Global research trends and hotspots of exosome-mediated drug delivery across the blood-brain barrier: a bibliometric study from 2015 to 2025

Abstract

Background:

The blood-brain barrier (BBB) is a major obstacle to targeted drug delivery for central nervous system (CNS) diseases. Although liposomes and polymeric nanoparticles have improved brain drug delivery, limitations remain in BBB targeting, long-term biocompatibility, and in vivo clearance. Exosomes are endogenous nanoscale extracellular vesicles with favourable biocompatibility, low immunogenicity, and BBB-crossing potential. Therefore, this bibliometric study summarises the current research status, future research trends, and challenges in the more specific field of exosome-mediated BBB drug delivery.

Methods:

A comprehensive search was conducted across the Web of Science Core Collection (WoSCC), PubMed, and Embase databases for relevant English-language literature on exosome-mediated drug delivery across the blood-brain barrier from 2015 to 2025. WoSCC served as the primary source for bibliometric analysis. PubMed and Embase databases were used for supplementary validation. Software such as VOSviewer, CiteSpace, and R-bibliometrix was employed for literature visualisation analysis.

Results:

This study included 1,365 relevant articles from the WoSCC database, and the annual publication volume showed a steady upward trend. China and the United States significantly lead in both the number of publications and the number of core contributing institutions in this field. Co-occurrence analysis of keywords showed that research hotspots are mainly focused on exosomes, the blood-brain barrier, drug delivery, and Alzheimer’s disease. PubMed and Embase were used as supplementary validation databases, including 1,089 and 1,517 records, respectively. Their annual publication trends, major countries/regions, core journals, and keywords/themes were generally consistent with WoSCC, supporting the macro-level stability of the bibliometric findings.

Conclusion:

Unlike previous bibliometric analyses that mainly focused on overall trends in CNS exosome research, this study focuses specifically on the direction of exosome-mediated drug delivery across the BBB. The findings show a shift from basic vesicle characterisation toward engineered delivery systems, CNS disease applications, and translational evaluation. Mammalian-derived exosomes remain dominant, while plant-derived vesicles, AI-assisted design, biomimetic hybrid nanovesicles, and gut–brain axis strategies are emerging areas of focus. Future research should prioritise systematic platform comparisons, standardised evaluation, quality control, scalable production, long-term safety, and regulatory pathways.

1 Introduction

With the acceleration of global population ageing, central nervous system (CNS) diseases, represented by Alzheimer’s disease (AD), Parkinson’s disease (PD) and brain tumours, have become major factors contributing to high disability and mortality rates worldwide (Dong, 2018). Epidemiological data indicate that over the past 30 years, the absolute number of deaths caused by CNS diseases has increased by 39%, and disability-adjusted life years (DALYs) have risen by 15%, with this trend being particularly evident in low- and middle-income countries (Feigin et al., 2020). However, existing treatments still face significant challenges, among which the physiological barrier of the blood-brain barrier (BBB) serves as a core bottleneck that limits the effective delivery of drugs to brain lesions and hinders precise treatment of CNS diseases (Zhang et al., 2025; Yang, 2025; Abbott, 2013). The BBB is a semi-permeable barrier between the brain and the peripheral circulatory system, and, as a core component of the neurovascular unit (NVU), it maintains homeostasis within the brain microenvironment through selective permeability mechanisms while blocking exogenous toxins, pathogens, and abnormal immune cells from entering the central nervous system (Chaulagain et al., 2023; Banks, 2016; Ayloo and Gu, 2019).

