Abstract
Background
Sonodynamic therapy (SDT) is increasingly investigated for cancer treatment. This study aimed to characterize the knowledge structure, thematic evolution, and publication trajectory of cancer SDT research.
Methods
English-language articles and reviews published from January 1, 2001, to December 31, 2025, were retrieved from the Web of Science Core Collection (WoSCC) on January 16, 2026. CiteSpace 6.4.R1 and VOSviewer 1.6.20 were used for collaboration, co-occurrence, co-citation, and burst analyses. Candidate Poisson log-link models with zero, one, or two changepoints were compared using the Akaike information criterion and Bayesian information criterion (BIC), with BIC as the principal selection criterion. Negative-binomial models and exclusion of 2025 were used for sensitivity analyses.
Results
A total of 1,053 publications were included. China had the highest publication count (819) and total citation count (36,281), whereas Israel had the highest citations per publication (110.17) among the ten most productive countries. The Chinese Academy of Sciences had the highest institutional publication count (104), with 10,691 citations and 102.80 citations per publication. Among the most prolific authors, Chen Yu had the highest publication count (40), whereas Liu Zhuang had the highest citations per publication (162.26). The primary Poisson model identified one changepoint after 2020, separating 2001–2020 from 2021–2025. Mean annual publication output was 11.0 during 2001–2020 and 166.8 during 2021–2025. Earlier research focused mainly on sonosensitizers and ultrasound-related mechanisms, whereas more recent work increasingly emphasized nanomaterial-assisted delivery, the tumor microenvironment, immunotherapy, oxidative stress, regulated cell death, and multimodal treatment
Conclusion
Cancer SDT research entered a substantially higher-output phase after 2020. In terms of research content, more recent work expanded beyond conventional sonosensitizers and ultrasound-related mechanisms to include nanomaterial-assisted delivery, tumor microenvironment modulation, immune regulation, and regulated cell death.
Keywords: Sonodynamic therapy, Bibliometric analysis, Changepoint analysis, Publication growth, Nanomaterials, Immunotherapy
Introduction
Sonodynamic therapy (SDT) is a noninvasive therapeutic approach in which ultrasound activates sonosensitizers within target tissues, leading to the generation of reactive oxygen species (ROS) and subsequent tumor cell injury [1–4]. Compared with light-dependent therapies, ultrasound provides greater tissue penetration and can be combined with other treatment modalities, extending the potential application of SDT to deep-seated tumors [1, 2, 5].
The antitumor effects of SDT are commonly attributed to acoustic cavitation and sonosensitizer activation. Acoustic cavitation can produce mechanical effects within target tissues, while activated sonosensitizers generate ROS, including singlet oxygen, superoxide anion, and hydroxyl radicals [1, 2, 6]. These processes have been associated with apoptosis, ferroptosis, immunogenic cell death, and other forms of cellular injury [6, 7]. Since the early 2000s, cancer-related SDT research has expanded from conventional sonosensitizers and basic mechanistic studies to nanomaterial-based sensitizers and delivery systems [5, 8, 9]. Combination strategies involving photodynamic therapy, chemodynamic therapy, and immunotherapy have received increasing attention [10–13], alongside research on tumor microenvironment modulation and regulated cell death [7, 14, 15].
Bibliometric analysis provides a quantitative framework for examining publication output, collaboration networks, citation structures, and the evolution of research themes [16]. Such approaches have been widely applied across biomedical research fields [17–22]. Previous reviews of SDT have focused mainly on mechanisms, sonosensitizer design, nanomaterials, and therapeutic applications [1, 2, 6, 8], while existing bibliometric research has mapped major research themes and development trends in the field [23]. Quantitative analysis of annual publication output and growth modeling can further characterize the temporal development of the field and identify stage-specific changes in publication trajectories.
This study aimed to (i) characterize publication patterns and collaboration networks across countries/regions, institutions, and authors, together with journal and disciplinary distributions; (ii) quantify annual publication trends and identify stage-specific changes using growth modeling; and (iii) map the temporal evolution of research themes and identify emerging research hotspots.
Methods
Data sources and search strategy
The Science Citation Index Expanded (SCI-EXPANDED) within the Web of Science Core Collection (WoSCC) was used because it provides standardized citation metadata and export formats suitable for bibliometric mapping. The search was executed on January 16, 2026, in Advanced Search mode and was limited to records published from January 1, 2001, through December 31, 2025. The searched fields were Title (TI) and Author Keywords (AK). The exact query was: #1 TI=(sonodynamic* OR sonosensitizer*) OR AK=(sonodynamic* OR sonosensitizer*); #2 TI=(cancer* OR tumor* OR tumour* OR neoplasm* OR malignan*) OR AK=(cancer* OR tumor* OR tumour* OR neoplasm* OR malignan*); final query: #1 AND #2. The search identified 1,072 records. Eligibility screening excluded 12 records with ineligible document types or publication status and seven non-English-language publications. The remaining 1,053 records comprised 922 articles and 131 reviews. Duplicate checking in CiteSpace identified no duplicate records (n = 0), leaving 1,053 records for analysis (Fig. 1). The final dataset represents WoSCC-indexed records identified through this TI/AK search strategy.
Fig. 1.
