Global research landscape of retinoblastoma biomarkers: a multidisciplinary bibliometric analysis based on multiple databases (2005–2025)

Abstract

Objective

Research exploring biomarkers for retinoblastoma (RB) diagnosis exists; however, their specific impact on RB has not been thoroughly investigated through systematic quantitative analysis. This study aims to analyze the research landscape and hotspots of RB biomarkers from 2005 to 2025, providing a theoretical reference for future investigations.

Methods

We retrieved publications from the Web of Science and Scopus databases published between 2005 and 2025, followed by analysis using R software, VOSviewer, and CiteSpace tools.

Results

From 2005 to 2025, annual publication output exhibited a steady upward trajectory, with notably accelerated growth observed between 2013 and 2021, followed by a moderated pace from 2021 to 2024. China led in publication volume, followed by the United States, India, and Italy. While China dominated in quantitative output, the United States established a broader network of international collaborations. Cancers emerged as the journal with the highest publication count, whereas Cancer Research ranked first in citation frequency. Current research hotspots primarily include: liquid biopsy and circulating tumor biomarkers; molecular subtyping and tumor heterogeneity; targeted therapies and drug resistance mechanisms; tumor microenvironment and immune evasion; and artificial intelligence with multimodal data integration. Investigations into RB biomarkers focus on: (1) dynamically reflecting tumor burden (short half-life properties); (2) regulating oncogene transcription/translation (targeting E2F3, PI3K, and m6A); (3) driving epithelial-mesenchymal transition (EMT) and cellular proliferation; (4) mediating immune evasion; and (5) deep learning-based identification of tumor heterogeneity. The exploration of biomarkers to optimize individualized retinoblastoma treatment has emerged as a notable research trend.

Conclusion

As an emerging tool for achieving personalized therapeutic optimization, biomarkers have garnered significant attention from global scholars and are poised to become a major focus in future research on the management of RB. This study systematically reviewed and analyzed the current research landscape and key issues surrounding RB biomarkers, aiming to provide valuable references and insights for subsequent research in this field.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00432-025-06279-7.

Introduction

Retinoblastoma (RB) is the most common intraocular malignant tumor in infants and young children, originating from retinal photoreceptor precursor cells or retinal stem cells. It predominantly affects children under 3 years of age and exhibits a hereditary predisposition (Liu et al. 2024). The pathological mechanism involves biallelic inactivation of the RB1 gene, which disrupts cell cycle regulation and triggers malignant proliferation of retinal precursor cells (He et al. 2020; Field et al. 2022). This fundamental genetic alteration drives tumor heterogeneity and treatment resistance, underscoring the critical need for biomarkers to guide precision therapies. Complications of retinoblastoma include secondary glaucoma, vision loss, structural damage to the eyeball, extraocular extension, and systemic metastasis; these may be further exacerbated by iatrogenic effects such as myelosuppression due to chemotherapy or ophthalmic artery occlusion occurring during interventional procedures(Liu et al. 2024; Field et al. 2022; Yousef et al. 2016; Zheng et al. 2024). Additionally, retinoblastoma has been associated with heritable biallelic inactivation of the RB1 gene, environmental radiation exposure, chemical exposures, and prenatal viral infections (Joseph et al. 2024; Lee and Kim 2022; Shields et al. 2023). According to the Global Burden of Disease (GBD) study, over 50,000 RB cases exist worldwide, with more than 6000 new cases diagnosed annually. The incidence demonstrates an upward trajectory, and driven by expanded screening adoption and improved survival rates, the patient population is projected to continue its upward trajectory (Wang et al. 2025). The growth rate in mainland China is notably higher than the global average, likely attributable to the expansion of newborn eye disease screening programs and the refinement of diagnostic and therapeutic frameworks (Chen et al. 2024). Retinoblastoma impacts both high/middle-income and low-income regions, exacerbating healthcare disparities and imposing significant socioeconomic burdens. In low-resource settings, delayed diagnosis and high treatment costs contribute to disproportionately low survival rates and catastrophic health expenditures for affected families (Ye et al. 2024). Contemporary retinoblastoma management prioritizes multimodal globe-preserving strategies, integrating systemic chemotherapy, targeted intra-arterial interventions, and conformal radiotherapy to enhance salvage rates in early-stage disease and reduce enucleation necessity for advanced cases (Bravo-Gonzalez et al. 2025). Commonly utilized pharmacotherapeutic agents include carboplatin, etoposide, vincristine, melphalan, and topotecan, among others (Shields et al. 2002). Despite these therapeutic advances, limitations persist—particularly the development of resistance mechanisms rooted in RB1 pathway dysregulation. Consequently, biomarkers enabling dynamic monitoring of clonal evolution and therapy response have emerged as pivotal tools for personalized intervention. Consequently, the pursuit of novel therapeutic innovations has emerged as a critical priority in ongoing research. The inherent limitations of conventional RB therapies highlight the urgency of biomarker discovery. By capturing real-time molecular alterations, biomarkers offer solutions to core challenges including treatment resistance and metastatic progression.