This interception function is primarily manifested physiologically as a synergistic effect between physical and biochemical barriers. The physical barrier is composed of brain microvascular endothelial cells, the basal membrane, pericytes, and astrocytes (Pandit et al., 2020). Among them, the tight junctions between endothelial cells form the core of the physical barrier, directly restricting the free diffusion of polar molecules and macromolecules and almost completely blocking the paracellular pathway for hydrophilic substances (Greene and Campbell, 2016; Bauer et al., 2014). The biochemical barrier mainly refers to the highly expressed ABC transporter protein family on the endothelial cell membrane, among which the most typical are P-glycoprotein (P-gp) and breast cancer resistance protein (BCRP) (Tournier and Langer, 2025; Chai et al., 2022). These transporter proteins utilise energy provided by ATP to actively expel lipophilic substances that have already crossed the cell membrane into endothelial cells back into the blood against the concentration gradient (Nicolazzo and Banks, 2011). The combined effect of this physical barrier and biochemical efflux results in extremely low drug entry efficiency into the brain parenchyma. Research data indicate that nearly 100% of macromolecular drugs (such as monoclonal antibodies, recombinant proteins, and gene therapies) and over 98% of small-molecule drugs cannot cross the BBB to achieve effective therapeutic concentrations (Pandit et al., 2020; Banks, 2016). This extremely low drug delivery efficiency has become the main bottleneck currently restricting CNS drug development.

In order to circumvent or physically open the BBB, researchers have attempted methods such as hyperosmotic mannitol, focused ultrasound, increasing drug lipophilicity or increasing the dosage; however, these strategies are often accompanied by risks such as neurotoxicity, cerebral oedema, non-targeted distribution and systemic toxicity (Han, 2021; Wu et al., 2023; Nájera-Maldonado et al., 2025). Therefore, nanomedicine delivery systems have gradually become an important direction in CNS drug delivery research. Synthetic nanocarriers such as liposomes and polymer nanoparticles offer advantages in improving drug stability and enhancing drug loading capacity, but still face challenges with BBB penetration efficiency, biocompatibility, and in vivo clearance (Ekhator et al., 2023). In comparison, exosomes, which are endogenous extracellular vesicles with diameters of 30–150 nm and a lipid bilayer structure derived from host cells, exhibit extremely low immunogenicity and greater circulatory stability (Ha et al., 2016; Mehdizadeh et al., 2025). Furthermore, exosomes can cross the BBB via receptor-mediated transcytosis and accumulate in damaged brain tissue or neuronal regions (Alvarez-Erviti et al., 2011; Abdelsalam et al., 2023). Based on these characteristics, exosomes have received continuous attention as drug delivery carriers for CNS diseases.

Existing bibliometric studies have analysed the overall development trends of CNS exosome research, but these studies mostly focus on the general research landscape of exosomes in CNS diseases, with insufficient attention to the more specific drug-delivery scenario of “exosome-mediated crossing of the BBB”. Unlike previous studies, this study employs a bibliometric approach, not focusing on the annual updates of general CNS exosome research, but systematically analysing the literature related to exosome-mediated BBB-crossing drug delivery, outlining publication trends, collaboration networks, keyword evolution, and research hotspots, and further discussing future research directions and translational challenges in this field, thereby providing a more focused reference for the design of exosome-based BBB-targeting drug delivery systems and clinical translational research.

2 Materials and methods2.1 Search strategy

This study primarily used the WoSCC database for retrieval. PubMed and Embase databases were introduced as external validation sets for clinical relevance. Comparative verification ensured the comprehensiveness and reliability of the retrieval results. Due to differences in field structure and indexing rules across the three databases, the literature retrieved from them was not merged after retrieval. Data extraction and analysis were conducted separately to clarify retrieval differences and ensure consistent research results.