Literature identification, eligibility screening, and analysis workflow. The diagram summarizes the WoSCC search strategy, record exclusions, final bibliometric dataset, and analysis tools. WoSCC, Web of Science Core Collection; SCI-EXPANDED, Science Citation Index Expanded; TI, Title; AK, Author Keywords; *, wildcard character
Bibliometric and statistical analysis
A prespecified multi-tool workflow was used. CiteSpace 6.4.R1 was used for duplicate checking, keyword-burst detection, keyword timeline and timezone visualization, and cited-reference co-citation analysis. VOSviewer 1.6.20 was used to calculate country/region co-authorship metrics and construct the country/region, institutional, author, and author-keyword networks. bibliometric.com supplied the country/region chord diagram and geographic map retained in Fig. 3C and E. R 4.5.0 was used to reconstruct annual indicators; generate the country/region, institution, journal, publisher, disciplinary, and hierarchical Citation Topic panels; fit changepoint models; calculate uncertainty; and perform sensitivity analyses. Country/region labels were harmonized before country-level analysis. Full counting was used, self-citations were retained, and no VOSviewer thesaurus file was loaded. Detailed software settings, model outputs, and the R script are provided in the Supplementary Information.
Fig. 3.
Country/region output and co-authorship patterns. A Publication counts and shares for the top 10 countries/regions; (B) annual publication counts for those countries/regions, with point size and color intensity representing yearly output; (C) international co-authorship chord diagram; (D) VOSviewer country/region co-authorship network; (E) geographic co-authorship map. In the network and map panels, larger nodes indicate greater publication volume, thicker lines indicate stronger collaboration, and colors indicate clusters.
For VOSviewer analyses, the country/region co-authorship map required at least one document per country/region; institution and author maps required at least five documents per item; and author-keyword co-occurrence analysis required at least five occurrences, retaining 301 keywords. Default advanced normalization, layout, and clustering settings were retained. Network-specific thresholds, limits on the number of co-authors per document, connected-component handling, and eligible-item counts are reported in Supplementary Tables S1–S2. Node size represents item frequency, line thickness represents link strength, and color represents cluster membership or average publication year, depending on the visualization [24].
CiteSpace analyses covered 2001–2025 with one year per time slice. Keyword and cited-reference networks were analyzed separately using the g-index selection criterion (k = 25), with LRF = 2.5, L/N = 10, LBY = 5, e = 1.0, and no pruning for the displayed maps. Burst detection displayed the 25 strongest keyword and cited-reference bursts. Complete map-specific settings and network-output statistics are provided in Supplementary Tables S3–S4 [25, 26]. Annual publication counts and citation totals by publication-year cohort were reconstructed from the WoSCC PY and TC fields. For a given publication year, the citation total is the cumulative number of citations accrued by all records published in that year as of January 16, 2026; it does not represent citations received during that calendar year.
Outputs from the different software platforms were treated as complementary visualizations.
Annual publication counts were analyzed as count data. For year t in model-selected segment j, with first year sⱼ and last year eⱼ, the primary specification was a Poisson regression with a log link:
Yₜ ∼ Poisson(λₜ).
log(λₜ) = αⱼ + βⱼ(t − sⱼ), sⱼ ≤ t ≤ eⱼ.
Here, Yₜ denotes the observed number of publications in year t; λₜ is the expected annual count; sⱼ and eⱼ are the first and last years of segment j; αⱼ is the log expected count at sⱼ; and βⱼ is the segment-specific log-linear slope. Each segment had its own intercept and slope, so the specification allowed both level and slope changes at the boundaries and did not impose continuity. A reported changepoint year denotes the final year of the preceding segment. Segmented publication-growth models have been used to characterize changes in scientific output, and established changepoint methods provide a basis for selecting and estimating breakpoints [27–31]. Candidate models with zero, one, or two changepoints were searched exhaustively, with at least four years required in each segment.
As an overdispersion sensitivity analysis, the same segment-specific log-linear mean structure was fitted with a negative-binomial distribution:
Yₜ ∼ NB(µₜ, θ), log(µₜ) = αⱼ + βⱼ(t − sⱼ), sⱼ ≤ t ≤ eⱼ.
Var(Yₜ) = µₜ + µₜ²/θ.
Here, µₜ is the expected annual count and θ is the negative-binomial size parameter; under this parameterization, Var(Yₜ) = µₜ + µₜ²/θ.
For each model-selected segment, the annualized growth rate was calculated as:
AGRⱼ = [exp(βⱼ) − 1] × 100%.
The 95% Wald confidence interval for the annualized growth rate was obtained by transforming βⱼ ± 1.96 SE(βⱼ) through the same exponential function. Candidate models with zero, one, or two changepoints were compared using AIC and BIC, with estimated changepoint locations included in the parameter count. BIC was used to select the primary model. Full model-comparison procedures and outputs are provided in the Supplementary Information.
Model adequacy was assessed using Pearson and deviance dispersion statistics, studentized residuals, and Cook’s distance. Negative-binomial modeling and exclusion of 2025 were used as sensitivity analyses.
The timing of the selected publication-growth transition was compared descriptively with keyword-burst and co-citation patterns.
Results
Literature identification and inclusion
The search identified 1,072 records. After exclusion of 12 records with ineligible document types or publication status and seven non-English-language publications, 1,053 records remained. CiteSpace identified no duplicates. The final dataset comprised 922 articles and 131 reviews (Fig. 1).
Annual publication and citation trends
Annual publication output remained low during the early years, increased progressively through the 2010s, and rose markedly after 2020 (Fig. 2). The selected changepoint separated 2001–2020 from 2021 to 2025. Citation totals peaked for the 2021 cohort and declined for more recent cohorts because of shorter citation-accumulation periods.
Fig. 2.
Annual publication output and citation totals by publication year in cancer SDT research, 2001–2025. Purple bars show annual publication counts, the blue line shows cumulative citations accrued by each publication-year cohort as of January 16, 2026, and the dashed line marks the selected changepoint after 2020
Country/region publication and collaboration patterns
After harmonization of country/region names, China accounted for 819 publications (77.8% of the 1,053 records), followed by the United States with 67 (6.4%), Iran with 42 (4.0%), South Korea with 29 (2.8%), the United Kingdom with 26 (2.5%), and Italy with 26 (2.5%) (Fig. 3A; Table 1).
