Research trends and hotspots of upconversion nanoparticles for drug delivery and biomedical targeting in photodynamic therapy

Abstract

Aim

This study presents a bibliometric analysis of research trends and emerging hotspots in upconversion nanoparticles (UCNPs) for drug delivery and biomedical targeting in photodynamic therapy (PDT).

Methods

We retrieved 474 publications from the Web of Science (2007–2025) and analyzed publication trends, contributions from countries, institutions, and authors, keyword co-occurrence, citation bursts, and co-citation networks using CiteSpace, VOSviewer, HistCite, and R software.

Results

UCNP-PDT research progressed through three distinct phases, with China and the USA contributing the most publications. Nanoscale published the highest number of articles, while high-impact reviews appeared in Chemical Society Reviews and Advanced Materials. Emerging research focuses include the tumor microenvironment, metal–organic frameworks, combination therapies, and antimicrobial PDT.

Conclusion

UCNP-PDT has evolved from simple photosensitizer loading to multifunctional nanoplatforms, expanding applications to cancer therapy, infection control, and immune modulation. Clinical translation remains limited by size–efficiency trade-offs, low-power irradiation, hypoxia, tissue heating under 980 nm excitation, GMP-compliant production, and insufficient long-term biosafety data. Future advances require coordinated strategies integrating material optimization, mechanistic insight, and translational considerations to achieve safe, scalable, and effective UCNP-PDT for clinical use.

Supplementary Information

The online version contains supplementary material available at 10.1186/s11671-025-04402-8.

Introduction

Photodynamic therapy (PDT) is a well-established treatment modality that employs light to activate photosensitizers (PSs) within target tissues. Upon excitation, PSs generate reactive oxygen species (ROS), leading to localized cytotoxicity and therapeutic effects in conditions such as cancer and wound healing [1–3]. Unlike traditional ionizing radiation therapies, PDT utilizes non-ionizing light, which is harmless to normal tissues that have not absorbed PSs. In the absence of light, PSs exhibit minimal toxicity, reducing long-term adverse effects [4]. PDT offers advantages over conventional therapies, including minimal invasiveness and precise spatiotemporal control, which together contribute to improved patient tolerance and overall treatment acceptability [5, 6]. According to predictions based on GLOBOCAN 2022 data, the number of cancer cases may rise to 35 million by 2050, highlighting the urgency for novel treatments [7]. Combining PDT with other therapies can enhance efficacy, supporting its role as a complementary or alternative cancer treatment [8].

The clinical application of PDT primarily entails the interaction among PSs, light, and oxygen, with its efficacy largely contingent upon the precise and targeted delivery of PSs. PDT can generally be grouped into 3 types: superficial PDT for skin lesions with shallow light penetration [9], interstitial PDT using fibers or catheters for tumors beyond 1 cm [10, 11], and deep PDT, which seeks to overcome light penetration limits. In this context, upconversion nanoparticles (UCNPs) are particularly advantageous for deep PDT. They can convert tissue-penetrating near-infrared (NIR) light into visible or UV light, thereby enabling localized activation of PSs in otherwise inaccessible regions [12]. Beyond their optical conversion ability, UCNPs also function as versatile drug carriers. Their surfaces can be modified with hydrophilic groups to improve biocompatibility or with biomolecules such as folic acid, DNA, or enzymes for targeted delivery [13, 14]. Compared with traditional fluorescent probes like quantum dots and organic dyes, UCNPs exhibit high chemical stability, sharp emission bands, and excellent photostability, while avoiding photobleaching and autofluorescence interferences [15–17]. Additionally, UCNPs can be combined with other materials, such as magnetic nanoparticles [18] or photosensitive semiconductors like TiO₂ [19], to augment ROS production and achieve multimodal imaging and therapeutic capabilities, thereby optimizing the efficacy of PDT while minimizing damage to healthy tissues. Recent advancements include integrating UCNPs with porous materials such as metal–organic frameworks (MOFs) and porous organic polymer nanofibers [20], further expanding their potential in PDT drug delivery systems.

UCNPs have demonstrated efficacy in treating solid tumors [21–23], skin diseases [24, 25], bacterial infections [26–28], cardiovascular conditions [29], and autoimmune disorders like rheumatoid arthritis [30]. Despite significant interest in the expanded applications of UCNPs for PDT, there is a lack of systematic analysis regarding the progress, trends, and emerging topics in this field. In this study, we perform a bibliometric analysis of UCNP-PDT for drug delivery and biomedical targeting from 2007 to 2025 using the Web of Science (WOS) database. We analyze annual publications, countries, institutions, authors, journals, keyword co-occurrence, and co-citation to provide a global overview of research activity. Keyword burst analysis is further applied to identify hotspots and emerging trends. We also summarize the main obstacles limiting drug delivery and biomedical targeting applications of UCNP-PDT. Together, these findings offer a visualized map of the field, highlight emerging research hotspots—including MOFs, tumor microenvironment (TME), combination therapy, and antimicrobial photodynamic therapy (aPDT)—and provide guidance for future studies and clinical translation.

