Search terms were expanded using the Medical Subject Headings (MeSH) vocabulary in PubMed according to the study topic, yielding the following search strategy: (“Periodontitis” OR “Periodontal Diseases”) AND (“Extracellular Vesicle*” OR “Exovesicle*” OR “Exosome*” OR “Apoptotic Bod*” OR “Microvesicle*”). The publication period ranged from 2015 to 2025, document types were restricted to articles and reviews, and the language was limited to English. To ensure comprehensive literature coverage, parallel searches were conducted across three major databases: Web of Science Core Collection (WOSCC) using the Topic field tag (n = 308), Scopus employing the TITLE-ABS-KEY field (n = 414), and PubMed with restrictions to Title/Abstract (n = 281). All searches were executed on January 18, 2026, to minimize temporal bias. The detailed search strategies are listed in Supplementary Table 1.

Records retrieved from the three databases were harmonized and merged by converting them to the WOSCC format using RStudio and the Bibliometrix package. Duplicate records were removed based on digital object identifier (DOI) matching. For records lacking a DOI, manual deduplication was performed using available bibliographic information. Because WOSCC records generally provided more complete metadata, they were prioritized for retention when duplicate entries were identified. Following data cleaning and deduplication, the final dataset comprised 489 unique publications.

Manual screening was carried out to ensure the accuracy of the dataset. Two researchers independently reviewed the titles and abstracts of all retrieved records to assess their eligibility for inclusion. Publications addressing EVs and periodontitis/periodontal disease were included. Records were excluded when the title or abstract clearly indicated that they were unrelated to EVs or periodontitis/periodontal disease. For records whose relevance could not be determined from the title and abstract alone, the full text was retrieved and searched for EV-related and periodontitis-related terms. Records containing terms from both categories were included. Any discrepancies between reviewers were resolved through consultation with a third reviewer. Ultimately, 340 publications were included in the bibliometric analysis [Figure 1].

To facilitate the analysis using bibliometric software, the formats of the 340 publications were standardized. CiteSpace software was used to convert records from Scopus and PubMed into the WOSCC plain text format. Missing critical metadata fields were manually curated to ensure completeness. Bibliometrix subsequently validated key field quality, confirming all reached “good” or “excellent” ratings. Bibliometrix (v5.2.0), VOSviewer (v1.6.20), and CiteSpace (v6.4.R1) were employed for subsequent bibliometric analysis [Figure 1].

Bibliometrix (v5.2.0) was used to analyze the retrieved collection^[29]. General bibliometric information was extracted, and a logistic growth model was used to fit the cumulative publication count over time: P(t) = K/ [1+exp (-b (t- t₀))], where P(t) represents the predicted cumulative number of publications at time t, and K, b, and t₀ denote the estimated maximum cumulative number, growth-rate parameter, and inflection point, respectively. The inflection point corresponds to the time at which the rate of publication growth is maximal. Annual publication output was calculated from the first derivative of P(t). Contributions by country were ranked according to corresponding author affiliation, and collaboration networks were visualized using a minimum edge threshold of 1. Institutional and journal rankings were based on publication counts, and the annual publication dynamics of the top 10 entities were plotted in Microsoft Excel. Multidimensional collaborations among leading institutions, countries, and authors were visualized using Sankey diagrams. Bradford’s law was applied to identify core journals, and their h-indexes were analyzed. Trend topics were identified through term frequency analysis following data cleaning. Citation analysis was performed on documents and references from 2015 to 2025. To further identify emerging research hotspots, documents between 2023 and 2025 underwent additional citation analysis. Documents were ranked according to normalized total citations, and the top 10 were visualized using ggplot2 (v4.0.0).

VOSviewer (v1.6.20) was used for bibliometric analysis and visualization of the retrieved collection^[30]. Node sizes were proportional to their assigned weights. An institutional co-authorship network was constructed, with node weights proportional to document counts. Institutions were filtered using a minimum threshold of one publication per institution. Two journal citation networks were generated, with node weights proportional to document counts and citation frequencies, respectively. Journals were filtered using minimum thresholds of two publications and five citations per journal. Document bibliographic coupling was subsequently analyzed. Node weights were scaled according to citation frequencies for the 2015-2025 corpus and normalized citation counts for the 2023-2025 corpus. Both corpora were subjected to a minimum citation threshold of five citations per document. A reference co-citation network was also constructed, with node weights proportional to citation frequencies. The minimum citation threshold was set at five citations per reference.