Table 1.
Top 10 countries/regions by publication volume
| Rank | Country/region | Documents (n)a | Total citationsb | Citations/publicationc | TLSd |
|---|---|---|---|---|---|
| 1 | China | 819 | 36,281 | 44.30 | 54 |
| 2 | United States | 67 | 4,644 | 69.31 | 45 |
| 3 | Iran | 42 | 1,029 | 24.50 | 6 |
| 4 | South Korea | 29 | 2,553 | 88.03 | 7 |
| 5 | United Kingdom | 26 | 1,396 | 53.69 | 10 |
| 6 | Italy | 26 | 979 | 37.65 | 9 |
| 7 | Japan | 22 | 1,178 | 53.55 | 2 |
| 8 | Singapore | 19 | 1,166 | 61.37 | 11 |
| 9 | Israel | 12 | 1,322 | 110.17 | 4 |
| 10 | Germany | 12 | 173 | 14.42 | 10 |
|
aCountries/regions were ranked by the number of included documents; ties were ordered by total citations. Full counting was used; a publication with authors from more than one country/region was credited to each participating country/region, so document totals are not mutually exclusive. b Total citations were accrued as of January 16, 2026 c Citations/publication was calculated as total citations divided by the number of publications; values are descriptive and were not normalized by publication year or research field d TLS, total link strength in the VOSviewer country/region co-authorship network | |||||
Figure 3B shows annual publication counts for the 10 leading countries/regions from 2001 to 2025; China had the highest annual publication output, particularly after 2020. The collaboration panels in Fig. 3C–E show co-authorship links between China and the United States, Singapore, Germany, the United Kingdom, and other countries.
China had the highest publication count (819), total citation count (36,281), and country/region TLS (54). Although Israel ranked ninth in publication output, it had the highest citations per publication among the top 10 countries/regions (110.17), followed by South Korea (88.03) and the United States (69.31) (Table 1).
Institutional publication and collaboration patterns
The top 10 institutions by publication volume were all located in China. The Chinese Academy of Sciences had the highest institutional publication count (104) (Fig. 4A; Table 2).
Fig. 4.
Institutional publication volume and co-authorship network. A Top 10 institutions by publication volume, using the same reconciled counts reported in Table 2; (B) VOSviewer institutional co-authorship network map. Node size represents publication volume, line thickness represents collaboration strength, and colors indicate clusters
Table 2.
Top 10 institutions by publication volume
| Rank | Institution | Country | Documents (n)a | Total citationsb | Citations/publicationc | TLSd |
|---|---|---|---|---|---|---|
| 1 | Chinese Academy of Sciences | China | 104 | 10,691 | 102.80 | 67 |
| 2 | Shanghai University | China | 53 | 2,371 | 44.74 | 46 |
| 3 | Chongqing Medical University | China | 45 | 1,764 | 39.20 | 6 |
| 4 | Shanghai Jiao Tong University | China | 44 | 2,186 | 49.68 | 34 |
| 5 | Tongji University | China | 42 | 2,730 | 65.00 | 35 |
| 6 | University of Science and Technology of China | China | 41 | 2,967 | 72.37 | 27 |
| 7 | Fudan University | China | 37 | 1,697 | 45.86 | 50 |
| 8 | Sun Yat-sen University | China | 37 | 1,229 | 33.22 | 7 |
| 9 | Huazhong University of Science and Technology | China | 33 | 2,309 | 69.97 | 7 |
| 10 | Zhejiang University | China | 32 | 1,323 | 41.34 | 9 |
|
aInstitutions were ranked by the number of included documents; ties were ordered by total citations. Full counting was used; a publication with more than one institutional affiliation was credited to each participating institution, so document totals are not mutually exclusive b Total citations were accrued as of January 16, 2026 c Citations/publication was calculated as total citations divided by the number of publications; values are descriptive and were not normalized by publication year or research field d TLS, total link strength in the VOSviewer institutional co-authorship network | ||||||
The Chinese Academy of Sciences had 104 publications, 10,691 total citations, 102.80 citations per publication, and the highest institutional TLS (67). Shanghai University had 53 publications, followed by Chongqing Medical University (45), Shanghai Jiao Tong University (44), and Tongji University (42). Figure 4B shows the principal institutional collaboration clusters.
Author publication and collaboration patterns
A total of 5,267 authors were involved in the relevant studies.
Chen Yu had the highest publication count (40), followed by Cheng Liang with 29 and Geng Bijiang with 23 (Table 3). Liu Zhuang and Wang Pan each had 19 publications; after tie-breaking by total citations, Liu Zhuang ranked eighth and had the highest citations per publication among the listed authors (162.26). Figure 5 shows 17 co-authorship clusters, with denser links within several author groups.
Table 3.
Top 10 authors by publication volume
| Rank | Author | Institution | Documents (n)a | Total citationsb | Citations/publicationc |
|---|---|---|---|---|---|
| 1 | Chen, Yu | Shanghai University | 40 | 4,881 | 122.03 |
| 2 | Cheng, Liang | Soochow University | 29 | 3,745 | 129.14 |
| 3 | Geng, Bijiang | Shanghai University | 23 | 1,012 | 44.00 |
| 4 | Wang, Zhigang | Chongqing Medical University | 21 | 1,258 | 59.90 |
| 5 | Lin, Jun | Chinese Academy of Sciences | 20 | 2,970 | 148.50 |
| 6 | Wang, Xiaobing | Shaanxi Normal University | 20 | 926 | 46.30 |
| 7 | Liu, Quanhong | Shaanxi Normal University | 20 | 836 | 41.80 |
| 8 | Liu, Zhuang | Soochow University | 19 | 3,083 | 162.26 |
| 9 | Wang, Pan | Shaanxi Normal University | 19 | 884 | 46.53 |
| 10 | Ma, Ping’an | Chinese Academy of Sciences | 18 | 2,340 | 130.00 |
|
a Authors were ranked by the number of included documents; ties were ordered by total citations. Full counting credited each co-authored publication to every listed author b Total citations were accrued as of January 16, 2026 c Citations/publication was calculated as total citations divided by the number of publications; values are descriptive and were not normalized by publication year or research field | |||||
Fig. 5.