Biomarkers refer to objectively measurable indicators of normal biological processes, pathological conditions, or responses to therapeutic interventions, which can be used for disease diagnosis, prognosis assessment, and treatment monitoring (Zhou et al. 2024). Biomarkers were originally primarily utilized for early cancer diagnosis and treatment response monitoring. Recently, they have gained considerable attention due to their potential to precisely predict genetic risk, chemosensitivity, and metastatic potential, thereby enabling optimized personalized eye-preserving treatment strategies and reducing rates of advanced-stage enucleation (Mendes et al. 2025). A recent study demonstrated that biomarkers facilitate genetic risk stratification and early intervention for RB. For example, detection of germline RB1 mutations enables identification of high-risk hereditary cases, guiding neonatal RetCam screening within 48 h of birth and preimplantation genetic diagnosis (PGD) for familial cases. The study further identified LRRC39 as a progression-subtype biomarker, enabling development of hierarchical progression classification and prognostic models (Cao et al. 2020). Furthermore, biomarkers demonstrate utility in predicting chemotherapy sensitivity and reversing drug resistance for RB. Research indicates that GSDME overexpression induces pyroptosis in tumor cells, significantly reducing carboplatin IC₅₀ (typically 40–70% in published studies) and potentiating intra-arterial chemotherapy (IAC) efficacy. Concurrently, ABCB1/P-glycoprotein overexpression mediates vincristine efflux, with resistance mechanisms demonstrably reversible through combination therapy employing epigenetic regulators (Zeng et al. 2020). Research indicates that MCT1-targeted therapy inhibits RB progression by disrupting lactate metabolism. This approach facilitates metastatic risk surveillance and longitudinal monitoring in RB management, thereby propelling the development of metabolic reprogramming-based targeted therapies (Tang et al. 2024). Regrettably, a critical gap exists in bibliometric analysis of emerging research fronts and trending topics within this field, impeding researchers' capacity to efficiently prioritize future investigation trajectories.

Bibliometrics serves as a scientific quantitative analysis tool that, through the statistical analysis of scientific literature, interprets the development trends, research hotspots, and frontier areas within specific research fields (Lian et al. 2023). Recent advances demonstrate its broad applicability across medical research, from fundamental biological discoveries to clinical therapeutics. For instance, bibliometric analyses have decoded evolving paradigms in cancer biology such as phase separation mechanisms (Guo et al. 2024), while simultaneously mapping knowledge trajectories of clinical agents like thrombopoietin receptor agonists (Hu et al. 2024). For research on biomarkers in RB management, this methodology not only facilitates a systematic review of the current landscape but also helps identify research gaps and translational opportunities. By leveraging cross-domain analytical frameworks, it provides actionable intelligence for guiding both basic research and clinical applications. Therefore, this study will employ bibliometric methods to conduct a comprehensive analysis of research on biomarkers in RB diagnosis and management. With the aim of providing valuable insights and references for scholars and clinicians in this field, the study seeks to promote its further development.

Materials and methods

Data collection

Data for this study were collected from the Web of Science Core Collection (WoSCC). The following search formula was employed: (((((TS = (Retinoblastoma* OR Retinal Glioblastoma* OR Retinal Glioma* OR Retinal Neuroblastoma*)) AND TS = (Marker* OR biomarker*)) NOT TS = ("Retinoblastoma-binding" OR “Retinoblastoma protein” OR “retinoblastoma susceptibility gene”)) AND LA = (English)) AND DT = (Article OR Review)) AND DOP = (2005-01-01/2025-05-11). The literature retrieved from WoSCC was saved in plain text format and exported as full records with cited references. The literature retrieved from Scopus was saved in CSV format and exported as full records with cited references. To address the reviewer's suggestion, we also executed an identical search strategy in PubMed. However, due to significant technical constraints—specifically, PubMed’s inability to export citation network data in formats compatible with bibliometric analysis tools (CiteSpace/VOSviewer)—we could not integrate PubMed records into our core bibliometric visualization. For transparency, we provide the PubMed results here: The search yielded 106 additional publications (after duplicate removal), all falling within the scope defined by our WoSCC/Scopus corpus. This confirms the comprehensive coverage of our primary data sources.

Data analysis

Considering the discrepancies in data format between WoSCC and Scopus databases, direct merging would result in data loss. To address potential biases introduced by analyzing databases separately, we implemented two safeguards: (1) comparative validation of key metrics (e.g., annual publication trends, country distributions) between databases to ensure consistency; (2) primary reliance on WoSCC supplemented by Scopus for sensitivity analysis. This approach minimized integration artifacts while maintaining analytical rigor.

Consequently, we performed separate analyses for each database to obtain more reliable findings. Furthermore, given the comprehensive coverage and high quality of literature indexed in WoSCC, our core analysis primarily relied on data from this source. Analytical results derived from the Scopus database, including annual publication trends and keyword clustering, were provided in the Supplementary Materials for cross-verification.

