The first comprehensive bibliometric map of this field. This study represents the first bibliometric analysis specifically focused on biomarkers in lung cancer immunotherapy, systematically analyzing 6,180 publications from the Web of Science Core Collection spanning January 2001 through March 2025. The dataset includes 4,477 original research articles and 1,703 reviews, produced by 29,446 authors, with an average of 9 to 10 authors per paper.
Rapid exponential growth. The field exhibited a 28.87% annual publication growth rate, reflecting surging scientific interest in predictive biomarkers. Publications appeared in 917 distinct journals, with a mean citation count of approximately 35 per paper and a citation lifecycle of 3.75 years. These metrics together confirm that biomarker research in lung cancer immunotherapy has achieved both high volume and sustained academic influence.
Three developmental phases identified. Bibliometric trend analysis using Bibliometrix, VOSviewer, and CiteSpace identified three distinct phases: an incubation period from 2003 to 2014 with limited output; a rapid expansion phase from 2015 to 2022 driven by landmark clinical approvals and trials; and a maturation phase from 2023 to 2025 characterized by consolidation, clinical translation, and multi-omics integration. The 2015 inflection point corresponds precisely to FDA approval of nivolumab for advanced NSCLC.
A field at a critical juncture. The recent moderation in growth rate signals a transition from exploratory biomarker discovery to rigorous validation and clinical integration. Despite remarkable productivity, the field faces translational challenges: knowledge is fragmented, global collaboration is uneven, and clinical validation of emerging biomarkers lags behind discovery. This bibliometric synthesis provides an evidence-based roadmap for addressing these gaps.
China leads in volume, US in impact. China produced the most publications with 9,394 papers, followed by the US (6,336) and Italy (2,153). However, the US dominated in citation impact with 93,888 total citations versus China's 41,882, and achieved a dramatically higher H-index of 358 compared to China's 183. This productivity-impact gap reflects differences in research type: US contributions include a higher proportion of seminal clinical trials and mechanistic studies that attract more citations, while Chinese output leans toward validation studies and cohort analyses.
Collaboration patterns reveal strategic opportunity. East Asian nations including China, Japan, and South Korea showed high publication volume but relatively low rates of multi-country collaboration (MCP). In contrast, Western nations showed stronger international partnership rates, and higher MCP correlates with higher citation impact. This suggests that increased cross-national collaboration could amplify the global influence of high-volume Asian research programs.
Harvard University leads institutional rankings. Harvard University ranked first in both publication volume (532 papers) and citation impact (69,247 citations), with an H-index of 117, establishing it as the field's preeminent research institution. The University of Texas System and UNICANCER were also top producers. Notably, US institutions dominated citation rankings despite representing fewer publications, while Chinese institutions dominated publication counts.
High-impact journals as field anchors. The Journal for Immunotherapy of Cancer led with 8,511 citations, followed by Clinical Cancer Research (8,210) and Nature Reviews Clinical Oncology (7,806). Bibliometric Bradford's Law analysis identified 15 core journals. Dual-map overlay analysis confirmed a clear flow from clinical and immunological research frontiers toward molecular biology knowledge foundations, mapping the disciplinary trajectory of the field.
Most cited works defined the field's conceptual backbone. The top cited reference is a 2016 Nature Reviews Cancer paper by Topalian et al. titled 'Mechanism-driven biomarkers to guide immune checkpoint blockade in cancer therapy' (2,063 citations), which categorized multidimensional biomarkers and established a mechanistic development framework. The second most cited (2016, Science Translational Medicine, 1,897 citations) systematically elucidated PD-1/PD-L1 pathway mechanisms and resistance biomarkers.
TMB established by a landmark New England Journal of Medicine study. The third most cited reference is the 2018 NEJM paper 'Nivolumab plus Ipilimumab in Lung Cancer with a High Tumor Mutational Burden' (1,894 citations) by Hellmann et al., which prospectively validated tumor mutational burden (TMB) as a predictive biomarker for dual immunotherapy and shaped subsequent clinical practice guidelines. This paper, along with the POPLAR trial and pan-tumor genomic biomarker studies, forms the empirical core of current biomarker practice.
Citation burst analysis reveals paradigm shifts in real time. The earliest citation burst traces to a 2012 NEJM study establishing PD-L1 expression as a lung cancer immunotherapy biomarker. The longest sustained burst (5 years) belongs to a 2015 Science paper validating TMB as an independent predictive biomarker -- its methodological framework continues to shape biomarker discovery pipelines. Seven references remained in active citation burst phases at the time of analysis, including the CheckMate 816 trial, five-year KEYNOTE-024 follow-up, and IMpower010 trial.
From advanced to resectable NSCLC. Co-citation cluster analysis revealed a clear temporal shift in research focus: early research emphasized predictive biomarkers generally, mid-term research focused on PD-L1 and lung adenocarcinoma, and current research has bifurcated into advanced NSCLC (survival prolongation) and resectable NSCLC (recurrence prevention). Each clinical context demands distinct biomarker strategies and represents a different translational challenge.
Four emerging keyword clusters define the future. Analysis of the top 30 keyword co-occurrences and trend topic visualization identified four emerging research frontiers: genomics, gut microbiome, soluble PD-L1, and immune infiltration patterns. These were identified as the most recently active and highest-growth keyword clusters, signaling a paradigm shift from single-molecule biomarker discovery toward multidimensional, systems-level integration of tumor biology, host immunity, and the microenvironment.