Methods

Data source and search strategies

WOS was selected as the primary database for this study due to its extensive coverage of over 12,000 academic journals and its widespread use among researchers. Compared to other databases such as Scopus, Medline, and PubMed, WOS offers the most comprehensive and reliable platform for bibliometric analysis [31]. Consequently, a topic search was conducted within the Web of Science Core Collection, targeting articles published from January 1, 2007, to September 14, 2025. The citation indexes included were the Science Citation Index Expanded and the Social Sciences Citation Index. The search was limited to publications in English and included only articles and reviews. Our search strategy combined Medical Subject Headings terms and keywords such as "Photodynamic Therapy," "Upconversion Nanoparticles," "Drug Delivery System," "Cancer," "Near-Infrared," "Targeting Agents," "Targeted," and "Therapy." The comprehensive search terms employed are detailed in Supporting Information. Manual screening was performed based on titles, abstracts, and, when necessary, full texts to exclude studies that were not substantially related to PDT or UCNPs, or that focused only on UCNP synthesis, luminescence properties, or non-therapeutic applications such as bioimaging and biosensing. The screening process is summarized in Fig. 1. Relevant articles were exported and stored in plain text format (.txt), including the full record and cited references for further analysis.

Fig. 1.

Flowchart of publication screening

Statistical analysis

To minimize interference from duplicated entities such as authors, institutions, and keywords, the text data downloaded from the WOS were meticulously processed. This involved data conversion, deduplication, merging, cleaning, and re-exporting. All relevant data were then imported into Microsoft Office Excel 2019, R software (version 4.2.2) [32], HistCite Pro (version 12.03.17) [33], VOSviewer (version 1.6.20) [34], and CiteSpace (version 6.4.R1) [35] for visual analysis.

HistCite was specifically used to evaluate the contributions of various institutions to the literature by analyzing publication counts, along with the total local citation score (TLCS) and the total global citation score (TGCS) for each publication year.

VOSviewer was used to generate visual representations of the bibliometric network, showcasing categorized authors, institutions, keywords, and cited references. In these visualizations, node colors represent different clusters, while font size and node size correlate with frequency of occurrence. The lines connecting nodes indicate the strength of their relationships. Similar principles were applied in the analyses performed with CiteSpace.

CiteSpace was utilized for a comprehensive analysis of UCNPs for drug delivery in PDT. This tool categorized and identified highly cited keywords and references, showing significant citation bursts within specific periods. Clustering results were deemed satisfactory when the mean silhouette score exceeded 0.5 and the modularity Q was above 0.3. Intermediary centrality quantifies the frequency with which an element serves as a conduit in the shortest path between other elements, thereby reflecting the element's significance. Higher intermediary centrality denotes pivotal nodes that are highly interconnected, as represented by the purple outer circle in the map. Additionally, Microsoft Office Excel 2019 and R software (version 4.2.2) were used for plotting graphs and performing descriptive statistics.

Results

Analysis of temporal distribution

Initially, 1205 articles were identified according to the established screening criteria between January 1, 2007, and September 14, 2025. Following an extensive review of the literature, an additional 63 relevant studies that were not covered in the initial search were included. Ultimately, 474 studies were retained for analysis, which included 298 articles and 176 reviews (Fig. 1). Figure 2 illustrates the annual publication output related to UCNP-PDT drug delivery and biomedical targeting. The number of papers increased from only a few in 2007 to more than 60 around 2020, corresponding to a compound annual growth rate of about 9.35%. This trend indicates a substantial expansion of research activity in this field, despite fluctuations in recent years. The growth in the number of publications during this period can be divided into three stages: the first stage (2007—2011), with relatively few publications, reflecting the initial exploration of UCNPs as carriers for PDT; the second stage (2012—2017), marked by a steady expansion from 9 to 46 papers, demonstrating growing research interest; and the third stage (2018—2025), during which publication numbers peaked at 54 in 2019 but subsequently declined to 10 in 2025. This overall trajectory indicates that research on UCNP-PDT drug delivery and biomedical targeting has attracted substantial scientific attention, while the recent decline in output may reflect a shift in research focus or early signs of field maturation. The list of the 474 publications is provided in Table S1.

Fig. 2.

Annual publication output of upconversion nanoparticles for drug delivery and biomedical targeting in photodynamic therapy

Analysis of most productive countries/regions

We analyzed publication counts by country and region to identify the main contributors to research on using UCNP-PDT drug delivery and biomedical targeting. A total of 33 countries have participated in this field of study. Figure 3a illustrates the leading 10 countries in publication output for this domain, with China contributing the largest number of publications (n = 358, 58.02%), followed by the USA (n = 58, 9.04%), Singapore (n = 45, 7.29%), South Korea (n = 21, 3.40%), and others (Table 1). Figure 3b illustrates the collaborative networks among these countries. For instance, China has established close collaborations with the USA, Singapore, and Saudi Arabia, while Singapore maintains active partnerships with the USA. Overall, this figure highlights the frequency of international collaborations, potentially indicating shared research interests between countries.

Fig. 3.

Leading countries/regions and collaboration in upconversion nanoparticles for drug delivery and biomedical targeting in photodynamic therapy. a Top 10 countries by publication volume; b country collaborations, with line thickness correlating with the intensity of closeness

Table 1.