VOSviewer author collaboration network map. Node size represents author publication volume, line thickness represents collaboration strength, and different colors represent different collaboration clusters (17 clusters in total)
Distribution of journals and publishers
Chemical Engineering Journal published the largest number of records (51), followed by ACS Applied Materials & Interfaces (49) and Advanced Functional Materials (33) (Fig. 6A). Among publishers, Elsevier accounted for 352 records, Wiley for 211, and the American Chemical Society for 125; the top 10 publishers together accounted for 88.8% of the included records (Fig. 6B).
Fig. 6.
Distribution of publication venues. A Top 12 journals by publication volume; (B) top 10 publishers by publication volume, with the line showing their cumulative share of the 1,053 included records
Distribution of research areas
Figure 7A reports the WoS Research Areas assigned to the records. Materials Science (484 publications), Science/Technology (408), and Chemistry (405) were the most frequent broad areas, followed by Engineering (178), Pharmacology/Pharmacy (151), Physics (148), and Oncology (69). Figure 7B reports the separate WoS Categories field, in which Nanoscience & Nanotechnology (393), Materials Science, Multidisciplinary (329), and Chemistry, Multidisciplinary (313) ranked highest.
Fig. 7.
Distribution of disciplinary classifications. A Top 20 WoS Research Areas by publication volume; (B) top 12 WoS Categories by publication volume. The panels represent distinct WoS classification systems
WoS Research Areas and WoS Categories are distinct indexing schemes with different levels of aggregation, so their counts are not directly comparable. The two panels provide complementary descriptions of the disciplinary profile. Their shared emphasis on materials science, chemistry, and nanoscience indicates strong representation of interdisciplinary materials and nanotechnology research in cancer SDT.
The disciplinary distribution places cancer SDT research at the interface of physical science, materials science, engineering, pharmacology, and oncology.
Evolution of the knowledge structure
Keyword co‑occurrence analysis
Author-keyword co-occurrence analysis was performed in VOSviewer. Keywords occurring at least five times were retained, resulting in 301 terms. Node size represents occurrence frequency, line thickness represents co-occurrence strength, and colors indicate clusters or temporal position. Tumor-type labels were derived from author-supplied keywords rather than standardized MeSH terms and were interpreted descriptively within clusters rather than used to generate an overall tumor-type ranking.
The density map (Fig. 8A) was centered on “sonodynamic therapy” and “ultrasound.” Additional high-density regions included “nanoparticles,” “drug delivery,” “combination therapy,” “immunotherapy,” “immune checkpoint blockade,” “photodynamic therapy,” “focused ultrasound,” and “ferroptosis.” The temporal-overlay visualization (Fig. 8B) and CiteSpace timezone view (Fig. 8D) showed earlier prominence of foundational terms such as sonodynamic therapy, photodynamic therapy, focused ultrasound, hematoporphyrin, and cancer, whereas tumor microenvironment, immunotherapy, immune checkpoint blockade, nanocarriers, ferroptosis, hypoxia, and oxidative stress became more prominent in later years.
Fig. 8.
Author-keyword structure and temporal evolution. A VOSviewer density visualization; (B) VOSviewer temporal-overlay visualization; (C) VOSviewer seven-cluster co-occurrence map; (D) CiteSpace keyword timezone visualization, 2001–2025. Warmer colors in panel A indicate higher keyword density, whereas colors in panel B progress from earlier to later average publication years
The seven-cluster co-occurrence map (Fig. 8C) showed distinct thematic groupings. Cluster 1 (red; 70 keywords) emphasized inorganic nanomaterials, including metal–organic frameworks, black phosphorus, MXenes, bismuth, and manganese dioxide, together with terms related to glutathione depletion, ferroptosis, cuproptosis, and gas therapy. Cluster 2 (green; 59 keywords) emphasized combination therapy, nanocarriers, liposomes, nanobubbles, hypoxia, and the tumor microenvironment. Cluster 3 (blue; 51 keywords) contained immunotherapy-related terms, including antitumor immunity, immune checkpoint blockade, cGAS–STING signaling, and immunogenic cell death.
Cluster 4 (yellow; 50 keywords) linked photodynamic and sonodynamic therapy through photosensitizer terms such as hematoporphyrin, phthalocyanine, and porphyrin. Breast cancer, colorectal cancer, and melanoma appeared as application-related terms within this cluster and were interpreted as cluster-specific terms rather than as a dataset-wide tumor-type ranking. The raw keyword “invitro” was standardized to “in vitro” during manual cleaning. Cluster 5 (purple; 30 keywords) emphasized tumor-biology processes and drug mechanisms; Cluster 6 (cyan; 27 keywords) emphasized focused ultrasound, microbubbles, glioma, glioblastoma, and the blood–brain barrier; and Cluster 7 (orange; 14 keywords) emphasized oxidative stress, apoptosis, autophagy, ferroptosis, pyroptosis, and synergistic therapy.