Genomics enables non-invasive multi-dimensional prediction. The keyword 'genomics' is the most contemporary focus, with innovations including blood-based Genomic Immune Subtyping (bGIS) that stratifies ICI responders from non-responders using circulating tumor DNA, and single-cell sequencing studies revealing that FGFBP2-positive NK cells predict immunotherapy response while identifying two distinct drug resistance mechanisms (Treg-enriched and insufficient immune activation). Combining genomics with CT radiomics and deep learning achieved AUC of 0.68 for treatment response and 0.64 for pneumonia risk prediction.
Gut microbiome as a modifiable biomarker system. The gut microbiome influences ICI response through three mechanisms: microbiota-derived metabolites directly activate dendritic cells and CD8+ T cells; dysbiosis disturbs pulmonary immune microenvironment via the gut-lung axis; and microbial antigen mimicry of tumor antigens may compromise immunotherapy efficacy. A predictive model integrating nine microbial metabolites via multi-omics achieved AUC of 0.87. Fecal microbiota transplantation has enhanced ICI responses in melanoma, with lung cancer-specific trials underway.
Soluble PD-L1 and immune infiltration as dynamic biomarkers. Soluble PD-L1 (sPD-L1) circulates in blood and directly suppresses ICI efficacy by binding PD-1 and promoting immunosuppressive macrophage polarization via the PI3K-AKT-mTOR pathway. Baseline sPD-L1 predicts OS in NSCLC patients receiving immunotherapy. Immune infiltration patterns -- particularly spatial aggregation of tumor-infiltrating immune cells -- better reflect antitumor immune activity than conventional density measurements. A distinct 'MT2' immune subpopulation rich in M1-like macrophages and CD8+ T cells correlated strongly with improved response to PD-L1 inhibitors in small cell lung cancer.
Quality-over-quantity strategies drive lasting impact. RAMALINGAM SS exemplifies the quality-over-quantity paradigm: with only 20 publications, he achieved 5,033 citations -- the highest among all analyzed authors. This contrasts with the highest-volume authors (WANG Y with 97 publications, ZHANG Y with 95) who achieved substantially lower per-paper citation impact. The dual model of high sustained productivity (WANG Y) and high-impact landmark research (RAMALINGAM SS) reflects the complementary strategies needed in a translational field.
US-China gap calls for strategic partnerships. The bibliometric disparity between Chinese publication volume and US citation impact suggests that merely increasing research output is insufficient for global influence. Strategic recommendations include breaking down academic barriers through multinational cooperation, integrating data from diverse racial and regional populations via multi-center cohort studies, and focusing Chinese research output on high-risk, high-reward transformative questions rather than validation studies alone.
Advanced versus resectable NSCLC as parallel research arenas. Advanced NSCLC research focuses on survival prolongation through combination biomarker strategies, while resectable NSCLC research emphasizes recurrence prevention through neoadjuvant/adjuvant biomarker-guided therapy. The MRD-EDGE platform, which combines whole-genome sequencing with machine learning to detect molecular residual disease months or years after surgery, achieved a remarkable AUC of 0.98 -- demonstrating that AI-driven liquid biopsy surveillance may transform postoperative monitoring.
Tumor microenvironment architecture as the next clinical frontier. Single-cell sequencing has identified specific cellular signatures -- FCRL4+FCRL5+ B cells and CD16+CX3CR1+ monocytes -- that predict pathological complete response after neoadjuvant immunotherapy. Spatial transcriptomics has enabled detailed mapping of immune cell architecture within tumors. These technologies move beyond population-level biomarkers to address intratumoral spatial heterogeneity, which single-site biopsies cannot capture.
From reductionist to systems-level biomarker science. The central finding of this bibliometric analysis is a documented paradigm shift from reductionist single-molecule biomarker pursuit toward multidimensional integration of genomics, gut microbiome signatures, soluble circulating mediators, and spatial immune architecture. This is not merely an incremental expansion of topics but a qualitative change in how the field conceptualizes the patient-tumor-immune system interplay.
Three actionable future directions. Based on the full bibliometric analysis, the authors prioritize: first, establishing international consortia to validate biomarkers across ethnically and geographically diverse populations to minimize regional bias; second, advancing AI-enhanced multi-omics integration that combines genomic, radiomic, metabolic, and microbiome data streams into comprehensive predictive models; and third, designing prospective trials that incorporate spatially resolved biomarkers such as immune infiltration patterns to predict responses to combination therapies in specific clinical contexts.
Dynamic non-invasive monitoring as a clinical pathway. The convergence of rising prominence in soluble PD-L1, ctDNA, and gut microbiome signatures points toward a tangible clinical pathway: blood-based longitudinal monitoring of biomarkers that adapt to treatment response and tumor evolution, rather than static pre-treatment tissue assessments. This shift from cross-sectional to dynamic monitoring represents the most clinically actionable implication of the bibliometric findings.
Limitations and future bibliometric updates. This analysis is confined to Web of Science and English-language publications only, potentially missing important non-English contributions. Recently published high-impact studies may be underrepresented as citation counts take time to accumulate. Future analyses should incorporate multi-database searches, funding landscape analysis to understand how research investment shapes output, and consideration of non-English language research particularly from East Asian institutions.