Top 10 central countries studying upconversion nanoparticles for drug delivery and biomedical targeting in photodynamic therapy

Rank	Country	Documents	Percentage	Citation
1	China	358	58.02%	36566
2	USA	58	9.04%	8737
3	Singapore	45	7.29%	6354
4	South Korea	21	3.40%	2415
5	India	20	3.24%	1461
6	Spain	12	1.94%	572
7	Australia	11	1.78%	2023
8	Canada	10	1.62%	695
9	Germany	9	1.46%	2489
10	Netherlands	9	1.46%	1003

Analysis of research institutions

TLCS represents the number of citations a set of papers receives from other papers within the same dataset, while TGCS indicates the number of times these papers have been cited across the entire WOS database [36]. Compared to TGCS, TLCS better reflects the influence of a specific group of papers within a particular field. Therefore, TLCS was used to rank the top 10 most influential institutions (Table 2), with the Chinese Academy of Sciences, National University of Singapore, and Soochow University occupying the top 3 positions.

Table 2.

Top 10 central institutions studying upconversion nanoparticles for drug delivery and biomedical targeting in photodynamic therapy

Rank	Institution	Country	Count	Percent	TLCS1	TGCS2
1	Chinese Academy of Sciences	China	90	10.70%	1195	12,687
2	National University of Singapore	Singapore	26	3.09%	584	4938
3	Soochow University	China	13	1.55%	436	3678
4	Harbin Engineering University	China	33	3.92%	336	3406
5	University of Chinese Academy of Sciences	China	26	3.09%	334	3263
6	University of Amsterdam	Netherlands	6	0.71%	184	870
7	Jilin University	China	19	2.26%	182	1689
8	Fudan University	China	14	1.66%	167	2113
9	New Mexico Institute of Mining and Technology	USA	2	0.24%	105	595
10	University of Massachusetts	USA	8	0.95%	95	1043

¹TLCS, total local citation score

²TGCS, total global citation score

Additionally, we constructed a collaboration network among institutions (Fig. S1), focusing on those with 5 or more publications. In this network, nodes of the same color represent closer collaborative relationships, while node size corresponds to the number of publications. Clustering was performed with a minimum cluster size of 3, resulting in 4 main clusters, with a fifth group of gray nodes representing institutions with sparse collaborations that did not form a cluster. The orange cluster includes institutions such as the Chinese Academy of Sciences, the University of Chinese Academy of Sciences, and Jilin University. The purple cluster features the National University of Singapore, Nanyang Technological University, and Soochow University. The blue cluster encompasses the Harbin Institute of Technology and the University of Massachusetts, while the green cluster includes Harbin Engineering University and Shanghai Jiao Tong University. These clusters may suggest common research interests among the institutions, providing valuable insights for researchers seeking potential collaboration partners.

Analysis of contributing journals

Publications related to UCNPs for PDT drug delivery and biomedical targeting were found in 146 different journals. Table 3 lists the top 10 journals with the highest number of relevant articles, accounting for 36.92% of the total publications. Nanoscale is the journal with the most publications in this field, accounting for 6.33% of the total, followed by ACS Applied Materials & Interfaces and Journal of Materials Chemistry B. The h-index, developed by J.E. Hirsch and published in Proceedings of the National Academy of Sciences of the United States of America, is based on a list of publications ranked in descending order by the number of citations. The h-index corresponds to the number of papers (N) in the list that have been cited N or more times [37]. The h-index reflects both the quantity of papers and the frequency with which they are cited, and it is less influenced by a single highly cited paper, making it a more balanced metric for evaluating scientific impact. Among the top 10 journals, Advanced Materials, Chemical Society Reviews, and Biomaterials have the highest h-index values. Additionally, according to the 2024 Journal Citation Reports, all of the top 10 journals are classified as Q1, except for ACS Applied Materials & Interfaces and Journal of Materials Chemistry B, which are classified as Q2.

Table 3.

Top 10 core journals publishing upconversion nanoparticles for drug delivery and biomedical targeting in photodynamic therapy

Rank	Source	Count	Percent	Country	H-index	IF1 (2024)	JCR2 (2024)
1	Nanoscale	30	6.33%	UK	303	5.1	Q1
2	ACS Applied Materials & Interfaces	25	5.27%	USA	169	8.2	Q2
3	Journal of Materials Chemistry B	24	5.06%	UK	71	5.7	Q2
4	Biomaterials	20	4.22%	Netherlands	337	12.9	Q1
5	Small	16	3.38%	Germany	195	12.1	Q1
6	ACS Nano	14	2.95%	USA	310	16.0	Q1
7	Advanced Materials	14	2.95%	Germany	447	26.8	Q1
8	Chemical Society Reviews	11	2.32%	UK	432	39.0	Q1
9	Theranostics	11	2.32%	China	69	13.3	Q1
10	Chemistry Reviews	10	1.90%	Netherlands	254	23.5	Q1

¹IF, impact factor

²JCR, journaI citation reports

Analysis of most productive authors

A total of 2,258 authors have made contributions to research in this field. The top 10 researchers with the highest publication counts are presented in Table 4. The most prolific author is Yong Zhang from the National University of Singapore, who has contributed 33 articles to this field. Jun Lin from the Chinese Academy of Sciences, however, holds the highest citation count and h-index, making him the most influential author in this domain. We observed close collaborations among several authors, which were grouped into six clusters based on their connectivity (Fig. S2). The clustering network not only reveals the collaborative relationships among researchers studying UCNP-PDT drug delivery and biomedical targeting but also provides insight into the different research directions currently being pursued in this field.

Table 4.