Due to the significant increase in research on exosome engineering, nanodrug delivery systems, and BBB-targeted delivery after 2015, this period better reflects exosomes moving from basic research to brain-targeted drug delivery. Meanwhile, 2025, as the most recent year with complete data available, helps illustrate the changes in the knowledge structure in the more specific research direction of exosomes crossing the BBB over the past decade. Therefore, the search period was set from January 1, 2015, to December 31, 2025. The search strategy was based on MeSH terms and free-text terms used in related thematic studies, and was preliminarily constructed around the three conceptual groups of “exosomes/extracellular vesicles,” “blood-brain barrier,” and “delivery/penetration/transport/targeting” to ensure the search focused on BBB-related drug delivery research rather than generalised CNS exosome studies (Mehdizadeh et al., 2025; Pedder et al., 2025; Rehman et al., 2023; Das et al., 2019). The final search strategy was constructed based on Boolean logic: in WoSCC, TS =(“Exosome*” OR “Extracellular Vesicle*” OR “EVs” OR “Exosomal” OR “Microvesicle*”) AND (“Blood-Brain Barrier” OR “BBB” OR “Blood-brain-barrier”) AND (“Cross*” OR “Penetrat*” OR “Deliver*” OR “Transport*” OR “Transcytosis” OR “Targeting”) was used; in PubMed, a supplementary search was conducted using (“Exosomes” [Mesh]) AND (“Blood-Brain Barrier” [Mesh]); in Embase, the search strategy was based on three core concept groups and adjusted according to its Emtree vocabulary and field rules. All three databases were limited to English-language publications, and only Article and Review document types were retained, with retracted publications deleted. Given that this study is a bibliometric analysis, the aim is to depict the macro-knowledge structure and research trends in exosome-mediated drug delivery across the BBB, rather than to evaluate intervention effects or the quality of evidence in individual studies. Therefore, the search results were limited by publication date, language, and document type using the database’s built-in filtering functions, and retracted articles were excluded without manual secondary screening or risk-of-bias assessment. Ultimately, 1365, 1089, and 1517 articles were selected from WoSCC, PubMed, and Embase, respectively. Supplementary Figure S1 presents the database search and screening process using a PRISMA 2020-style flowchart (Page et al., 2021). All data used in this study were directly sourced from public literature databases; ethical committee approval was not required.

In addition, AI, machine learning, and other related terms were not included as primary search terms in this study to avoid excessively narrowing the scope of the literature. Given that AI-related content is an emerging topic in discussions of future trends, a targeted supplementary search was conducted using the following terms: “exosome” OR “extracellular vesicle”, “drug delivery”, “artificial intelligence” OR “machine learning” OR “deep learning” OR “nanomedicine”. The results are used solely to support the discussion in Section 4.4 and are not included in the main text bibliometric analysis.

2.2 Analysis methods

The articles in the WoSCC database were exported in the formats “Full Record and Cited References” and “Plain Text”, with the file name set to “download_XXX.txt”. In this study, we employed tools such as CiteSpace (version 6.1. R6), VOSviewer (version 1.6.18), and R-bibliometrix (version 4.2.3) for bibliometric analysis. Specifically, CiteSpace was used to perform dual-map overlay analysis, identify citation bursts and keyword bursts, and generate a keyword clustering timeline map, with the time slice set from January 2015 to December 2025. VOSviewer was primarily used for the following analyses: country and institution analyses, co-citation analyses of authors, journals, and documents, and keyword co-occurrence analyses. The R-bibliometrix package was used to present the top ten publishing countries, authors, and journals, and to analyse the evolution of thematic trends in articles retrieved from the WoSCC database. In addition, Microsoft Excel 2019 was used for quantitative analysis of publication output.

The literature in the PubMed database was exported in PubMed format for comparative verification with WoSCC results, focusing on annual publication trends, country/region distribution, institutions, authors, journals, and keyword analysis. The literature from the Embase database was exported in RIS format and was not included in the complete bibliometric analysis, but was used for the final multi-database robustness verification in the results section. This verification mainly compared the consistency between WoSCC and PubMed, and WoSCC and Embase in terms of annual publication trends, major countries/regions, keywords, and core journal sources. Annual publication trends were evaluated using Pearson correlation analysis. For countries/regions, keywords, and journal sources, this study extracted the Top 10 items from each database, computed the Jaccard similarity coefficient to assess item overlap, and used Spearman’s rank correlation to assess the consistency of rankings across common items.

3 Results3.1 Overview of publications

Initially, this study retrieved 1456 and 1161 articles from the WoSCC and PubMed databases, respectively. After systematic database screening, 1365 and 1089 articles were ultimately included for bibliometric analysis. Figures 1, 2 show that, between 2015 and 2025, the annual and cumulative publication volumes in both databases increased steadily, indicating that exosome-mediated drug delivery across the BBB is receiving increasing attention.