CiteSpace (v6.4.R1) was also used to analyze and visualize the retrieved collections^[31]. The time span was set from 2015 to 2025, with a time slice of one year. Collaboration networks of countries and institutions were constructed separately, and the betweenness centrality of each network node was calculated. Betweenness centrality was defined by the number of shortest paths passing through a given node. Nodes with a betweenness centrality > 0.1 were considered central nodes. A keyword co-occurrence network was constructed, and node labels were displayed for keywords with a frequency greater than 20 after synonym merging. Node size was proportional to keyword frequency, while nodes surrounded by a purple outer ring represented central nodes. Keyword clustering analysis was performed, and a timeline visualization was generated. Clustering performance was evaluated using the modularity Q value and the mean silhouette score (S value). It is generally accepted that Q > 0.3 indicates significant community structure, whereas S reflects cluster homogeneity, with S > 0.5 indicating acceptable clustering quality and S > 0.7 indicating high reliability^[32,33]. Furthermore, burst detection analysis was conducted to identify references exhibiting the strongest citation bursts.

A total of 340 documents regarding EVs in periodontitis were retrieved between 2015 and 2025 [Figure 2A]. A logistic growth model implemented in Bibliometrix was used to characterize the research life cycle of the field. Such models provide insights into the developmental stage of a research topic, forecast upcoming trajectories, and project the timeline for the research domain to plateau. The model yielded a coefficient of determination (R²) of 0.97, indicating an excellent agreement between the observed and fitted publication data [Figure 2B]. The estimated maximum saturation capacity (K) was 1,241 publications [Figure 2C]. The ratio of cumulative publications to the K value indicates the developmental stage of the field. As of the last observation year (2025), the cumulative number of publications had reached 340, accounting for 27.4% of K. This indicates that the field is in a rapid growth phase (defined as 10%-50% of saturation) and remains a highly active research direction. The inflection point, corresponding to the maximum publication growth rate (50% of K), is projected to occur in 2027, with an estimated 134 publications. Thereafter, the growth rate is predicted to decline as the research topic approaches maturation. The time interval between the emergence of the field (10% of K) and near-saturation (90% of K) is estimated to be 10.2 years.

Geographical analysis identified 31 contributing countries based on corresponding author affiliations. The top 10 countries ranked by the number of publications (11 due to tied rankings) are shown in Supplementary Figure 1A and Table 1. China ranked first with 210 publications, accounting for 61.8% of the total collection. However, its average citation rate was relatively low at 24.8 citations per article. The USA ranked second with 22 publications (6.5%) but achieved the highest average citation rate, with 69.9 citations per article. These findings indicate that while China is the most productive country in this field, its modest average citation rate suggests a need to increase its research impact. Alternatively, the low average citation rate may partly result from the recent increase in publication output, as newer articles have had less time to accumulate citations. The collaboration map [Supplementary Figure 1B] illustrates a centralized international network dominated by several core countries. High-frequency collaboration pairs include China-USA, China-Japan, USA-Australia, and USA-Japan. CiteSpace analysis identified the USA (Centrality, C = 0.77), China (C = 0.34), Australia (C = 0.34), Germany (C = 0.16), and Iran (C = 0.14) as the five central nodes in the collaboration network [Table 1]. Notably, the USA exhibited the highest centrality, highlighting its prominent bridging and hub role within the international collaboration network, while China and Australia also demonstrated substantial network connectivity.

The institutional affiliation analysis identified core research forces and potential collaborative targets. A closely-knit collaboration network dominated by Chinese universities, with limited international interaction, was identified among 427 total contributing institutions [Supplementary Figure 2A]. Table 2 presents the top 10 institutions, nine of which are based in China. Sichuan University ranks first with 109 articles (32.1% of the total) and, with a centrality of 0.15, constitutes a critical hub in the institutional collaboration network. Air Force Medical University ranks second with 51 articles (15.0%), followed by Wuhan University (11.2%). Supplementary Figure 2B shows that most of the top 10 entities entered this field between 2019 and 2021 with a stable output. Sichuan University entered this research field in 2019 and subsequently showed a steady increase in publication output. Notably, Seoul National University, the only non-Chinese institution among the top 10 contributors, maintained publication activity from 2020 to 2023 but produced no publications during 2024 and 2025. Furthermore, Supplementary Figure 2C highlights the dominant position of China, with outgoing publication flows substantially exceeding those of other nations. Sichuan University ranks first in publication output among Chinese institutions. Scholars including Tian Weidong, Guo Shujuan, Liu Qian, and Huo Fangjun are highly productive researchers affiliated with Sichuan University.