Top 10 productive authors studying upconversion nanoparticles for drug delivery and biomedical targeting in photodynamic therapy

Rank	Author	Institution	Documents	Citations	H-index
1	Yong Zhang	National University of Singapore	33	4762	60
2	Piaoping Yang	Harbin Engineering University	32	3386	77
3	Jun Lin	Chinese Academy of Sciences	31	5002	143
4	Fei He	Harbin Engineering University	26	2614	61
5	Shili Gai	Harbin Engineering University	25	2534	52
6	Chunxia Li	National University of Singapore	25	3794	52
7	Ruichan Lv	Harbin Engineering University	19	1366	29
8	Bin Liu	Chinese Academy of Sciences	14	1786	92
9	Hanjie Wang	Tianjin University	14	525	51
10	Yunlu Dai	Nanyang Technological University	13	2106	60

Analysis of co-cited references

In this field, a total of 26,639 co-cited references were identified. Table 5 presents the top 10 co-cited references, with the most frequently cited work being In Vivo Photodynamic Therapy Using Upconversion Nanoparticles as Remote-Controlled Nanotransducers by Niagara Muhammad Idris, cited 172 times [38]. This is followed by Near-Infrared Light Induced In Vivo Photodynamic Therapy of Cancer Based on Upconversion Nanoparticles by Chao Wang [39] and Upconversion Nanoparticles: Design, Nanochemistry, and Applications in Theranostics by Guanying Chen [40]. Figure 4a illustrates the co-citation relationships among these references, forming 3 distinct clusters. Using CiteSpace, these co-cited references were further analyzed, yielding a mean silhouette score of 0.882 and a modularity Q score of 0.719, indicating strong clustering effects and high homogeneity. As shown in Fig. 4b, the references were divided into 13 clusters, such as #0 (size = 127) "excited lanthanide-doped upconversion nanoparticle," #1 (size = 97) "recent advance," and #3 (size = 65) "light-mediated theranostics," among others. Additionally, the timeline view in Fig. 4c depicts the evolutionary trajectory of the research themes, illustrating the historical and temporal span of co-cited references within each cluster. Citation bursts in references were first observed in 2007, with the latest bursts appearing in 2021 (Fig. 4d).

Table 5.

Top 10 co-cited references among the 474 publications

Rank	Author	Year	Source	Citations	DOI
1	Niagara Muhammad Idris	2012	Nature Medicine	172	10.1038/nm.2933
2	Chao Wang	2011	Biomaterials	125	10.1016/j.biomaterials.2011.05.007
3	Guanying Chen	2014	Chemical Reviews	113	10.1021/cr400425h
4	Dennis E J G J Dolmans	2003	Nature Reviews Cancer	105	10.1038/nrc1071
5	Peng Zhang	2007	Journal of the American Chemical Society	92	10.1021/ja0700707
6	Kai Liu	2012	ACS Nano	89	10.1021/nn300436b
7	Haisheng Qian	2009	SMALL	84	10.1002/smll.200900692
8	Sisi Cui	2013	ACS NANO	83	10.1021/nn304872n
9	Jing Zhou	2012	Chemical Society Reviews	81	10.1039/c1cs15187h
10	François Auzel	2004	Chemical Reviews	80	10.1021/cr020357g

Fig. 4.

Co-cited references between 2007 and 2025 in upconversion nanoparticles for drug delivery and biomedical targeting in photodynamic therapy. a Co-citation network; b clusters; c Timeline of the top 13 clusters; d Top 25 references with the strongest citation bursts

Analysis of keyword co-occurrence

Keywords encapsulate the essence of academic research. Table 6 presents the top 14 keywords ranked by their frequency of occurrence. In this study, we utilized VOSviewer to construct a keyword co-occurrence network, manually merging synonymous terms to refine the analysis. Among the 1,626 keywords present in the included literature, 35 keywords with a co-occurrence frequency of over 20 times were categorized into these 3 clusters (Fig. S3a). To further validate the research hotspots, we employed CiteSpace for additional clustering, which yielded 10 clusters (Fig. S3b). The clusters are numbered starting from 0, with smaller numbers indicating higher frequency and importance of the keywords within the cluster. The 3 largest clusters are #0 (size = 44) "combination therapy," #1 (size = 38) "contrast agents," and #2 (size = 37) “photodynamic therapy.” Fig. S3c provides a detailed view of the temporal trajectories and specific keywords within each cluster. Additionally, a burst analysis of keywords identified 18 terms with significant citation bursts. The earliest burst term was "singlet oxygen," which began in 2007. From 2013 onwards, terms like "in-vivo photodynamic therapy," "ray computed tomography," and "near-infrared light" emerged as prominent burst keywords. Notably, terms such as "upconversion" and "metal–organic framework have continued to show strong citation bursts, extending into 2025. These analyses are instrumental in uncovering research trends, highlighting the focal points of research during different periods (Fig. S3d).

Table 6.