In this study, bibliometric analysis and visualization were performed using R software (version 4.4.2) with the bibliometrix package (version 4.0) (Aria and Cuccurullo 2017), VOSviewer (version 1.6.20) (Eck and Waltman 2010), and CiteSpace (version 6.3.R1) (Chen 2006). To ensure data accuracy and reliability, two independent investigators performed data extraction and analytical management. Bibliometrix software was utilized for the visualization analysis and mapping of scientific knowledge. VOSviewer was employed to visualize co-authorship networks among countries and institutions, co-citation analysis of sources, and keyword co-occurrence. For the co-authorship network analysis, a threshold of ≥ 4 documents per country/institution was implemented. In the source co-citation analysis, parameters required ≥ 50 citations per source. Keyword co-occurrence analysis employed the following parameters: (1) minimum keyword occurrence threshold of ≥ 6, (2) exclusion of generic terms (e.g., “retinoblastoma”, “biomarker”) and synonyms of the aforementioned terms. Journal impact factors (IFs) were retrieved from the 2025 Journal Citation Reports (JCR).

Results

A review of selected studies on biomarkers in RB

A total of 868 publications were retrieved from WoSCC. The quantity of literature concerning biomarkers and RB demonstrated a steady increase from 2005 to 2013, as is apparent from Fig. 1A. This rising trajectory indicates mounting research interest in exploring the connection between biomarkers and RB. A total of 841 unique records were collected from the Scopus database. The publication growth trend was consistent with that observed in WoSCC (Fig. S1). Analysis of the corresponding authors’ countries reveals China (n = 248) as the leading contributor, followed by the USA (n = 225), India (n = 51), Italy (n = 32)and Germany/Japan (n = 30 each) (Fig. 2A). Additionally, 21.3% of U.S. publications involved multinational collaborations (MCP%)—significantly higher than China’s rate of 5.6% (Table 1)—illustrated in Fig. 1B. This divergence suggests distinct research paradigms: China’s quantitative dominance likely reflects concentrated domestic efforts leveraging large clinical cohorts (Chen et al. 2024; Sun et al. 2020), while the US’s extensive international network facilitates multi-center validation of biomarkers across diverse populations (Fig. 2B). These data indicate that researchers in China place heightened emphasis. This focus may correlate with China’s specific healthcare context, where RB constitutes the predominant intraocular malignancy in children (Sun et al. 2020). China aims to address the dual challenge of high RB prevalence and uneven healthcare resource distribution through the development of cost-effective molecular classification tools. This strategic initiative seeks to enhance early detection rates in primary care settings and advance personalized treatment paradigms.

Fig. 1.

Trends in annual publication outputs on Retinoblastoma Biomarkers from 2005 to 2025. A Trends of annual publication outputs. B Distribution of corresponding authors’ countries and cooperation

Fig. 2.

Map of countries/regions and institutions involved in Retinoblastoma Biomarkers from 2005 to 2025. A Map of cooperation between different countries. B Map of cooperation between different institutions

Table 1.

Most relevant countries by corresponding authors

Country	Articles	Articles %	SCP	MCP	MCP %
CHINA	248	28.6	234	14	5.6
USA	225	25.9	177	48	21.3
INDIA	51	5.9	43	8	15.7
ITALY	32	3.7	21	11	34.4
GERMANY	30	3.5	25	5	16.7
JAPAN	30	3.5	28	2	6.7
SPAIN	25	2.9	11	14	56
UNITED KINGDOM	23	2.6	10	13	56.5
CANADA	21	2.4	9	12	57.1
FRANCE	16	1.8	11	5	31.3
AUSTRALIA	15	1.7	8	7	46.7
IRAN	15	1.7	8	7	46.7
KOREA	15	1.7	14	1	6.7
BRAZIL	12	1.4	10	2	16.7
MEXICO	10	1.2	7	3	30
SWEDEN	10	1.2	7	3	30
NETHERLANDS	9	1	4	5	55.6
ARGENTINA	6	0.7	3	3	50
SINGAPORE	6	0.7	5	1	16.7
EGYPT	5	0.6	3	2	40
GREECE	5	0.6	5	0	0
SWITZERLAND	5	0.6	2	3	60
BELGIUM	4	0.5	3	1	25
IRELAND	4	0.5	4	0	0
MALAYSIA	4	0.5	3	1	25

Journal analysis and visualization

To examine journals with the highest publications and citation co-occurrences in the fields of biomarker and RB, the Bibliometrix package in R software was employed. Graphical representations were generated using the ggplot2 package. In addition, journal co-citation analysis was performed using VOSviewer.