Identifying the journals that most frequently publish research in this field assists scholars in selecting appropriate publication venues. The 340 publications were distributed across 154 journals [Figure 2A]. Table 3 presents the top-ranking journals (12 journals due to tied rankings). The International Journal of Molecular Sciences [impact factor (IF) = 4.9, h-index = 12] ranked first, with 23 publications. Other prominent journals in the molecular/nanomedicine area include the International Journal of Nanomedicine (14 publications) and the Journal of Nanobiotechnology (13 publications).‍‍‌‍ ‍‍‍‍‍‍Four stomatology journals were represented, notably the Journal of Clinical Periodontology, a widely recognized journal in periodontology. Immunology and stem cell biology are represented by Stem Cell Research & Therapy (14 publications) and Frontiers in Immunology (13 publications). This distribution across multiple disciplines indicates that EVs research in periodontitis closely integrates traditional dentistry with immunology, materials science, and nanomedicine. Remarkably, 11 of the top 12 journals are classified as Journal Citation Reports (JCR) Q1, including high-IF journals such as the Journal of Nanobiotechnology (IF = 12.6) and Advanced Science (IF = 14.1). These findings indicate that research on EVs in periodontitis is being disseminated through influential journals across several disciplines.

In citation networks, the International Journal of Molecular Sciences leads in number of publications [Supplementary Figure 3A], while Frontiers in Immunology leads in the total number of citations [Supplementary Figure 3B]. Regarding annual trends [Supplementary Figure 3C], the International Journal of Molecular Sciences has published steadily since 2016, whereas most other top journals began publishing around 2022.

Bradford’s law of scattering, one of the three major laws in bibliometrics, describes the distribution of publications across journals within a given research field. When journals are ranked in descending order based on the quantity of publications within a specific field, they can be categorized into three distinct areas: a core area, a related area, and an unrelated area, with each area containing equal numbers of articles^[34,35]. The 154 journals found in this study were divided into three zones according to Bradford’s law, resulting in the identification of 10 core journals [Supplementary Figure 3D and Table 4].

The keyword co-occurrence network generated by CiteSpace comprised 305 nodes, displaying labels with frequency > 20 [Figure 3A]. “Extracellular vesicles”, “mesenchymal stem cells”, and “periodontitis” emerged as core keywords. Nodes with high betweenness centrality included “differentiation”, “cells”, “activation”, and “expression”, acting as pivotal structural bridges within the network. Spatially, the upper region links to pathogenic mechanisms (“outer membrane vesicles”, “Porphyromonas gingivalis”, and “inflammation”), whereas the lower region focuses on their regenerative and therapeutic roles (“regeneration”, “angiogenesis”, and “osteogenic differentiation”). A clustering analysis grouped keywords into 10 clusters [Figure 3B]. This clustering structure showed high robustness (modularity Q = 0.4858 > 0.3) and excellent homogeneity (weighted mean silhouette S = 0.7716 > 0.7). The major clusters were primarily associated with two research themes: therapeutics (#0 tissue engineering and #9 engineered exosomes), and pathogenesis (#2 outer membrane vesicles, #3 Porphyromonas gingivalis, #7 alveolar bone loss, and #8 microbiota dysbiosis).

Figure 3C and D show an evolutionary trajectory over the past decade. Initially, research on EVs in periodontitis focused on pathogenic mechanisms (Clusters #2, #3, #7 and #8), characterized by “outer membrane vesicles”, “Aggregatibacter actinomycetemcomitans”, and “gram-positive bacteria.” Notably, themes in Clusters #2 and #3 have recently resurged. Subsequently, the focus shifted to bone regeneration-oriented therapies utilizing MSC-EVs (Cluster #0), evidenced by “regeneration”, “osteogenesis”, and “angiogenesis.” Recently, the research frontier has pivoted toward “drug delivery” and “immunomodulation”. Engineered EVs for drug delivery represent an emerging but still scarce area (Cluster #9). Meanwhile, immunomodulation (Cluster #4) has become a focal point, emphasizing macrophage polarization and the use of macrophage-derived EVs as therapeutics. Significantly, trend topics highlight a growing clinical interest in EVs as diagnostic biomarkers [Figure 3D].