Top 14 keywords on the research of upconversion nanoparticles for drug delivery in photodynamic therapy

Rank	Keyword	Occurrences	Rank	Keyword	Occurrences
1	Upconversion nanoparticles	334	8	Nanoparticles	906
2	Photodynamic therapy	288	9	Near-infrared light	878
3	Nanomaterials	172	10	Upconversion	794
4	Drug delivery	145	11	Singlet-oxygen	697
5	Cancer	108	12	Photothermal therapy	778
6	Luminescence	107	13	In-vivo	700
7	Photosensitizers	95	14	Delivery systems	650

Discussion

Global trends on UCNP-PDT drug delivery and biomedical targeting

UCNPs, owing to their anti-Stokes emission, high chemical stability, and favorable biocompatibility, have emerged as a key platform driving research in PDT-based drug delivery. Over the past 20 years, the field has undergone significant evolution in both focus and scientific questions. Our bibliometric analysis indicates three major stages of development: initial exploration (2007—2011), rapid growth (2012—2017), and optimization and integration (2018—2025). During the initial stage, studies primarily addressed material synthesis and basic validation, yet energy transfer efficiency and in vivo applications remained limited [41, 42]. In the rapid growth phase, strategies such as Nd³⁺ doping and dye sensitization [43] markedly improved UCNP brightness and safety. In the current optimization and integration stage, research has shifted toward composite systems, including MOFs, and combination therapy approaches (Fig. 5). Although overall publication volume shows a declining trend, the increasing proportion of in vivo studies signals a pivotal transition from material-focused development toward therapeutic validation and clinical translation (Fig. S4). This trajectory not only reflects an evolution in research priorities but also signifies the field's growing maturation.

Fig. 5.

Evolutionary overview of upconversion nanoparticles for drug delivery and biomedical targeting in photodynamic therapy (2007–2025). UCNPs, upconversion nanoparticles; PDT, photodynamic therapy; NIR, near-infrared (The scheme is designed by BioRender, https://www.biorender.com/)

Among the top 10 most productive institutions, 6 are based in China, which partly explains why China contributes the majority of publications in UCNP-PDT drug delivery and biomedical targeting. Notably, although Germany and Australia have a moderate number of publications, their high citation counts indicate that their work exerts significant influence in the field. This underscores the importance of establishing top-tier research institutions to enhance a country's academic standing.

Among the top 10 most productive authors, Yong Zhang has published the largest number of papers, while Jun Lin has the highest h-index. Collaboration networks show that Piaoping Yang, Fei He, Shili Gai, Bin Liu, and Yanlu Dai collaborate closely, as do Jun Lin and Chunxia Li. These collaborations appear to facilitate the production of high-impact research and reflect the existence of distinct research teams with different focuses.

Specifically, Yong Zhang focuses on in vivo photodynamic therapy for deep tumors and remotely controlled nanosystems, with expertise in controllable synthesis, surface modification, and functionalization of nanofluorescent materials [41, 44, 45], and has contributed some of the most influential articles in the field [38]. Piaoping Yang specializes in UCNP structural optimization and catalysis-driven therapies [46–49]. Jun Lin studies high-doped UCNPs for imaging-guided photodynamic therapy [50–52]. Ruichan Lv focuses on multifunctional UCNPs for combined PDT and photothermal therapy (PTT) [46, 48, 53]. Hanjie Wang is skilled in designing safe and effective delivery strategies for these nanosystems in vivo [54–56]. These teams have distinct strengths in materials design, functional optimization, and clinical translation, forming a diverse set of research hotspots [50, 57–59].

Regarding journals, although Nanoscale publishes the most articles, Chemical Society Reviews and Advanced Materials have the highest h-index and impact factors. This reflects a functional division where Nanoscale serves as a core repository for primary research, whereas broader-scope top journals stage the field's most influential reviews and advances.

Overall, these analyses provide a clear picture of the research landscape and major contributors in the UCNP-PDT field, offering insights for future collaborations and research directions.

Structural and compositional optimization of UCNP-PDT

Keyword networks (Fig. S3a) suggest that contemporary research largely falls into 2 categories: structural and compositional optimization of UCNPs, and the construction of integrated imaging-therapy platforms. The latter will be discussed in the Hotspots section; here, we first focus on structural and compositional aspects.

Upconversion in UCNPs involves multiple processes—including excited-state absorption, energy transfer upconversion (ETU), cooperative sensitization upconversion, cross relaxation, and photon avalanche—with efficiency governed by host lattice, sensitizer, and activator interactions [60, 61]. In our dataset, fluoride hosts account for 84.42% (n = 233) of studies due to their low phonon energy, which minimizes nonradiative losses [40]. Recent studies show that lattice engineering and rare-earth doping can markedly improve the performance of UCNP-PDT systems. Substituting the rigid NaYF₄:Yb,Er lattice with the softer Na₃ZrF₇:Yb,Er,Ce enhanced both flexibility and biodegradability, while boosting optical performance with a 2.5-fold increase in upconversion luminescence (UCL) intensity. In vivo, strong NIR-IIb signals accumulated in the liver and spleen but declined sharply within 48 h, indicating efficient clearance [62]. Lattice-matched hosts such as LiLuF₄:Yb,Er,Ce@LiYF₄@LiLuF₄:Nd have further achieved a quantum yield of 21.7% at 0.1 W/cm², nearly five times higher than NaLuF₄ controls. These systems also improved ROS generation, with a singlet oxygen yield of 0.469 versus 0.208 for conventional UCNP–Ce6, and reduced A549 cell viability to 21.9% under 808 nm irradiation. In vivo, real-time NIR-IIb imaging enabled precise tumor monitoring, and treatment achieved a 94% tumor suppression rate without obvious toxicity. Together, these findings highlight how rational lattice design can deliver both enhanced emission efficiency and biodegradability, making UCNP-PDT more effective and safer for biomedical applications [63].