Keyword burst analysis
Keyword bursts identify terms whose frequency increases sharply during a defined interval. Bursts beginning in 2001–2004 included “hematoporphyrin,” “focused ultrasound,” “gallium porphyrin complex,” “enhancement,” and “mechanism,” reflecting early interest in sensitizers, acoustic approaches, and basic mechanisms.
Bursts beginning in 2007–2013 included “protoporphyrin IX,” “cavitation,” “in vitro,” “damage,” “apoptosis,” “5-aminolevulinic acid,” and “low intensity ultrasound.” More recent bursts, with onset years between 2015 and 2022, included “photodynamic therapy,” “delivery,” “TiO₂ nanoparticles,” “microbubbles,” “sinoporphyrin sodium,” “release,” “breaking,” “singlet oxygen,” and “oxidative stress.”
Keyword-burst patterns (Fig. 9A, B) showed a gradual thematic shift from early sensitizer and acoustic mechanism research toward nanomaterial-enabled delivery, combination therapy, immune regulation, and regulated cell death.
Fig. 9.
CiteSpace keyword-burst and timeline analyses. A Top 25 keywords with the strongest bursts; red segments indicate burst intervals. B Keyword timeline map
Reference co-citation analysis
CiteSpace identified nine major co-cited-reference clusters (Fig. 10A). Clusters labeled “sonosensitizers” and “5-aminolevulinic acid” represented sensitizer development and evaluation; “oxygen evolution reaction” represented ROS-generation mechanisms; “cuproptosis” and “immunogenic cell death” represented emerging cell death and immune themes; and “immunotherapy,” “photothermal therapy,” and “trastuzumab” represented combination or targeted approaches.
Fig. 10.
Co-cited-reference cluster and burst analyses. A Nine-cluster co-citation map; (B) top 25 references with the strongest citation bursts
The co-citation structure was organized around sensitizer design, ROS generation, regulated cell death, and multimodal therapy.
The strongest reference bursts (Fig. 10B) included Qian et al. [32] and Deepagan et al. [33], both published in 2016 and focused on nanomaterial-assisted SDT. Earlier bursts included studies of basic SDT mechanisms and ultrasound-mediated therapy [34, 35], whereas later bursts increasingly involved multifunctional nanoplatforms, combination therapy, and newer sonosensitizers [36, 37].
Among the 1,040 records assigned a Web of Science Citation Topic, the dominant meso-level topics were Nanoparticles (649; 62.4%) and Ultrasound in Medicine (204; 19.6%), which mapped primarily to the micro-level topics Magnetic Nanoparticles (642; 61.7%) and Microbubbles (204; 19.6%), respectively (Fig. 11). Lower-frequency topics were grouped as Other in the hierarchical display.
Fig. 11.
Hierarchical flow of Web of Science Citation Topics. Meso-level topics are shown on the left and micro-level topics on the right. Ribbon widths are proportional to publication counts among the 1,040 records assigned a Citation Topic; lower-frequency categories are combined as Other
Changepoint analysis of publication growth
Primary changepoint model
BIC selected a one-changepoint Poisson model with a boundary after 2020 (BIC = 151.1), defining the periods 2001–2020 and 2021–2025. Complete comparisons of the zero-, one-, and two-changepoint models and all sensitivity models are provided in the Supplementary Information.
Segment-specific estimates and uncertainty
The log-link slopes were 0.200 for 2001–2020 and 0.177 for 2021–2025. These corresponded to annualized growth rates of 22.2% (95% CI, 18.3%–26.1%) and 19.3% (95% CI, 13.6%–25.3%), respectively. The observed publication totals were 219 and 834, yielding mean annual outputs of 11.0 and 166.8 publications (Table 4; Fig. 12).
Table 4.
Segment-specific estimates from the selected Poisson log-link changepoint model for annual publication counts
| Period | αⱼ (SE)ᵃ | exp(αⱼ)ᵇ | βⱼ (SE)ᶜ | Annualized growth (%)ᵈ | 95% CIᵉ | Mean annual outputᶠ |
|---|---|---|---|---|---|---|
| 2001–2020 | −0.102 (0.251) | 0.90 | 0.200 (0.016) | 22.2% | 18.3%–26.1% | 11.0 |
| 2021–2025 | 4.732 (0.068) | 113.54 | 0.177 (0.025) | 19.3% | 13.6%–25.3% | 166.8 |
|
Abbreviations: CI, confidence interval; SE, standard error Values in parentheses are standard errors a For segment j, sⱼ denotes the first year of the segment, and αⱼ is the log expected publication count at sⱼ b exp(αⱼ) is the expected publication count in the first year of the segment c βⱼ is the segment-specific log-linear slope d The annualized growth rate was calculated as [exp(βⱼ) − 1] × 100% e The 95% Wald confidence interval was obtained by transforming βⱼ ± 1.96 × SE(βⱼ) through the same exponential function f Mean annual output is the arithmetic mean of the observed annual publication counts within each model-selected segment | ||||||
Fig. 12.
Selected changepoint model and annual publication growth. A Observed annual counts and fitted means from the one-changepoint Poisson model, with a boundary after 2020; (B) breakpoint-adjusted BIC values for Poisson and negative-binomial models with zero, one, or two changepoints; (C) segment-specific annualized growth rates with 95% Wald confidence intervals; (D) mean annual publication output in the selected periods
Model diagnostics and sensitivity analyses
The Pearson and deviance dispersion statistics were 1.36 and 1.38, respectively. The maximum absolute studentized residual was 2.54 in 2011, and the largest Cook’s distance was 0.758 in 2020. After exclusion of 2025, the 2020 boundary remained selected and the final-period growth estimate was 19.1% (95% CI, 10.8%–28.0%). Complete diagnostic and sensitivity results are provided in the Supplementary Information.