Figure 4A displays the top 10 cited publications in the field between 2015 and 2025, highlighting the therapeutic and pathogenic roles of EVs in periodontitis. Early studies demonstrated that human mesenchymal stem cell-derived EVs (MSC-EVs) promote periodontal tissue and alveolar bone regeneration^[36]. MSC-EVs retain the immunomodulatory properties of their parent cells^[37] and have been implicated in modulating macrophage polarization toward the M1 or M2 phenotype^[38]. EVs from dental pulp stem cells (DPSCs)^[39] or tumor necrosis factor-alpha (TNF-α)-preconditioned gingival mesenchymal stem cells (GMSCs)^[40] can facilitate M2 macrophage polarization, thereby supporting tissue repair. In addition, adipose tissue-derived EVs have entered clinical trials for periodontal pocket treatment^[41]. Beyond mammalian sources, plant-derived EVs (e.g., those isolated from ginger) show therapeutic potential through inhibition of Porphyromonas gingivalis (P. gingivalis) pathogenicity^[42]. In contrast, bacterial OMVs are pathogenic^[43]. P. gingivalis OMVs trigger periodontitis by inducing a metabolic shift toward glycolysis in macrophages^[44]. Aggregatibacter actinomycetemcomitans (A. actinomycetemcomitans) OMVs can cross the blood-brain barrier and induce TNF-α production, suggesting a pathophysiological link to Alzheimer’s disease^[45]. Further details of these highly cited publications are provided in Supplementary Table 2.

To overcome time-lag bias, articles from 2023-2025 were ranked by normalized total citations [Figure 4B]. Among the top 10, one review highlighted salivary EVs as periodontal biomarkers^[46]. One study investigated the effects of bone marrow mesenchymal stem cell (BMSC) derived EVs on alveolar bone metabolism under inflammatory conditions^[47]. The remaining studies focused on therapeutic applications. Sources of therapeutic EVs have expanded beyond MSCs^[37,41], with EVs derived from Schwann cells^[48] and M2 macrophages^[49,50] demonstrating their capacity to enhance bone regeneration. In addition, EVs are increasingly being explored as drug delivery systems^[51]. For example, melatonin-loaded M2 macrophage-derived EVs have been shown to reduce endoplasmic reticulum stress and the unfolded protein response, thereby restoring the osteogenic and cementogenic differentiation capacity of inflamed periodontal ligament cells^[50]. Additionally, the optimization of hydrogels as EV carriers remains an active area of research^[52,53]. Collectively, these advancements further drive the clinical application of EV treatments. A detailed summary of the ten publications is presented in Supplementary Table 3.

A total of 20,159 cited references were identified, constituting the knowledge base of this research field [Figure 2A]. Figure 4C and D display the 10 most cited and the 10 top references with the strongest citation bursts. Notably, the study by Nakao et al. ranked first among both the most cited documents [Figure 4A] and the most cited references [Figure 4C]^[40]. The highly cited references can be categorized into three themes. The first encompasses disease background and theoretical consensus, including the immunopathogenesis of periodontitis^[54,55], EV biogenesis^[9,10], MISEV guidelines^[56], and the role of periodontal ligament stem cells (PDLSCs) in tissue engineering^[57]. The second theme involves disease mechanisms, with studies demonstrating how EVs derived from periodontal pathogens and stressed host cells contribute to disease progression^[58,59]. The third theme focuses on therapeutic applications, highlighting the potential and effectiveness of MSC-EVs from diverse sources^{[36,39,40,60-65]}. Figure 4E-G shows citation-weighted VOSviewer clustering networks for documents in the time periods 2015-2025 and 2023-2025 and cited references. These networks further illustrate the central position of the highly cited publications and references identified in Figure 4A-C within the knowledge structure of the field.

This study analyzed 340 articles on EVs in periodontitis published between 2015 and 2025. The life cycle analysis of scientific production indicated that this research area is currently in a growth phase. According to the logistic growth model, the field is projected to peak in 2027 and to take approximately 10.2 years from emergence to near-saturation. These findings suggest research on EVs in periodontitis remains an active area of investigation. This rapid growth can be attributed to several factors. First, the increasing clinical and social demand for regenerative periodontal therapies has accelerated the translational exploration of EVs. Second, advances in EV isolation, characterization, and detection technologies have improved feasibility and reproducibility, thereby reducing technical barriers. Furthermore, integrating EV research with bioengineering has broadened the scope of potential applications.