Ion doping critically influences both emission wavelength and intensity. In our bibliometric analysis of 276 selected articles, 166 studies (60.14%) employed Yb³⁺–Er³⁺ pairs, which predominantly produce green and red emissions, whereas 73 studies (26.45%) used Tm³⁺ doping to extend emission into the near-UV and blue regions. Within these studies, optimizing Tm³⁺ content can enhance 475 nm emission by 11.5-fold [64]. Core–shell structures remain central to enhancing brightness and energy transfer efficiency. Applying inert or gradient-doped shells minimizes surface quenching and optimizes energy transfer pathways, enabling strong upconversion at reduced excitation power densities [40, 65, 66]. For instance, Yb³⁺/Er³⁺ core–shell UCNPs show a 7.4-fold enhancement, and Yb³⁺/Tm³⁺ core–shell UCNPs exhibit a 29.6-fold increase under identical conditions [67], substantially reducing reliance on high-power lasers. These findings highlight how rational ion selection and core–shell engineering, as reported in the literature, can significantly enhance UCNP performance while enabling lower-power, safer excitation.

Excitation wavelength selection further impacts the application. 808 nm excitation has steadily increased since 2012, reaching roughly 66.7% in 2023 (Fig. S5). This proportion reflects the fraction of publications using 808 nm relative to the total number of UCNP-PDT studies in that year, highlighting the growing adoption of Nd³⁺ sensitization to mitigate water absorption and local heating for deeper tissue penetration [68]. While 980 nm remains predominant due to its higher upconversion efficiency and stronger excitation of conventional sensitizers, its high power density requirement has historically hindered clinical translation. Dual-wavelength strategies using 808 and 980 nm have recently emerged [24, 69, 70], enabling independent activation of PSs and UCNP excitation to meet both therapeutic and imaging demands. Early UCNP systems often exceeded the skin's maximum permissible exposure (e.g., > 0.73 W/cm² at 980 nm and > 0.33 W/cm² at 808 nm) [71, 72], limiting clinical feasibility [73]. Recent advancements, however, such as Nd³⁺ sensitization, dye sensitization, and plasmonic enhancement, now enable effective emission at clinically relevant power densities of 0.2–0.5 W/cm² [63, 74–77], indicating a clear trajectory toward clinical applicability.

Despite improvements in UCNP design, energy transfer from UCNPs to PSs remains a major challenge. UCNP emission spectra often overlap well with the absorption spectra of various PSs, and the most efficient energy transfer occurs via Förster resonance energy transfer (FRET), which relies on nonradiative dipole–dipole coupling [29]. Nevertheless, PS loading efficiency and UCNP brightness are still limited, particularly under low NIR excitation, resulting in low quantum yields and extinction coefficients. This can be partially improved by optimizing the local environment of lanthanide dopants or by using dye-sensitization, plasmonic structures, or quantum dots to enhance emission [78]. Effective FRET requires PSs to be in close proximity to UCNPs, implying very thin shells, yet even at optimal distances, the overall quantum yield may remain low [79, 80]. Moreover, PS release is often unstable: physical adsorption can lead to aggregation and self-quenching, whereas excessive modification may shield UCNP emission and reduce energy transfer efficiency [81]. Strategies such as coating UCNPs with porous SiO₂ or MOF shells can provide high surface-area loading sites while preserving optical performance [82, 83], and stimuli-responsive systems (pH or NIR-triggered) enable controlled spatiotemporal release.

Comparisons with alternative excitation strategies highlight both strengths and limitations of UCNPs. Direct NIR PSs (800–900 nm) are easy to use but face inherent drawbacks such as broad emission peaks, short lifetimes, photobleaching, and limited tunability, in addition to potential photothermal side effects [84]. Two-photon excitation achieves deep penetration [85], but most PSs show very low cross-sections (< 50 GM), thereby restricting ROS production [86, 87]. Scintillators convert X-rays into visible light and thus increase penetration, but they expose healthy tissues to ionizing radiation. Sonodynamic approaches can also activate PSs in deep tissue, but sonoluminescence is weak, unpredictable, and lacks wavelength tunability, making control of treatment challenging [84]. By contrast, UCNPs employ efficient ETU mechanisms and can be engineered for strong emission enhancement—for instance, a thin NaYF₄ shell boosted visible UC emission by 7.4 × (Er³⁺) and 29.6 × (Tm³⁺) [88], while Ca²⁺ substitution achieved a 121-fold increase [89]. UCNPs further enable orthogonal excitation (dual-wavelength PDT and imaging) and high-resolution NIR-IIb imaging, and they can be functionalized with folate or magnetically guided for targeted tumor accumulation [90]. These advances underscore the superior controllability, spectral flexibility, and biosafety of UCNP-based PDT, even though optimizing efficiency and ROS generation in vivo remains an ongoing challenge.

Together, structural optimization, excitation strategy, and comparative advantages underscore the pressing scientific challenge: achieving efficient upconversion and stable ROS generation under clinically permissible low-power irradiation. These material-level improvements directly impact the therapeutic performance of UCNP-PDT systems. In particular, their translation into cancer treatment highlights both the opportunities and the challenges of applying optimized UCNPs in vivo.