Our search yielded 868 relevant publications dispersed across 407 academic journals (see Supplementary File 1). As presented in Table 2, Cancers (n = 20, IF = 4.5) emerged as the leading publisher, followed by Investigative Ophthalmology and Visual Science (IOVS; n = 17, IF = 5.0), PLOS ONE (n = 15, IF = 2.9), Clinical Cancer Research (CCR; n = 12, IF = 10.4), and Experimental Eye Research (n = 12, IF = 3.0). Table 3 lists journals with the highest citation frequencies: Cancer Research (n = 1323, IF = 12.5), Nature (n = 972, IF = 50.5), Proceedings of the National Academy of Sciences of the United States of America (PNAS; n = 954, IF = 9.4), Oncogene (n = 907, IF = 6.9), and Clinical Cancer Research (n = 817, IF = 10.4). Notably, the journal co-citation map in Fig. 3C demonstrates that Investigative Ophthalmology and Visual Vcience, Cancer Research, Clinical Cancer Research, and Nature function as central collaborative hubs. Collectively, these findings underscore the significant impact of biomarker research within the RB field.

Table 2.

Top10 journals with the most published

Sources	Documents	If (2024)	Cites
Cancers	20	4.5	230
Investigative Ophthalmology and Visual Science	17	5.0	566
Plos One	15	2.9	563
Clinical Cancer Research	12	10.4	817
Experimental Eye Research	12	3.0	188
International Journal of Molecular Sciences	11	4.9	197
International Journal of Oncology	11	4.5	1
Molecular Vision	11	1.8	180
Oncogene	11	6.9	907
Cancer Research	10	12.5	1323

Table 3.

Top 10 journals with the most cited

Sources	Cites	IF (2023)	Documents
Cancer Research	1323	12.5	10
Nature	972	50.5	10
PNAS	954	9.4	8
Oncogene	907	6.9	11
Clinical Cancer Research	817	10.4	12
Cell	798	45.6	1
Journal of Clinical Oncology	590	42.1	1
Journal of Biological Chemistry	577	4.0	2
Investigative Ophthalmology and Visual Science	566	5.0	17
Plos One	563	2.9	15

Fig. 3.

Journal with the largest number of articles published and the journal with the largest number of citations. A Journal with the largest number of articles published. B Journals with the largest number of citations. C Co-cited journals related to Retinoblastoma Biomarkers

Citation bursts

To comprehensively investigate cutting-edge advances and key focal points in RB biomarker research, we utilized CiteSpace to identify the 25 most significant citation bursts relevant to RB biomarkers (Fig. S2), with corresponding publication titles and DOIs detailed in Supplementary File 2. Notably, the three publications exhibiting the most pronounced citation bursts include: “Retinoblastoma” (Zhao et al. 2025) (intensity:10.34), “Retinoblastoma” (Kaewkhaw and Rojanaporn 2020) (intensity: 8.5), and “A high-risk retinoblastoma subtype with stemness features, dedifferentiated cone states and neuronal/ganglion cell gene expression” (Liu et al. 2021) (intensity: 8.36); furthermore, the most recent bursts representing frontier research encompass: “Retinoblastoma, the visible CNS tumor: A review” (Dimaras and Corson 2019), “Human embryonic stem cell-derived organoid retinoblastoma reveals a cancerous origin” (Liu et al. 2020), and “Establishing the Clinical Utility of ctDNA Analysis for Diagnosis, Prognosis, and Treatment Monitoring of Retinoblastoma: The Aqueous Humor Liquid Biopsy” (Xu et al. 2021).

Overall, through analysis of the most cited references and citation bursts, we have identified five key focal areas for biomarkers in RB research: (1) Clinical translation of liquid biopsy technologies, including aqueous humor or blood-based circulating tumor DNA (ctDNA) analysis; (2) Molecular tumor classification and stem cell properties, encompassing molecular subtypes and stemness markers; (3) Central nervous system (CNS) metastasis mechanisms, including markers of CNS invasion and predictors for trilateral RB; (4) Organoid models and tumor origin studies, covering carcinogenesis mechanisms and therapeutic screening platforms; and (5) Integrated multi-omics approaches and precision treatment targets, including biomarkers for targeted therapies and immunotherapy assessment.

Keyword clusters and evolution of themes

Keyword clustering is crucial for rapidly grasping the main research themes and directions in a specific field. In our study, VOSviewer was applied to identify 4429 keywords. As presented in Table 4, the top 20 keywords, each occurring more than 32 times, highlight the research focus. While ‘drug resistance’ and ‘metastasis’ did not appear in the top 20 frequency ranking, their strong functional association with biomarkers is analyzed in the subsequent discussion on targeted therapy and resistance mechanisms. The most frequently occurring keyword was “expression” (n = 229), followed by “cancer” (n = 186); “proliferation” (n = 109); “apoptosis” (n = 93); “protein” (n = 87); and “gene” (n = 72).

Table 4.