Discussion
Sonodynamic therapy (SDT) uses ultrasound to activate sonosensitizers and generate reactive oxygen species (ROS), enabling therapeutic effects in relatively deep tissues with spatial control [1, 2]. In oncology, SDT has been investigated in several solid tumors, including glioma and breast cancer [4, 12]. Recent studies have increasingly examined SDT in combination with immune-based treatment strategies, particularly in breast cancer and brain tumors [13, 38]. Research in this field has expanded from the early evaluation of sonosensitizers and ultrasound-induced cytotoxicity to drug delivery systems, tumor microenvironment modulation, combination therapy, and immune regulation.
Nanotechnology has broadened the design of SDT systems. Nanosonosensitizers, liposomes, microbubbles, metal–organic frameworks, and stimuli-responsive delivery systems have been developed to improve sonosensitizer delivery, tumor accumulation, and ROS generation [5, 39]. Stimuli-responsive nanosonosensitizers can exploit tumor-specific microenvironmental conditions to improve treatment selectivity [8]. Hypoxia, redox regulation, and the tumor microenvironment have also become important mechanistic considerations [40, 41]. SDT-induced immunogenic cell death and ferroptosis have linked oxidative injury with antitumor immune responses [7], while nanoplatform-based SDT has been explored for modulation of the tumor immune microenvironment [10]. Sono-immunotherapy and immune checkpoint blockade have also emerged as active areas of investigation [37, 42].
In the present study, formal count-based changepoint modeling identified a boundary after 2020 in the publication trajectory of cancer SDT research from 2001 to 2025. The model separated 2001–2020 from 2021 to 2025, with mean annual publication outputs of 11.0 and 166.8, respectively. The substantially higher mean annual output during 2021–2025 indicates that cancer SDT research entered a higher-output phase after 2020.
Research themes also evolved over time. Earlier studies focused mainly on sonosensitizers, cavitation, ROS, apoptosis, and ultrasound-related mechanisms [43], whereas more recent work increasingly emphasized the tumor microenvironment, immunotherapy, immune checkpoint blockade, ferroptosis, hypoxia, and oxidative stress. Temporal keyword analyses showed greater prominence of these topics in later years, while keyword-burst analysis highlighted delivery systems and nanomaterial-related strategies. Co-citation analysis further identified immunogenic cell death, cuproptosis, immunotherapy, and multimodal treatment as components of the recent knowledge structure.
The thematic evolution reflects increasingly integrated biological and therapeutic questions. Early studies primarily examined whether ultrasound could activate sonosensitizers and induce ROS-mediated tumor cell injury [44]. Subsequent research increasingly addressed sonosensitizer delivery, hypoxia and tumor microenvironment modulation, regulated cell death, and antitumor immunity. Nanomaterial-based platforms have allowed drug delivery, microenvironment modulation, and combination therapy to be incorporated within the same therapeutic system [45]. Increasing interest in immunogenic cell death, ferroptosis, and cuproptosis has also extended the biological scope of SDT beyond conventional apoptosis and direct oxidative injury [37].
Publication output increased markedly after 2020, while research themes in more recent years broadened from sonosensitizers and basic acoustic mechanisms toward nanomaterial-assisted delivery, tumor microenvironment modulation, immune regulation, and multimodal treatment. Changepoint modeling characterized the change in publication trajectory, whereas keyword and co-citation analyses described the evolution of research emphasis.
Limitations
This study has three main limitations. First, the analysis was restricted to WoSCC records retrieved through Title and Author Keyword fields, so relevant studies indexed elsewhere or described only in abstracts may have been missed. Second, citation totals, full-counting networks, and the 2025 publication count are affected by publication age, database indexing, and analytical thresholds. Third, changepoint estimates are based on a short annual series and identify changes in publication counts rather than their causes; updated analyses will be needed as additional years of data become available.
Conclusions
Cancer sonodynamic therapy research expanded substantially from 2001 to 2025. Formal changepoint modeling identified a boundary in the publication trajectory after 2020, with substantially higher research output during 2021–2025. Temporal keyword, keyword-burst, and co-citation analyses showed that, in more recent years, research emphasis broadened beyond conventional sonosensitizers, cavitation, and ROS-related mechanisms to include nanomaterial-assisted delivery, the tumor microenvironment, immunotherapy, and regulated cell death. These findings characterize the recent evolution of cancer SDT research in terms of both publication trajectory and thematic development.
Acknowledgements
Not applicable.
Author contributions
WANG Xiaoting: literature search, data curation, investigation, visualization, writing – original draft. YANG Ke: methodology, software analysis, validation, writing, review, and editing. DENG Peitong: literature search, data curation. ZHANG Bo: funding acquisition, writing – review & editing. All the authors have read and approved the final manuscript.All authors reviewed the manuscript.
Funding
This study was supported by the National Natural Science Foundation of China (General Program) under Grant No. 22478448.
Data availability
No datasets were generated or analysed during the current study.
Code availability
The Supplementary Information includes the annual publication-count table and the R script used for the statistical analyses. No new software package was developed.
Declarations
Competing interests
The authors declare no competing interests.
Ethical approval
This study did not involve human participants, human tissue, animal experiments, or clinical intervention. Therefore, ethical approval was not required.
Consent to participate
This study did not involve human participants. All data were obtained from bibliographic databases and published literature records. Therefore, informed consent to participate was not required.
Consent to publish
Not applicable.
