tumor volume. Previous studies have indicated that pre-treatment plasma EBV DNA levels are correlated with the clinical stage of patients, including T stage, N stage, and M stage. This indicates that EBV DNA can not only reflect the tumor burden of the primary lesion but also represent the burden of regional lymph node involvement and the risk of distant metastasis [31]. Moreover, EBV DNA can provide information about tumor biological behavior that traditional TNM staging cannot fully encompass, including tumor invasiveness, metastatic potential, and systemic disease burden [32]. Therefore, an increase in baseline EBV DNA levels can be understood as an overall increase in tumor burden caused by both the primary and metastatic lesions; it also suggests a stronger biological aggressiveness of the tumor, which biologically explains why patients with high viral loads have poorer prognoses. Following the initiation of immunotherapy, the clinical value of plasma EBV DNA is further expanded. Multiple studies have shown that an early, rapid decline or conversion to negative of EBV DNA is closely associated with better treatment responses and survival outcomes [33, 34]. Since plasma EBV DNA primarily originates from EBV-positive tumor cells, its decline can reflect the elimination of tumor cells mediated by immunotherapy, a reduction in tumor-derived DNA release, and an improvement in disease control. Conceptually, changes in EBV DNA during treatment represent the trajectory of disease evolution. A sustained decline usually indicates effective tumor control. This may reflect both a reduction in tumor burden and the therapeutic effects of immune-mediated clearance. However, such changes should not be regarded as a direct equivalent of immune-mediated clearance. Short-term fluctuations may result from transient DNA release after tumor-cell destruction, and they may also be affected by viral reactivation, which can introduce non-tumor-derived signals. A more cautious interpretation is therefore warranted. Baseline EBV DNA largely reflects overall tumor burden, whereas rapid clearance during treatment is more likely to indicate immunotherapy-related tumor control [35]. In contrast, a single transient increase should be interpreted together with serial monitoring, imaging findings, and other clinical information [36]. Heterogeneity and challenges in standardization The clinical application of EBV DNA in NPC immunotherapy continues to face substantial challenges in standardization. The first major issue lies in sample heterogeneity. Plasma and whole blood are the two most commonly used specimen types. Plasma is generally regarded as the more reliable biomarker. This is because it more closely reflects tumor-derived cell-free DNA and may better indicate the burden of viable tumor throughout the body [37]. Although whole blood tends to yield a higher positivity rate [38], non-tumor-derived EBV DNA released from leukocytes may increase background noise, thereby contributing to false-positive findings and limiting precise quantification [39]. A second challenge arises from methodological variation. Real-time quantitative polymerase chain reaction (R-T qPCR) remains the predominant detection method, yet no unified standard has been established for key procedures such as DNA extraction, primer and probe design, or assay workflow. These methodological inconsistencies have further contributed to the lack of uniform clinical cut-off values. However, in key clinical trials such as RATIONALE-309 [24], the data showed that regardless of whether the baseline EBV DNA titer was higher or lower than 500 IU/mL, the combination of PD-1 inhibitor therapy could bring significant survival benefits. This result further indicates that using a single static cutoff value to predict treatment efficacy has obvious limitations. Nevertheless, these problems are not without potential solutions. A recent expert consensus [40] has proposed a relatively clear path toward standardization, including the unified use of the BamHI-W primer/probe set, standardized plasma separation procedures, consistent reagent systems, the inclusion of both positive and negative controls, and calibration against World Health Organization reference materials to minimize inter-laboratory variation as much as possible. At the same time, cut-off values should be determined in relation to treatment stage, testing purpose, and population characteristics, rather than relying on a single universal threshold detached from the clinical context. Taken together, current standardization challenges should not be regarded as grounds for dismissing the clinical potential of EBV DNA, but rather as technical integration tasks that need to be addressed as the field moves toward clinical implementation. Limitations Several limitations of the present study should be acknowledged. First, the analysis was conducted exclusively on the basis of the WoSCC. Although WoSCC provides a well-structured and comprehensive citation framework that is particularly suitable for knowledge-mapping analysis, reliance on a single database inevitably introduces coverage bias. Previous comparative studies of bibliographic databases have shown that WoSCC differs from Scopus, PubMed, and Embase in disciplinary scope and journal inclusion [41]. Future studies integrating multiple databases may therefore be needed to further verify the stability of the identified research hotspots and knowledge networks. Second, bibliometric analysis itself is subject to inherent confounding biases, particularly excessive self-citation and salami-sliced publication. Excessive self-citation may lead to overestimation of the influence of certain authors or institutions and may slightly distort collaboration networks and co-citation patterns. Salami-sliced publication, in contrast, may artificially inflate the number of publications within specific themes and thereby interfere with annual publication trends and keyword co-occurrence statistics. In addition, recently published studies have had less time to accumulate citations, which may place emerging researchers and newly developing topics at a disadvantage in citation-based analyses. To mitigate these issues, quality control was implemented during data screening, and only original articles and reviews indexed in the WoSCC were included, whereas non-research documents such as conference abstracts and letters were excluded. This approach was expected to reduce, at least in part, the bias introduced by fragmented publication. At the same time, the conclusions of the present study were intended to focus on overall trends, structural patterns, thematic evolution, and emerging hotspots in the field, rather than on evaluating the academic influence of any individual author or institution; nor should these findings be taken to automatically indicate higher-quality clinical evidence. Nevertheless, the above biases remain inherent limitations of bibliometric analysis. In addition, studies on immunotherapy are inherently heterogeneous in terms of disease stage, treatment regimen, sampling time point, and outcome definition. As a result, studies within the same research topic often do not address fully comparable clinical questions. Therefore, the research hotspots and evolutionary trends identified in this study mainly reflect areas of concentrated academic attention. They do not indicate that a unified conclusion has been reached regarding the clinical application of EBV DNA in NPC immunotherapy. Conclusion Since 2017, research on EBV DNA in relation to NPC immunotherapy has proliferated, with China and the United States as the primary contributors, and the Journal of Clinical Oncology (J Clin Oncol) recognized as the most frequently cited publication in this domain. EBV DNA has emerged as a central biomarker in the management of NPC. Its role is evolving from a conventional virus-related circulating marker to a dynamic biomarker that integrates virology, tumor biology, and treatment-response assessment. Plasma EBV DNA primarily originates from short, cell-free DNA fragments released following the apoptosis or necrosis of EBV-positive tumor cells, and its baseline level largely reflects the overall tumor burden. In the era of immunotherapy, the clinical significance of EBV DNA has been further extended. A sustained decline during treatment is generally associated with a reduction in tumor-derived DNA release and favorable tumor control. In contrast, transient fluctuations or renewed elevations may indicate the combined effects of tumor-cell destruction, potential viral reactivation, and host immune responses. Therefore, EBV DNA should not be regarded solely as a static marker for risk assessment; it may also serve as a continuous temporal reference for pretreatment evaluation, on-treatment monitoring, and prediction of disease trajectory. Currently, its most established value lies in prognostic stratification and dynamic monitoring. Further clinical translation will depend on a deeper understanding of its virological basis, the establishment of standardized detection systems, and prospective validation. As these issues are addressed, EBV DNA may become a core biomarker in NPC immunotherapy, possessing both biological rationale and practical clinical value. Supplementary Information Supplementary material 1. Supplementary material 2. Supplementary material 1. Supplementary material 2.