China has become the leading contributor to research in this field, driven largely by institutions such as Sichuan University, Air Force Medical University, and Wuhan University. As China’s top institutions for stomatology research, they represent key nodes within the current research network. Despite its substantial publication output, China exhibited a lower average citation rate than countries such as the USA, Japan, and Korea. This difference may partly reflect the recent increase in publication volume, as newer articles have had less time to accumulate citations. Nevertheless, this suggests potential to further increase the impact and visibility of research outputs from China while strengthening international collaborations.

Journal analysis provides insights into patterns of knowledge dissemination within a research field. Identifying core journals may help researchers select appropriate venues for publication. The International Journal of Molecular Sciences was the most productive journal in this field. Furthermore, journals identified as core journals according to Bradford’s Law are all classified as JCR Q1. Several of these journals, such as the Journal of Nanobiotechnology and Advanced Science, have IFs exceeding 10, indicating that research on EVs in periodontitis is being published in widely recognized, high-impact multidisciplinary journals. In addition, the concentration of publications in a relatively small number of core journals may facilitate literature retrieval and help researchers remain informed of emerging developments in the field.

Analysis of highly cited publications and references provides insight into the intellectual foundation of this research field [Figure 4], where keyword analysis facilitates the identification of major research themes and emerging trends [Figure 3]. To support a systematic interpretation of the bibliometric findings, the following discussion integrates the knowledge base and research themes identified in the analysis with evidence from the broader literature. In addition to highly cited studies and references with strong citation bursts, relevant studies from the wider body of literature are included to contextualize and complement the bibliometric results. Collectively, this synthesis reveals that EVs play three distinct roles in periodontitis: serving as potential therapeutic agents for periodontal regeneration, acting as mediators in the pathogenesis of periodontitis and its associated comorbidities, and functioning as diagnostic biomarkers with potential clinical utility [Figure 5]. The subsequent sections will elaborate on these three roles.

As a biofluid that directly interacts with the oral microenvironment and can be collected non-invasively, saliva has considerable potential as a diagnostic medium for periodontitis and for monitoring disease progression^[165]. Compared with whole saliva, which may contain contaminants such as food debris and microorganisms, salivary EVs can be separated from contaminants during the isolation process. Moreover, the lipid bilayer of salivary EVs protects their cargo from enzymatic degradation, including by salivary enzymes such as amylase^[15]. Consequently, salivary EVs may serve as more sensitive and specific biomarkers for oral diseases.

Current research identifies distinct salivary EV signatures in periodontitis, including mRNA, proteins, metabolites, microRNA, and other molecular cargoes. Salivary EVs from periodontitis patients show reduced levels of the bone turnover marker osterix (OSX) alongside increased expression of the pro-inflammatory cytokine TNF-α^[166]. Furthermore, programmed death-ligand 1 (PD-L1) expression correlates positively with severity of inflammation^[167]. EVs-related tetraspanins CD9/CD81 show significant downregulation in salivary EVs of patients with periodontitis^[168]. Notably, specific miRNAs enriched in salivary EVs (hsa-miR-140-5p, hsa-miR-146a-5p, and hsa-miR-628-5p) are markedly upregulated in periodontitis^[169]. Metabolomic analysis revealed distinct alterations in salivary EVs from children with localized aggressive periodontitis, including in acetylcholine, creatinine, L-pipecolic acid, and N-acetylneuraminate levels^[170]. Longitudinal studies further demonstrate the capacity of EVs to dynamically reflect therapeutic responses. Wang et al. investigated salivary EVs in patients with periodontitis who exhibited clinical improvement following non-surgical periodontal therapy. Their findings revealed a reduction in specific EV subpopulations (CD63⁺/CD24⁺/CD142⁺), a downregulation of pro-inflammatory cytokines (IL-6, IL-2), and an upregulation of the anti-inflammatory cytokine IL-10^[171]. Collectively, these findings highlight the potential of salivary EVs as a promising liquid biopsy tool for the diagnosis and monitoring of periodontitis.