UCNP-PDT applications across cancer types

Before exploring specific research hotspots, it is essential to briefly summarize the applications of UCNP-PDT across different tumor types. Among the 150 studies that explicitly targeted specific cancers, breast cancer accounted for the largest proportion (n = 60, 40.00%), followed by liver cancer (n = 24, 16.00%) and cervical cancer (n = 21, 14.00%) (Fig. S6). Notably, a substantial subset of breast cancer studies addressed triple-negative breast cancer (TNBC). As current therapies for advanced TNBC rely predominantly on chemotherapy, which provides limited survival benefits and often compromises quality of life, UCNPs present promising strategies and tools to advance PDT for TNBC management.

For instance, Jin et al. developed UCNP@TTD-cRGD, enhancing PS stability and tumor inhibition via αvβ3 targeting and deep NIR penetration [91]. Similarly, CREKA-modified UCNP nanoliposomes (CLIP-RB-PFOB@UCNP) co-delivering oxygen and PSs achieved significant in situ tumor destruction and lung metastasis suppression under NIR irradiation [92]. In addition, GdOF-based cRGD UCNP-doxorubicin probes combined pH-responsive drug release with synergistic PDT and PTT therapy, along with imaging guidance, demonstrating excellent theranostic potential in TNBC models [93]. Nonetheless, these studies remain largely preclinical, and clinical relevance requires further validation.

Challenges persist due to TNBC's hypoxic, acidic, and immunosuppressive microenvironment [94], which can limit PDT and immunotherapy efficacy. Future research should focus on overcoming immune suppression, remodeling the tumor microenvironment, and constructing multifunctional, responsive UCNP platforms. Additionally, complex vasculature in breast tissue and metastatic sites imposes further constraints on nanoparticle delivery, presenting ongoing formulation and translational challenges [94].

Hotspots and emerging frontiers in UCNP-PDT drug delivery and biomedical targeting

This study identifies four significant research frontiers based on the keyword timeline and clustering analysis. In keyword burst detection, terms that emerge later often indicate more recent hotspots, while clusters with smaller numerical labels in the map generally reflect higher importance. Moreover, clusters with lines extending further toward recent years highlight sustained and ongoing attention. Some frequently appearing terms, such as "upconversion nanoparticle," were excluded from the hotspot analysis because they were part of the initial search strategy and therefore do not carry additional interpretive value. According to our results, the most prominent emerging keywords in recent years include "metal–organic frameworks" and "tumor microenvironment." In addition, the clustering analysis highlights "combination therapy" and "antibacterial PDT" as important directions. Based on these findings, the following sections will analyze each research frontier in detail (Fig. 6).

Fig. 6.

Research frontiers of UCNPs for drug delivery and biomedical targeting in photodynamic therapy, with the inner circle representing recent research frontiers and the outer circle representing research prospects. UCNP, upconversion nanoparticle; PDT, photodynamic therapy; MOFs, metal–organic frameworks; TME, tumor microenvironment; aPDT, antimicrobial photodynamic therapy. The scheme is designed with biorender (https://www.biorender.com)

MOFs

The first major hotspot is MOFs. In our keyword burst analysis, MOFs have shown a strong and sustained burst signal since 2021, indicating that this topic has emerged as a key frontier in UCNP-PDT-based drug delivery and biomedical targeting. Notably, our bibliometric statistics reveal that only 7.38% (35/474) of the analyzed studies have investigated MOF–UCNP systems in the context of PDT, underscoring both the novelty of this research direction and its potential for further exploration.

MOFs combine the advantages of both inorganic and organic materials, featuring high surface area, tunable pore size, and favorable biocompatibility [95–97]. These characteristics make them particularly suitable for PS loading, drug delivery, and TME responsiveness. When integrated with UCNPs, the organic ligands of MOFs can serve as "energy relay stations," promoting Förster or lanthanide resonance energy transfer and simultaneously preventing aggregation-induced quenching of PSs in solution [83, 98]. Furthermore, the ordered arrangement within the rigid MOF structure provides an ideal intermolecular spacing and orientation for TTA upconversion, thereby improving exciton diffusion efficiency [83, 98].

At the application level, several UCNPs@MOFs platforms have been developed for both cancer therapy and infectious disease treatment. For instance, NH₂-MIL-53(Fe) can serve not only as a photosensitizer for PDT but also as a catalyst for the conversion of H₂O₂ to •OH through the Fe³⁺/Fe²⁺ cycle, thereby enhancing chemodynamic therapy (CDT). In addition, its porous structure enables efficient loading of chemotherapeutic drugs, resulting in a triple-therapy strategy with synergistic effects [99]. Another study constructed an antibacterial platform by embedding a UCNPs@ZrMOF-Pt composite in a hydrogel matrix. In this system, Zr-MOF not only stabilized the Pt complex but also displayed catalase-like activity, continuously generating O₂ in the H₂O₂-rich environment of infected wounds. This relieved hypoxia, enhanced PDT efficacy, and significantly accelerated wound healing [100]. These examples demonstrate how the porosity and catalytic activity of MOFs complement the deep-tissue excitation capacity of UCNPs, offering a practical pathway toward multifunctional theranostic platforms.

Nevertheless, UCNPs@MOFs platforms face several challenges. First, the loading efficiency within MOF pores remains limited, and partial loss of PSs during treatment can compromise PDT outcomes. Second, the spectral match between UCNP emission and MOF-based PSs is not yet fully optimized, leading to variability in energy transfer efficiency across different designs [101]. Third, the structural complexity of these composites reduces synthetic reproducibility and scalability, hindering clinical translation [102].