Line and bar chart showing annual publications and cumulative publication counts from 2015 to 2025. Annual publication numbers rise from 8 in 2015 to a projected 350 in 2025, with cumulative publications reaching 1,365. The exponential trendline equation and R squared value indicate strong growth correlation.

Annual publication volume and total cumulative publications (WoSCC).

Line and bar graph illustrating the number of publications per year and cumulative totals from 2015 to 2025, showing steady growth and a rapid increase after 2023, with an exponential regression line and equation included.

Annual publication volume and total cumulative publications (PubMed).

3.2 Country/region analysis

The distribution of publications by country/region is generally consistent between WoSCC and PubMed. Supplementary Figures S2, S3 show that in WoSCC, China ranks first with 577 publications, followed by the United States (218), India (72), Iran (55) and Italy (51); PubMed shows similar results, with China (379) still ranking first, followed by the United States (157), India (58), Iran (44) and South Korea (30).

The national cooperation network further shows that international collaboration is mainly concentrated among high-output countries. Among them, the United States has the most collaborations with other countries, totalling 228, followed by China (155) and India (93), indicating that these countries play an important role in the global cooperation network (Figure 3).

Circular network diagram showing international collaborations among countries, with colored clusters representing eight groups. Node size indicates total link strength, as shown in the legend. Multiple colored lines interconnect countries, visually representing collaborative relationships.

National collaboration network visualization map.

3.3 Institutional analysis

There are certain differences among the high-output institutions in the two databases, which may be related to differences in database coverage and the differing methods of institutional indexing. In the WoSCC database, Harvard University has the highest number of publications, followed by the Chinese Academy of Sciences and Shanghai Jiao Tong University, most of which are comprehensive research entities; in the PubMed database, Nanjing University of Chinese Medicine ranks first, followed by the University of Coimbra and Central South University, most of which are medical schools and affiliated hospitals (Table 1 and 2).

RankInstitutionCountryArticle1HARVARD UNIVERSITYUSA662CHINESE ACADEMY OF SCIENCESChina583SHANGHAI JIAO TONG UNIVERSITYChina544TABRIZ UNIVERSITY OF MEDICAL SCIENCEIran495NANJING UNIVERSITY OF CHINESE MEDICINEChina476UNIVERSITY OF CALIFORNIA SYSTEMUSA477UNIVERSITY OF TEXAS SYSTEMUSA468ZHEJIANG UNIVERSITYChina449FUDAN UNIVERSITYChina4310SICHUAN UNIVERSITYChina42

Top 10 institutions by publication volume in the WoSCC database.

RankInstitutionCountryArticle1NANJING UNIVERSITY OF CHINESE MEDICINEChina1872UNIVERSITY OF COIMBRAPortugal1553CENTRAL SOUTH UNIVERSITYChina1154CAPITAL MEDICAL UNIVERSITYChina905SOUTHERN MEDICAL UNIVERSITYChina816TIANJIN MEDICAL UNIVERSITY GENERAL HOSPITALChina817NANJING UNIVERSITYChina658SICHUAN UNIVERSITYChina649TABRIZ UNIVERSITY OF MEDICAL SCIENCESIran6210SOUTHEAST UNIVERSITYChina60

Top 10 institutions by publication volume in the PubMed database.

The institutional collaboration network shows that the Chinese Academy of Sciences is at the centre, with the highest connection strength (28), indicating its important bridging role in global collaboration. Harvard University has the highest cumulative citation frequency (2394), suggesting its strong academic influence in this field (Figure 4).

Network visualization graphic displays institutional collaborations, with nodes labeled for universities and research institutes. Node colors represent clusters, node size indicates collaboration strength, and connecting lines illustrate partnership links.

Institutional collaboration network visualization map (node size represents the total connection strength of the Institution, and the lines indicate the degree of collaboration/connection).

3.4 Author analysis

The volume of publications by authors shows consistency across the two databases, with Wang Y. and Zhang Y. both ranking among the most prolific authors (Supplementary Figures S4, S5).