Gingival crevicular fluid (GCF) is another common oral biofluid. Although GCF collection is more technically demanding than saliva, its site-specific nature allows it to more accurately reflect local periodontal activity^[172]. Patients with periodontitis exhibit significantly higher total concentrations of GCF-EVs, which are positively correlated with clinical measures of periodontal disease severity. Notably, no significant difference in EV concentration was observed in concurrently collected saliva samples^[173]. Furthermore, pregnant women with gestational diabetes mellitus and periodontitis show significantly elevated concentrations of GCF-EVs, with EV cargo enriched in proteins associated with immune responses and glucose and insulin signaling pathways^[174]. In summary, GCF-EVs represent promising diagnostic biomarkers for periodontitis, although the current evidence base remains limited.

Oral disease-associated EVs can enter the systemic circulation and may therefore serve as blood-based diagnostic biomarkers for periodontitis^[172]. Patients with chronic periodontitis exhibit more than twofold differential expression of 33 miRNAs in plasma-derived EVs^[175]. Another study identified miR-1304-3p, miR-200c-3p, C/D box small nucleolar RNA (SNORD) 57, and SNORD1771 from plasma-derived EVs as potential biomarkers for periodontitis^[176]. However, because circulating EVs are distant from the primary sites of many oral diseases, they are susceptible to systemic confounding factors. Moreover, disease-related signals may be diluted within the bloodstream, potentially reducing their diagnostic sensitivity^[172]. For example, in pregnant women with gestational diabetes mellitus and periodontitis, EV concentration and biomarker profiles were altered in GCF, whereas no significant differences were detected in plasma EVs^[174].

This study has several limitations. First, with regard to data sources and retrieval, despite integrating WOSCC, Scopus, and PubMed to maximize literature coverage, relevant studies indexed exclusively in other databases (e.g., Google Scholar and Dimensions) may have been overlooked. Moreover, only English-language publications were included, which may have excluded high-quality studies published in other languages and consequently underestimated the contributions of certain countries or regions. Only articles and reviews were considered, whereas other document types, such as conference proceedings, were excluded despite their potential to capture emerging research trends.

Second, regarding indicators and methodology, bibliometric indicators primarily reflect research productivity, visibility, and scholarly influence rather than methodological quality, scientific rigor, or clinical relevance of individual studies. Therefore, they cannot directly assess the substantive quality of specific publications. Citation-based indicators are also subject to temporal bias, as recently published studies may be underrepresented due to their limited time for citation accumulation. In addition, citation counts do not differentiate between supportive and critical citations. Furthermore, the identified research hotspots and evolutionary trends are inherently descriptive and retrospective. While they provide insights into publication patterns and the intellectual structure of the field during a defined period, they do not support causal inferences regarding the drivers of scientific development.

Future studies should incorporate additional databases, broader language coverage, and complementary methodologies, such as systematic reviews, content analyses, and expert evaluations, to provide a more comprehensive and robust understanding of this research field.

Building on the evolutionary trends identified in the bibliometric analysis, future research should prioritize the translational development of EV-based therapeutic and diagnostic strategies while further elucidating the roles of EVs in the pathogenesis of periodontitis and its associated comorbidities.

In line with current research priorities in immunomodulation and targeted drug delivery, MSC-EVs from diverse tissue sources have demonstrated the capacity to attenuate periodontitis-associated inflammation and promote periodontal tissue repair and regeneration. However, their therapeutic efficacy may vary according to cellular origin. Therefore, systematic comparative studies are needed for the rational selection and optimal clinical translation of MSC-EVs. Initial evidence indicates source-dependent differences. Ma et al. compared uptake by PDLSCs of EVs derived from DFCs, adipose-derived MSCs, BMSCs, and umbilical cord MSCs, and found that DFC-EVs were internalized most rapidly and reached uptake saturation earliest^[103]. Zheng et al. reported that SHEDs produced more EVs than DPSCs, while SHED-EVs more effectively promoted PDLSC proliferation, migration, and osteogenic differentiation^[69]. Similarly, GMSC-EVs exhibited greater immunomodulatory activity than BMSC-EVs in vitro^[177]. Nevertheless, current comparative evidence is largely limited to in vitro studies and is insufficient to establish the superiority of any single MSC-EV source for periodontal regeneration. Future systematic, head-to-head studies using standardized EV isolation, characterization, and quality-control procedures, as well as harmonized disease models and outcome measures, are required to define the relative advantages, limitations, and optimal clinical applications of MSC-EVs from different sources. Such studies may facilitate the development of source-specific EV therapies for periodontal inflammation control and tissue regeneration.