Footnotes
Publisher’s note
Springer nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Wan GY, Liu Y, Chen BW, et al. Recent advances of sonodynamic therapy in cancer treatment. Cancer Biol Med. 2016;13(3):325–38. 10.20892/j.issn.2095-3941.2016.0068. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Lafond M, Yoshizawa S, Umemura SI. Sonodynamic therapy: advances and challenges in clinical translation. J Ultrasound Med. 2019;38(3):567–80. 10.1002/jum.14733. [DOI] [PubMed] [Google Scholar]
- 3.Dolmans DE, Fukumura D, Jain RK. Photodynamic therapy for cancer. Nat Rev Cancer. 2003;3(5):380–7. 10.1038/nrc1071. [DOI] [PubMed] [Google Scholar]
- 4.An YW, Liu HQ, Zhou ZQ, et al. Sinoporphyrin sodium is a promising sensitizer for photodynamic and sonodynamic therapy in glioma. Oncol Rep. 2020;44(4):1596–604. 10.3892/or.2020.7695. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Yang M, Wang X, Peng M, et al. Nanomaterials enhanced sonodynamic therapy for multiple tumor treatment. Nanomicro Lett. 2025;17(1):157. 10.1007/s40820-025-01666-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Zhang M, Sun D, Huang H, et al. Nanosonosensitizer optimization for enhanced sonodynamic disease treatment. Adv Mater. 2024;36(46):e2409663. 10.1002/adma.202409663. [DOI] [PubMed] [Google Scholar]
- 7.Yuan H, Ma J, Huang W, et al. Antitumor effects of a distinct sonodynamic nanosystem through enhanced induction of immunogenic cell death and ferroptosis with modulation of tumor microenvironment. JACS Au. 2023;3(5):1507–20. 10.1021/jacsau.3c00156. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Jiang Z, Xiao W, Fu Q. Stimuli-responsive nanosonosensitizers for sonodynamic therapy. J Control Release. 2023;361:547–67. 10.1016/j.jconrel.2023.08.003. [DOI] [PubMed] [Google Scholar]
- 9.Dong C, Yang P, Wang X, et al. Multifunctional composite nanosystems for precise/enhanced sonodynamic oxidative tumor treatment. Bioconjug Chem. 2022;33(6):1035–48. 10.1021/acs.bioconjchem.1c00478. [DOI] [PubMed] [Google Scholar]
- 10.Zheng J, Sun Y, Long T, et al. Sonosensitizer nanoplatform-mediated sonodynamic therapy induced immunogenic cell death and tumor immune microenvironment variation. Drug Deliv. 2022;29(1):1164–75. 10.1080/10717544.2022.2058653. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Zheng H, Yin N, Lv K, et al. Defect-rich sonosensitizers based on CeO2 with Schottky heterojunctions for boosting sonodynamic/chemodynamic synergistic therapy. J Mater Chem B. 2024;12(17):4162–71. 10.1039/D4TB00084F. [DOI] [PubMed] [Google Scholar]
- 12.Wang P, Li C, Wang X, et al. Anti-metastatic and pro-apoptotic effects elicited by combination photodynamic therapy with sonodynamic therapy on breast cancer both in vitro and in vivo. Ultrason Sonochem. 2015;23:116–27. 10.1016/j.ultsonch.2014.10.027. [DOI] [PubMed] [Google Scholar]
- 13.Li X, He G, Jin H, et al. Ultrasound-activated precise sono-immunotherapy for breast cancer with reduced pulmonary fibrosis. Adv Sci (Weinh). 2025;12(5):e2407609. 10.1002/advs.202407609. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Singh N, Kim D, Min S, et al. Multimodal synergistic ferroptosis cancer therapy. Coord Chem Rev. 2025;522:216236. 10.1016/j.ccr.2024.216236. [DOI] [Google Scholar]
- 15.Yu P, Zhang X, Liu N, et al. Pyroptosis: mechanisms and diseases. Signal Transduct Target Ther. 2021;6(1):128. 10.1038/s41392-021-00507-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Aria M, Cuccurullo C. bibliometrix: an R-tool for comprehensive science mapping analysis. J Informetr. 2017;11(4):959–75. 10.1016/j.joi.2017.08.007. [DOI] [Google Scholar]
- 17.Wu H, Wang Y, Tong L, et al. The global research trends and hotspots on developmental dysplasia of the hip: a bibliometric and visualized study. Front Surg. 2021;8:671403. 10.3389/fsurg.2021.671403. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Xia D, Wu R, Wang S, et al. Global trends and prospects regarding exosomes in cancer immunology research over the past 10 years. Technol Cancer Res Treat. 2023;22:15330338231199892. 10.1177/15330338231199892. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Wan Y, Shen J, Hong Y, et al. Mapping knowledge landscapes and emerging trends of the biomarkers in melanoma: a bibliometric analysis from 2004 to 2022. Front Oncol. 2023;13:1181164. 10.3389/fonc.2023.1181164. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Wan Y, Shen J, Ouyang J, et al. Bibliometric and visual analysis of neutrophil extracellular traps from 2004 to 2022. Front Immunol. 2022;13:1025861. 10.3389/fimmu.2022.1025861. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Ma D, Yang B, Guan B, et al. A bibliometric analysis of pyroptosis from 2001 to 2021. Front Immunol. 2021;12:731933. 10.3389/fimmu.2021.731933. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Wu H, Tong L, Wang Y, et al. Bibliometric analysis of global research trends on ultrasound microbubble: a quickly developing field. Front Pharmacol. 2021;12:646626. 10.3389/fphar.2021.646626. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Wu Z, Cheng K, Shen Z, et al. Mapping knowledge landscapes and emerging trends of sonodynamic therapy: a bibliometric and visualized study. Front Pharmacol. 2023;13:1048211. 10.3389/fphar.2022.1048211. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.van Eck NJ, Waltman L. Software survey: VOSviewer, a computer program for bibliometric mapping. Scientometrics. 2010;84(2):523–38. 10.1007/s11192-009-0146-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Synnestvedt MB, Chen C, Holmes JH. CiteSpace II: visualization and knowledge discovery in bibliographic databases. AMIA Annu Symp Proc. 2005;2005:724–728. PMID:16779135IF: NA NA NA. [PMC free article] [PubMed]