Co-citation analysis of authors indicates that Zhang Y (429 times), Alvarez-Erviti L. (424 times), and Théry C. (424 times) are the most frequently co-cited scholars, suggesting that their research has had a broad impact on the field of exosome-mediated BBB drug delivery (Figure 5).

Network visualization graph generated in VOSviewer showing clusters of interconnected author names in distinct colors, indicating collaboration patterns and community structure in academic or research publications, with nodes sized by network centrality.

Network visualization map of author co-citation analysis (node size represents the total connection strength of the author, and the lines indicate the degree of collaboration/connection).

3.5 Journal analysis

Research in this field is mainly published in journals in molecular biology, drug delivery, nanomedicine, and pharmaceutics. According to the WOSCC database, the International Journal of Molecular Sciences (72 articles, h-index 28, impact factor 4.9) has the highest number of publications, followed by the Journal of Controlled Release (48 articles, h-index 19, impact factor 11.5) and Pharmaceutics (44 articles, h-index 20, impact factor 5.5); the top three journals in PubMed are largely consistent with WoSCC, indicating that the main publication platforms are fairly stable across different databases (Supplementary Figures S6, S7).

Journal co-citation analysis further shows that the Journal of Controlled Release (3,870 times), International Journal of Molecular Sciences (3,141 times), and Journal of Extracellular Vesicles (2,449 times) are highly influential journals in this field’s knowledge base (Figure 6).

Network visualization map illustrating citation relationships among scientific journals, with nodes representing journals and edges showing citation links. Clusters are color-coded: red, green, and blue, signifying distinct research fields or topics. Larger node size indicates higher citation frequency. Major journals like j control release, j biol chem, and p natl acad sci usa are prominently labeled. Generated using VOSviewer.

Network visualization map of journal co-citation analysis (node size represents the total connection strength of the journal, and the lines indicate the degree of collaboration/connection).

The dual-overlay analysis in Figure 7 indicates that this field is simultaneously supported by both material science and molecular biomedical research. The path from Physics, Materials, Chemistry to Molecular, Biology, Genetics highlights the significance of biomaterials and nanotechnology in the construction of exosome drug delivery systems; the path from Molecular, Biology, Immunology to Molecular, Biology, Genetics reflects the importance of biological mechanisms such as exosome transport, immune compatibility, and BBB penetration efficiency. This suggests that exosome-mediated BBB drug delivery is an interdisciplinary research direction connecting nanomedicine, molecular pharmacology, and CNS disease therapy.

Network diagram visualizing clusters of academic disciplines such as mathematics, physics, biology, and medicine connected by curved colored lines representing interdisciplinary relationships. Each cluster is labeled and color-coded with varying line thickness indicating connection strength.

Overlaying dual-map visualization of disciplinary field flows.

3.6 Citation analysis

Co-citation analysis of the literature can illustrate the overall distribution of foundational publications in this research field. As shown in Figure 8, among the core publications with high total link strength, the study by Alvarez-Erviti L published in Nature Biotechnology in 2011 (total link strength 4443, cited 381 times) ranks first, followed by the study by Haney MJ published in Journal of Controlled Release in 2015 (total link strength 3382, cited 243 times) and the study by Yang TZ published in Pharmaceutical Research in 2015 (total link strength 2771, cited 210 times). These highly co-cited publications mainly focus on targeted exosome delivery to the brain, engineered exosome drug loading, and drug transport across the BBB, indicating that these directions form an important part of the early knowledge base in this field.

Network visualization map generated with VOSviewer showing interconnected clusters of scientific publications represented by colored nodes and lines. Blue, red, yellow, and green clusters indicate thematic groupings, with node size reflecting publication influence and edge thickness depicting strength of citation linkage. Key references are prominently labeled within each cluster.

Network visualization map of literature co-citation analysis (node size represents the total connection strength of the literature, and the lines indicate the degree of collaboration/connection).