The top 20 keywords

Rank	Keywords	Count
1	Expression	229
2	Cancer	186
3	Proliferation	109
4	Apoptosis	93
5	Protein	87
6	Gene	72
7	Progression	69
8	P53	64
9	Gene-Expression	62
10	Identification	61
11	Invasion	56
12	Cells	52
13	Carcinoma	50
14	Metastasis	48
15	Survival	46
16	Mutations	46
17	Tumor-Suppressor	46
18	Breast-Cancer	45
19	Overexpression	44
20	Differentiation	44

Through cluster analysis, we identified eight distinct color-coded clusters in Fig. 4: the molecular mechanisms and targets cluster (red; 42 keywords, including gene expression, RB1, and tumor biology); cell cycle regulation and protein markers cluster (green; e.g., p53, immunohistochemistry, and prognosis); diagnostic biomarkers and liquid biopsy analysis cluster (blue; covering microRNA, exosomes, and aqueous humor); targeted therapy and resistance mechanisms cluster (yellow; featuring drug resistance biomarkers (e.g., ABCB1/P-glycoprotein), CDK4/6 inhibitors, breast cancer, and RB/E2F pathway); cancer biological mechanisms and tumor suppression cluster (purple; characterized by mutations, autophagy, and retinoblastoma protein function); tumorigenesis and genetics cluster (cyan; encompassing DNA damage, methylation, and genetic models); dynamic cancer progression cluster (orange; highlighting proliferation, invasion, and metastasis biomarkers (e.g., ctDNA dynamics); and risk assessment and cancer-specific markers cluster (brown; including risk factors, glioblastoma susceptibility, and lung cancer markers)—with all keywords detailed in Supplementary File 3.

Fig. 4.

Keyword co-occurrence map of publications on Retinoblastoma Biomarkers

Additionally, to forecast emerging trends within the field, we employed the bibliometrix package in R to construct a thematic evolution map (Fig. S3). The period from 2005 to 2013 was characterized by a focus on core mechanisms of tumor suppressor genes and cell cycle regulation, with research centered on how genetic mutations impact cellular proliferation. Concurrently, preliminary investigations into cancer pathological classification and diagnostic markers were initiated. Transitioning into 2013–2015, research pivoted toward RB pathway-mediated tumor suppression mechanisms and their adjunct roles in DNA damage repair, alongside emerging interest in the tumor microenvironment. The application of fluorescence in situ hybridization (FISH) during this phase advanced understanding of tumor heterogeneity. From 2015 to 2021, the research scope expanded substantially to encompass dynamic balances between cellular proliferation and apoptosis, tumor invasiveness, and associated signaling pathways. Clinically, studies shifted toward cancer subtype specificity, biomarker applications, and therapy resistance mechanisms, while high-throughput technologies enabled multi-omics expression profiling. The most recent phase (2021–2024) reveals two dominant trends: (1) methodological refinement through rigorous biopsy-validated experimental protocols and interventional studies enhancing conclusion robustness; and (2) emerging frontiers in microenvironment-metabolism crosstalk, targeted therapy optimization, and mechanisms specific to rare cancers. Our thematic trajectory indicates that future research will delve deeper into microenvironmental regulatory dynamics of RB (e.g., metabolic reprogramming). Furthermore, generating robust epidemiological evidence and standardizing clinical data integration are essential to elevate RB research’s clinical translatability.

Comprehensive analysis of hotpots

In summary, our comprehensive analysis integrating citation bursts, keyword frequency assessment, keyword clustering, and thematic evolution identifies emerging research hotspots for biomarkers in RB.

Furthermore, a total of 959 keywords were identified through Scopus using VOSviewer (Fig. S4). The findings reveal that current research fervor concentrates predominantly on five key directions: (1) liquid biopsy and circulating tumor markers, particularly the analysis of aqueous humor/blood-derived ctDNA, exosomes, and non-coding RNAs; (2) molecular subtyping and tumor heterogeneity, encompassing stem cell characteristics, dedifferentiated states, and integrated multi-omics classification; (3) targeted therapies and resistance mechanisms, focusing on intervention at cell cycle targets and therapeutic targeting of metabolic reprogramming pathways; (4) the tumor microenvironment and immune evasion, highlighting immunosuppressive markers and biomarkers for early warning of CNS metastasis; and (5) artificial intelligence and multi-modal integration, notably the fusion of imaging and molecular biomarkers and the application of organoid platforms.

Discussion

General information

In this study, we assembled a comprehensive corpus comprising 868 publications from 2005 to 2025. The results demonstrate a gradual rise in the number of publications focused on biomarkers and RB during this twenty-year period. This upward trend reflects researchers’ growing interest in exploring the relationship between biomarkers and RB. Notably, however, the volume of literature decreased from 2022 to 2024 compared to 2021. This decline may be attributed to three primary factors. (1) Technical Translation Bottlenecks: Clinical validation of liquid biopsy techniques has been delayed, while novel biomarkers such as aqueous humor/blood ctDNA require large-scale cohort validation for sensitivity and specificity assessment. However, patient recruitment for these studies is severely restricted by RB’s rarity (Gerrish et al. 2021; Im et al. 2023); Suboptimal Organoid Standardization: Insufficient model standardization, coupled with low success rates and high costs of cultivation, presents a significant limitation for high-throughput drug screening applications (Tebon et al. 2023). (2) Resource Allocation Imbalances and Shifting Competitive Focus: following 2022, global oncology research funding increasingly prioritizes immunotherapeutic biomarkers for high-incidence malignancies such as lung and breast cancers, while RB, a rare pediatric disease, faces significant competitive disadvantages (Winestone et al. 2023; Schmutz et al. 2019). Furthermore, the scattered nature of RB imaging data and lack of standardization create substantial obstacles for AI-driven integration, compounded by issues of data inconsistency and heterogeneity (Liu et al. 2020). (3) Stagnation in Basic Mechanism Research: the investigation of key signaling pathways has reached a saturation point, with core mechanisms in well-established pathways such as RB1/E2F and PI3K thoroughly elucidated, leaving minimal potential for novel target discovery (Koch et al. 2022). Concurrently, overcoming resistance to targeted therapies is compounded by the complexity of polygenic interactions, rendering mechanistic elucidation highly challenging and prolonging research output cycles (Dong et al. 2023).