- 26.Chen C. CiteSpace II: detecting and visualizing emerging trends and transient patterns in scientific literature. J Am Soc Inf Sci Technol. 2006;57(3):359–77. 10.1002/asi.20317. [DOI] [Google Scholar]
- 27.Bornmann L, Haunschild R, Mutz R. Growth rates of modern science: a latent piecewise growth curve approach to model publication numbers from established and new literature databases. Humanit Soc Sci Commun. 2021;8:224. 10.1057/s41599-021-00903-w. [DOI] [Google Scholar]
- 28.Su Y, Han LF. A new literature growth model: variable exponential growth law of literature. Scientometrics. 1998;42(2):259–65. 10.1007/BF02458359. [DOI] [Google Scholar]
- 29.Muggeo VMR. Estimating regression models with unknown break-points. Stat Med. 2003;22(19):3055–71. 10.1002/sim.1545. [DOI] [PubMed] [Google Scholar]
- 30.Kim J, Kim HJ. Consistent model selection in segmented line regression. J Stat Plan Inference. 2016;170:106–16. 10.1016/j.jspi.2015.09.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Truong C, Oudre L, Vayatis N. Selective review of offline change point detection methods. Signal Process. 2020;167:107299. 10.1016/j.sigpro.2019.107299. [DOI] [Google Scholar]
- 32.Qian X, Zheng Y, Chen Y. Micro/nanoparticle-augmented sonodynamic therapy (SDT): breaking the depth shallow of photoactivation. Adv Mater. 2016;28(37):8097–129. 10.1002/adma.201602012. [DOI] [PubMed] [Google Scholar]
- 33.Deepagan VG, You DG, Um W, et al. Long-circulating Au-TiO2 nanocomposite as a sonosensitizer for ROS-mediated eradication of cancer. Nano Lett. 2016;16(10):6257–64. 10.1021/acs.nanolett.6b02547. [DOI] [PubMed] [Google Scholar]
- 34.Gao Z, Zheng J, Yang B, et al. Sonodynamic therapy inhibits angiogenesis and tumor growth in a xenograft mouse model. Cancer Lett. 2013;335(1):93–9. 10.1016/j.canlet.2013.02.006. [DOI] [PubMed] [Google Scholar]
- 35.Wood AKW, Sehgal CM. A review of low-intensity ultrasound for cancer therapy. Ultrasound Med Biol. 2015;41(4):905–28. 10.1016/j.ultrasmedbio.2014.11.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Pan X, Bai L, Wang H, et al. Metal-organic-framework-derived carbon nanostructure augmented sonodynamic cancer therapy. Adv Mater. 2018;30(23):e1800180. 10.1002/adma.201800180. [DOI] [PubMed] [Google Scholar]
- 37.Zeng Z, Zhang C, He S, et al. Activatable cancer sono-immunotherapy using semiconducting polymer nanobodies. Adv Mater. 2022;34(28):e2203246. 10.1002/adma.202203246. [DOI] [PubMed] [Google Scholar]
- 38.Hutton DL, Burns TC, Hossain-Ibrahim K. A review of sonodynamic therapy for brain tumors. Neurosurg Focus. 2024;57(3):E7. 10.3171/2024.6.FOCUS24338. [DOI] [PubMed] [Google Scholar]
- 39.Du Z, Chen W, Tian M, et al. Aggregation-induced emission sonosensitizers: molecular engineering and biomedical applications. Chem Commun (Camb). 2025;61(80):15561–74. 10.1039/D5CC04713G. [DOI] [PubMed] [Google Scholar]
- 40.Zhu P, Chen Y, Shi J. Nanoenzyme-augmented cancer sonodynamic therapy by catalytic tumor oxygenation. ACS Nano. 2018;12(4):3780–95. 10.1021/acsnano.8b00999. [DOI] [PubMed] [Google Scholar]
- 41.Gong F, Cheng L, Yang N, et al. Ultrasmall oxygen-deficient bimetallic oxide MnWOx nanoparticles for depletion of endogenous GSH and enhanced sonodynamic cancer therapy. Adv Mater. 2019;31(23):e1900730. 10.1002/adma.201900730. [DOI] [PubMed] [Google Scholar]
- 42.Du Y, Yang J, Xu A, et al. Research progress on the induction of immunogenic cell death in tumor immunotherapy using a sonodynamic therapy nanoparticle delivery system. Front Immunol. 2025;16:1681773. 10.3389/fimmu.2025.1681773. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Araújo Martins Y, Zeferino Pavan T, Fonseca Vianna Lopez R. Sonodynamic therapy: ultrasound parameters and in vitro experimental configurations. Int J Pharm. 2021;610:121243. 10.1016/j.ijpharm.2021.121243. [DOI] [PubMed] [Google Scholar]
- 44.Chen H, Zhou X, Gao Y, et al. Recent progress in development of new sonosensitizers for sonodynamic cancer therapy. Drug Discov Today. 2014;19(4):502–9. 10.1016/j.drudis.2014.01.010. [DOI] [PubMed] [Google Scholar]
- 45.Wu X, Chen F, Zhang Q, et al. What is the magical cavitation bubble: a holistic perspective to trigger advanced bubbles, nano-sonocatalysts, and cellular sonosensitizers. BME Front. 2024;5:0067. 10.34133/bmef.0067. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
No datasets were generated or analysed during the current study.
The Supplementary Information includes the annual publication-count table and the R script used for the statistical analyses. No new software package was developed.