Citation bursts show that the study by Yang TZ, published in 2015, had the strongest burst intensity between 2016 and 2020, followed by the studies by Haney MJ, published in 2015, and Théry C, published in 2018 (Figure 9). These burst publications mainly focus on exosome drug delivery across the BBB, CNS disease delivery applications, and EV characterisation standardisation, indicating that early research hotspots focused on validating drug-delivery feasibility and establishing methodological standards.

Table titled “Top 25 References with the Strongest Citation Bursts” displays references by author, publication year, journal, DOI, citation burst strength, and period of activity. A horizontal bar graph represents burst duration in red against a timeline from two thousand fifteen to two thousand twenty-five, showing citation trends for each reference.

References outbreak. Note: The red section indicates the period during which the article was heavily cited.

3.7 Keyword analysis

Keyword analysis revealed common themes and differences between the two databases. In WoSCC, the keywords with the highest total link strength included “exosome”, “extracellular vesicles”, “blood–brain barrier”, “drug delivery” and “Alzheimer’s disease”; in PubMed, the high-frequency terms included “human”, “animal”, “exosome”, “blood–brain barrier” and “extracellular vesicles”, which is related to PubMed’s greater emphasis on biomedical subject indexing (Figures 10, 11).

Network visualization graph generated by VOSviewer showing clusters of interconnected research topics related to extracellular vesicles, with prominent nodes for exosome, blood-brain barrier, drug delivery, stroke, inflammation, and neurodegenerative diseases, categorized by color to represent thematic groupings.

Keyword co-occurrence map (node size represents the total connection strength of the author, and the lines indicate the degree of collaboration/connection, WoSCC).

Network visualization graphic showing clustered keywords related to exosomes, extracellular vesicles, drug delivery, blood-brain barrier, humans, animals, and mice. Nodes are color-coded by research topic and interconnected with lines representing relationships between subjects.

Keyword co-occurrence map (PubMed).

The keyword cluster evolution results in Figures 12, 13 show that the field is gradually shifting from basic molecular research towards drug delivery system design and disease applications. In WoSCC, Cluster #1 RNA and Cluster #2 drug delivery system reflect the significant association between basic microscopic molecular research and carrier design as well as drug delivery; Cluster #6 microbiota increased in activity after 2018, suggesting that gut–brain axis-related mechanisms began to receive more attention. In PubMed, cluster labels such as #3 blood-brain barrier, #6 traumatic brain injury, and #10 spinal cord injury indicate a greater focus on anatomical barriers, disease models, and translational applications. Overall, both databases established a research framework in which extracellular vesicles serve as carriers, with neurodegenerative diseases such as Alzheimer’s disease as the primary therapeutic targets.

Network visualization graphic displaying clusters of research topics related to ischemic stroke, RNA, drug delivery systems, Alzheimer's disease, extracellular vesicles, drug delivery, and microbiota from 2015 to 2025. Nodes represent topics, sized by significance, with multicolored connecting arcs indicating citation links over time, and a timeline running horizontally with color gradients representing each year from blue (2015) to red (2025). Cluster labels and a color legend appear on the right.

Keyword clustering evolution map (WoSCC).

Network visualization graphic displays interconnected research topics from 2015 to 2025. Colored lines and nodes correspond to years and clusters, with a legend labeling topics such as blood-brain barrier, extracellular vesicles, neurodegenerative diseases, and multiple sclerosis, indicating topic evolution and interconnections over time.

Keyword clustering evolution map (PubMed).

The keyword burst analysis further reflects the shift in research focus. “Mesenchymal stromal cells” and “mouse model” experienced bursts from 2018 to 2021 and 2019–2020, respectively, indicating that at that time, the clinical applications of exosomes were still at the stage of animal experiments in basic research. In recent years, keywords such as “permeability” and “proliferation” suggest that researchers are paying more attention to the penetration effect of exosome-mediated drugs across the blood–brain barrier (Figure 14). The thematic trend map also shows that “drug delivery”, “nanoparticles”, “blood-brain barrier” and “extracellular vesicles” have emerged as current hotspots, further supporting the view that the field is shifting from conceptual research on exosomes to operationally engineered drug delivery strategies (Figure 15).

Comments (0)

No login
gif