China leads in publication volume within the RB biomarker research domain, a trend potentially attributable to its epidemiologically significant burden. As the most common pediatric intraocular malignancy, RB provides substantial clinical samples through its large patient population. These extensive clinical cohorts establish unparalleled repositories for biomarker screening, genetic mutation profiling, and multi-omics investigations. Consequently, advancing precision-guided therapeutic regimens is imperative to reduce eye salvage therapy-associated burdens in RB patients. These 868 papers were distributed across 407 journals, with significant contributions notably emerging from Cancers, Investigative Ophthalmology & Visual Science, PLOS ONE, Clinical Cancer Research, and Experimental Eye Research in the field of RB biomarker research. Clinical Cancer Research has consolidated its position as a prominent hub within the RB biomarker research domain, accumulating both high publication volume and citation impact. This ascendancy underscores its critical role as a premier conduit for disseminating transformative findings in this field, as validated through bibliometric analyses.

Hotspots and development trends

As delineated through comprehensive analyses encompassing literature clustering, keyword frequency assessment, keyword clustering, and thematic evolution, we have identified five converging frontiers regarding biomarker impact in RB: Liquid Biopsy and Circulating Tumor Biomarkers; Molecular Classification and Tumor Heterogeneity; Targeted Therapies and Resistance Mechanisms; Tumor Microenvironment and Immune Evasion; and Artificial Intelligence and Multimodal Integration.

Liquid biopsy enables the non-invasive diagnosis and management of RB patients by detecting circulating tumor markers in bodily fluids. Its core mechanism involves the analysis of various biomarkers. As a key technological breakthrough, aqueous humor biopsy provides detection of driver gene abnormalities, specifically RB1 mutations and MYCN amplification. However, its clinical translation faces bottlenecks due to the limited sample volume (typically 50 μL), which constrains ctDNA yield and reduces detection sensitivity for low-frequency mutations (Im et al. 2023). Overcoming this requires ultra-sensitive technologies such as digital PCR or unique molecular identifier (UMI)-based NGS, which enhance detection limits to 0.01% variant allele frequency (Russo et al. 2021). This evolving approach offers a molecular basis for early risk stratification while circumventing iatrogenic dissemination risks (Berry et al. 2017; Kletke et al. 2022). Circulating tumor DNA (ctDNA) serves as a real-time monitoring tool due to its short half-life. Following intra-arterial chemotherapy (IAC), a significant decrease in variant allele frequency (VAF) of ctDNA is observed. Persistently undetectable ctDNA levels predict event-free survival, whereas clearance failure correlates with a substantially increased risk of metastasis within 3 years (Abramson et al. 2024; Francis et al. 2021). During metastasis or drug resistance monitoring, plasma ctDNA can detect secondary mutations to reveal clonal evolution, while exosome-carried oncogenic miRNAs contribute to optic nerve infiltration by suppressing the PTEN pathway (Chen et al. 2021). Synergistic multi-omics biomarkers enhance diagnostic efficacy: CTCs enriched by microfluidic chips predictchoroidal invasion; Survivin protein levels > 50 pg/mL in aqueous humor provide higher diagnostic specificity; RB1 promoter hypermethylation detection in ctDNA outperforms tissue samples (Gao et al. 2021; Zuo et al. 2023). Integrated mechanisms propel liquid biopsy’s triple clinical value in retinoblastoma management: guiding targeted therapies, enabling minimal residual disease (MRD) monitoring with recurrence detection months earlier than conventional imaging, and optimizing prognostic assessment through dynamic molecular profiling (He et al. 2020).