In studies of the pathogenic roles of EVs, P. gingivalis and OMVs have remained prominent research hotspots. Notably, the cluster #8 “microbiota dysbiosis” emerged during the early development of this research field but received relatively limited attention in subsequent studies. Given that periodontitis is not driven by a single microorganism but arises from the interplay among dysbiotic polymicrobial biofilms, host immune-inflammatory responses, and environmental factors^[126], further investigation into polymicrobial BEV interactions is warranted. Consequently, although studies focusing on P. gingivalis and its OMVs have provided important insights into specific virulence factors and molecular mechanisms, they cannot fully recapitulate the interspecies interactions and BEV-mediated communication networks that characterize human periodontitis. Future research should therefore progress from single-species BEV models toward multispecies-derived BEVs and, ultimately, BEVs isolated from patient-derived subgingival microbiota. Such approaches may provide a more comprehensive understanding of how polymicrobial BEV networks contribute to periodontal inflammation, alveolar bone loss, and the development of periodontitis-associated systemic diseases.

Research on EVs as clinical biomarkers is attracting increasing attention. Saliva is readily accessible and amenable to repeated, non-invasive sampling, facilitating broad implementation and large-scale longitudinal cohort studies. Consequently, salivary EVs have been investigated extensively in the context of periodontitis. However, salivary EVs are susceptible to dilution effects and interference from the oral microbiota, which may limit their ability to accurately localize periodontal lesions when used in isolation. Future diagnostic strategies should therefore move beyond single-analyte assays toward integrated multiparametric and multi-omics approaches, enabling comprehensive monitoring of disease dynamics, therapeutic response, and prognosis. Among the currently available sample types, GCF is collected closest to periodontal lesions. Consequently, GCF-EVs may provide greater site specificity and diagnostic sensitivity. However, systematic characterization of GCF-EVs, including proteomic profiling, remains in its early stages. Although blood collection is more invasive than saliva and GCF sampling, plasma-derived EVs may better reflect long-term systemic alterations and could be particularly valuable for detecting systemic disorders associated with periodontitis. Future studies should capitalize on the complementary strengths of these biofluids to establish integrated EV-based biomarker platforms. Nevertheless, substantial efforts are still required before EVs can be translated into routine clinical biomarkers for periodontitis.

In addition to these emerging research topics, the clinical translation of EV-related findings requires careful consideration of the limitations of currently available experimental models. Substantial differences exist between animal models and humans with respect to oral microbial composition and abundance, periodontal anatomy, and immune responses. Non-human primates more closely resemble humans in periodontal tissue structure and disease phenotypes^[178], but their high cost limits widespread use. In contrast, rodents, particularly mice, harbor oral microbiota that differ markedly from those of humans^[179], and experimental periodontitis typically requires ligature placement and/or inoculation with human periodontal pathogens^[180]. Therefore, EV-related mechanisms identified in animal models should be extrapolated to human periodontitis with caution. Their translational relevance requires further validation in human-derived samples and experimental systems with greater physiological relevance to human disease.

This study provides a comprehensive bibliometric analysis of research on EVs in periodontitis from 2015 to 2025, revealing a rapidly expanding field with substantial scientific influence, led primarily by China and several of its major research institutions. Current research hotspots are centered on immunomodulation and targeted therapeutic delivery. Collectively, the available evidence highlights multifaceted roles of EVs in periodontitis. From a therapeutic perspective, EVs promote periodontal tissue regeneration and modulate the immune microenvironment, while engineered EVs offer a versatile platform for targeted cargo delivery. In contrast, pathogenic EVs contribute to disease progression by amplifying inflammatory responses and disrupting bone homeostasis, thereby driving periodontal tissue destruction and potentially linking periodontitis to systemic comorbidities. Furthermore, although biofluid-derived EVs show considerable potential as diagnostic biomarkers, rigorous validation and the establishment of standardized multi-biomarker panels are still required to facilitate their clinical translation. Future research should focus on standardized comparisons of EVs from different cellular sources, clinically relevant polymicrobial pathogenic models, and prospective validation of EV-based diagnostic and therapeutic strategies.