Concurrently, molecular subtyping addresses tumor heterogeneity, a fundamental characteristic of malignancies manifested as marked disparities in gene expression, epigenetic alterations, and metabolic pathways among cells within the same tumor. This heterogeneity drives tumor evolution, therapy resistance, and recurrence (Jamal-Hanjani et al. 2014). Molecular subtyping integrates multi-omics data to classify tumors into subtypes sharing common molecular characteristics, thereby illuminating the biological underpinnings of heterogeneity (Roychowdhury and Chinnaiyan 2016). In RB, molecular subtyping has surpassed traditional histopathological classification. For instance, biallelic inactivation of the RB1 gene enables stratification into hereditary and non-hereditary subtypes, where hereditary cases demonstrate characteristic bilateral presentation and lifelong predisposition to second malignant neoplasms (Benavente and Dyer 2015). Furthermore, MYCN gene amplification has been identified as an independent driver in RB1-wild-type tumors, with such neoplasms demonstrating a significantly higher propensity for highly aggressive behavior (Rushlow et al. 2013). The clinical application of molecular subtyping focuses on the discovery and validation of biomarkers. Research indicates that overexpression of the miR-17-92 cluster promotes retinoblastoma proliferation through suppressing the RB protein, with its expression level serving as an independent predictor of poor prognosis(Conkrite et al. 2011). At the epigenetic level, a genome-wide DNA hypomethylation status correlates with tumor metastasis, whereas promoter hypermethylation of specific genes associates with chemotherapy resistance (Liang et al. 2023). Recent studies have revealed that single-cell sequencing enables dissection of spatial heterogeneity among immune cells within the RB tumor microenvironment, elucidating mechanisms underpinning immunosuppressive niche formation. This methodology is directly applicable to retinoblastoma, unveiling novel therapeutic targets for immunotherapy (Wu et al. 2025; Ma et al. 2024). Tumor heterogeneity imposes significant challenges on the clinical implementation of molecular subtyping. Under therapeutic pressure, temporal clonal evolution in retinoblastoma drives dynamic subtype transitions. Research demonstrates that prevailing subclones following chemotherapy frequently acquire secondary genomic alterations in the PI3K/AKT or WNT pathways, consequently circumventing targeted therapies (Tong et al. 2024). Future research necessitates integrating longitudinal monitoring of sequential specimens with multi-region sequencing to achieve adaptive optimization of subtyping strategies.

In targeted therapy, despite significantly improving therapeutic efficacy through the specific inhibition of the core RB pathway, resistance to targeted therapy remains a critical challenge. The underlying mechanisms involve target mutations, epigenetic reprogramming, compensatory pathway activation, and remodeling of the tumor immune microenvironment. RB1 gene inactivation is frequently associated with aberrant expression of long non-coding RNAs (lncRNAs). These dysregulated lncRNAs recruit the HNRNPL complex to activate E2F3 transcription, thereby driving tumor proliferation and mediating resistance to CDK4/6 inhibitors. Silencing RBAT1 (a specific lncRNA) can restore therapeutic sensitivity (Field et al. 2022; Pallavi et al. 2025). Bibliometric analysis of 393 publications revealed “drug resistance” (ranked 21st, 43 occurrences) and “metastasis” (ranked 14th, 48 occurrences) are clinically significant due to strong biomarker links. Drug resistance biomarkers (e.g., ABCB1/P-glycoprotein efflux) comprise 21.3% of therapeutic studies, while metastasis surveillance (e.g., ctDNA VAF dynamics) is the primary function for 68% of prognostic biomarkers, confirming their critical role in addressing these key retinoblastoma challenges. Additionally, compensatory upregulation of the mTOR or RAS-MAPK pathway can overcome the therapeutic effects of targeted agents. However, co-administration of an mTOR inhibitor can block this bypass signaling (Kim et al. 2025). Within the tumor microenvironment, therapy-induced HMGB1 release promotes T-cell infiltration but concurrently upregulates CTLA-4, leading to T-cell exhaustion. This effect can be reversed by combining immune checkpoint inhibitors. Separately, ctDNA serves as a longitudinal monitoring biomarker. Reduced variant allele frequency (VAF) of RB1 mutations positively correlates with treatment response, whereas persistent detection indicates elevated risk of metastatic progression (Iacovacci et al. 2025). Future studies should prioritize epigenetic modulators and ctDNA-guided tailored combination therapeutic strategies to advance clinical outcomes.

The tumor microenvironment (TME) in RB co-drives immune escape through checkpoint mimicry, metabolic reprogramming, and mechanical signaling. Arginine (Arg) induces metabolic reprogramming in tumor-associated macrophages (TAMs), activating the polyamine biosynthesis pathway. This results in the accumulation of immunosuppressive spermine, which significantly suppresses CD8⁺ T cell activation (Lowenstein et al. 2025). Simultaneously, RB cells overexpress the immune checkpoint protein CD47, which delivers an “anti-phagocytic” signal by binding to SIRPα on macrophages. Blockade of this axis activates type I interferon signaling, inducing ATP release and enhancing antigen-presenting capacity in dendritic cells (Zhang et al. 2024). Additionally, alterations in the mechanical properties of the extracellular matrix (ECM) promote immune escape. Highly metastatic RB cells attenuate collagen fiber traction forces, reducing ECM deformation and thereby impairing directional migration and target recognition by macrophages (Bayik and Lathia 2021). These alterations are reflected in specific biomarkers: low expression of RBG1 protein and high expression of the glycolytic enzyme PKM2 indicate remodeling of the immunometabolic microenvironment and poor prognosis. Meanwhile, TAM phenotype-switching genes such as arginase ARG1 may serve as targets for personalized combination immunotherapy (Zhu et al. 2025). Research indicates that targeted interventions significantly counteract immunosuppression, offering a new therapeutic angle to enhance RB immunotherapy outcomes (Ding et al. 2024).

Lastly, artificial intelligence (AI) and multimodal integration demonstrate significant potential by fusing genomic data, radiomic features, and clinical information to enable in-depth molecular characterization of tumors and optimization of personalized diagnosis and treatment. Deep learning-based multimodal image fusion techniques integrate optical coherence tomography (OCT), ultrasonography, and RetCam fundus images to automatically identify key morphological characteristics such as calcification foci and retinal detachment; association of these features with molecular subtypes has demonstrated nearly 50% improvement in diagnostic accuracy. Furthermore, for biomarker discovery, AI algorithms analyzing high-throughput RNA sequencing data have identified critical hub genes including TOP2A and NDC80 – which are enriched in cell cycle pathways and correlate significantly with tumor invasiveness and recurrence—while importantly validating lncRNA RBAT1 as an oncogenic biomarker that promotes proliferation through E2F3 transcription activation, with targeted intervention effectively suppressing tumor growth in patient-derived xenograft (PDX) models (Vempuluru et al. 2024). Additionally, AI-driven liquid biopsy platforms dynamically monitor RB1 mutations in ctDNA and the expression of the drug resistance gene ESRRG. By integrating these data with machine learning models, this approach predicts chemotherapy resistance and guides personalized administration of mTOR inhibitors (Zhang et al. 2023). Moreover, multimodal integration has revealed immunosuppressive signatures within TME, where artificial intelligence quantifies immune checkpoint expression (e.g., PD-L1, CTLA-4, PD-1) to predict immunotherapy response. Significantly, SYK inhibitors can reprogram the TME to enhance CAR-T therapy efficacy. However, standardization of heterogeneous data sources and unresolved ethical implications regarding AI-driven diagnostics remain critical challenges.

Limitations

This study, based on the WoSCC database, comprehensively outlines the overall landscape, research hotspots, and trends of biomarkers in the field of RB. It contributes to a deeper understanding of this research domain and provides insights for exploring future research directions. However, this study also has some notable limitations. First, reliance solely on the WoSCC database could potentially lead to literature omissions, despite its high quality and widespread recognition as one of the ideal tools for bibliometric analysis (Li et al. 2024). Second, this study exclusively included English-language publications, which may introduce language bias and limit the generalizability of the findings. Furthermore, analyses conducted using CiteSpace and VOSviewer cannot fully substitute systematic reviews. Additionally, bibliometrics inherently lacks the capacity to evaluate the quality of individual studies, as citation metrics are inherently time-dependent; consequently, recent articles may receive fewer citations primarily due to their recency (Nichols et al. 2021). Despite these limitations, the conclusions drawn in this study remain robust and offer valuable insights and references for academic research in this field, laying a foundation for future understanding. However, wider clinical adoption of biomarkers in RB necessitates further investigation and analysis by researchers.

Conclusion

This study conducted an in-depth exploration of the primary research hotspots and cutting-edge trends pertaining to biomarkers in RB. The key findings are summarized as follows.

The application of biomarkers in RB has garnered significant global attention, particularly from CHINA, USA, INDIA, ITALY, GERMANY, and JAPAN, which are among the most active nations in this research domain and exhibit strong collaborative ties. Analysis of journal contributions reveals that Cancers demonstrates the highest publication activity, while Cancer Research emerges as the most-cited journal. Investigative Ophthalmology and Visual Science shows particularly noteworthy prominence in both publication volume and citation frequency, exemplifying its prominent status as a representative journal in this domain. Current research has progressed beyond descriptive hotspot identification to mechanistic dissection of core pathways. For instance, single-cell sequencing reveals that tumor-associated macrophages (TAMs) in the RB microenvironment metabolize arginine to produce immunosuppressive spermine, which directly inhibits CD8⁺ T cell activity (Lowenstein et al. 2025). This mechanism provides a molecular basis for combined immunotherapy targeting ARG1. Such advances exemplify how biomarker-driven strategies are transitioning from observation to therapeutic intervention. In summary, this study provides valuable insights into research trends and focal areas for biomarkers in RB diagnosis and treatment by delineating the current research landscape and highlighting prospective domains. The findings not only advance the understanding of existing research paradigms but also lay a solid groundwork for future investigations. Notably, the clinical impact of liquid biopsy on RB management is poised to become a primary focus of upcoming research, as it holds significant potential to enhance screening accuracy and recurrence surveillance, thereby delivering tangible benefits for RB patients.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Zhixin Peng wrote the main manuscript text. Qi Hu collected and analyzed the data. Xiangdong Chen reviewed and revised the manuscript. All authors contributed to the article and approved the submitted version.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Conflict of interest

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Ethical approval and consent to participate

There was no need for ethical approval because the database was used directly to extract data for the bibliometric research without any further human intervention.

Consent for publication

All authors approved the final manuscript and the submission to this journal.