Despite these challenges, the integration of MOFs with UCNPs still holds considerable promise. Advances in reticular chemistry may drive a shift from simple drug carrier functions toward multifunctional reactive platforms. In parallel, expanding research into other porous frameworks, such as covalent organic frameworks and porous organic cages, could retain high porosity while improving biodegradability and energy transfer efficiency.

TME

TME, which consists of cancer cells, immune cells, fibroblasts, vasculature, and secreted factors, poses a major barrier to PDT due to its characteristic hypoxia and elevated glutathione (GSH) levels [103]. Hypoxia markedly decreases ROS production by PSs, while excessive GSH neutralizes ROS, thereby diminishing therapeutic efficacy [104]. Moreover, ROS generation itself consumes oxygen, further exacerbating hypoxia and limiting PDT effectiveness [105]. To mitigate these limitations, certain nanoplatforms have been designed to simultaneously supply oxygen and modulate redox balance within tumors. In the case of UA@CC nanoparticles, once they accumulate in tumor tissue, the CaO-Cu layer reacts under mildly acidic conditions to release oxygen. At the same time, Cu²⁺ ions are reduced to Cu⁺, which depletes intracellular GSH, while the endogenous H₂O₂ serves as an additional substrate to promote ROS formation. The incorporation of 5-ALA further boosts protoporphyrin IX levels, resulting in enhanced synergistic photodynamic and chemodynamic therapeutic effects in melanoma models [106]. Nevertheless, such approaches are inherently limited by the complexity of the TME.

To overcome these oxidative stress–related obstacles, recent UCNP-PDT strategies have employed multi-mechanistic approaches. Mitochondria-targeted platforms, such as NZ@TG, exploit the relatively oxygen-rich mitochondrial microenvironment, increasing total intracellular ROS by ~ 73% and mitochondrial ROS intensity by ~ 3.9-fold under UCNP-PDT, amplifying cytotoxicity even in hypoxic tumors [107]. Similarly, ZnO-Ce6 systems leverage catalase-like ZnO activity for in situ oxygen generation while inducing immunogenic cell death and ferroptosis via Ce6 excitation, reducing PDT dependence on light penetration [74]. These findings suggest that, rather than merely maximizing ROS production, more effective strategies may focus on weakening tumor defenses against ROS through multiple complementary mechanisms.

Building on these multi-mechanistic strategies, multi-target approaches have been developed to address hypoxia and metabolic resistance. For example, the UCNPs@MON@CQ/CHCA@HA platform integrates chloroquine to inhibit autophagy, CHCA to block lactate uptake, and disulfide bonds to consume GSH, thereby synergistically amplifying ROS effects [108]. Another innovative direction is light-controlled gene editing. The UCPP platform employs UCNPs to achieve spatiotemporal knockdown of HIF-1α under near-infrared irradiation, thereby mitigating hypoxia and reinforcing PDT, which substantially improved outcomes in deep-seated tumors [109].

In parallel with strategies aimed at overcoming TME-related barriers, an emerging approach is to exploit the distinctive physicochemical features of the TME for environment-responsive therapies. UCNP-based platforms have also been engineered to respond to specific TME cues, such as acidic pH, redox species, adenosine triphosphate (ATP), and metal ions (e.g., Zn²⁺, Fe²⁺/³⁺, Mn²⁺), achieving precise ROS generation or drug release at disease sites [103, 110–113]. For instance, Chen et al. developed a Zn²⁺-activated UCNP probe that selectively released ROS upon detecting intracellular Zn²⁺ imbalance, resulting in highly targeted cytotoxicity with an apoptosis rate exceeding 70% in breast cancer cells [114]. Li et al. designed triple-layer UCNP hydrogels that recognize elevated ATP concentrations via aptamer binding, thereby unlocking doxorubicin release in combination with PDT [115]. Building on these single-response designs, multifunctional platforms have been developed to integrate multiple TME cues for synergistic therapy. A representative example is the UC@mSiO2-RB@ZIF-90-DOX-PEGFA (URODF) system, which combines 808 nm UCNP excitation, O₂ release from a ZIF-90 shell, and pH-triggered doxorubicin delivery, achieving synergistic PDT and chemotherapy selectively in acidic tumor environments [116]. Together, these studies demonstrate the versatility of TME-responsive UCNP platforms and highlight their potential for precise, site-specific cancer therapy.

Among these strategies, oxygen-generating and GSH-depleting nanoplatforms have been widely validated in animal models, suggesting higher preclinical maturity [106, 117, 118]. In contrast, gene editing [29], metabolic modulation [119], and multi-responsive platforms remain largely at the proof-of-concept stage, though they represent promising directions for future translation. Taken together, the complexity of the TME represents both the greatest obstacle to PDT and a unique opportunity for UCNP functionalization. Future research should shift from simply enhancing local ROS generation toward integrated approaches involving immune modulation, metabolic intervention, and gene editing. In particular, the development of intelligent, multi-responsive platforms will be essential to address the heterogeneity of different tumor types.

Combination therapy

Our bibliometric analysis indicates that "combination therapy" represents the most prominent keyword cluster in the UCNP field, ranking first in importance. This suggests that UCNP-PDT combination strategies have emerged as a key focus for future research. Rather than serving solely as single-function PDT platforms, UCNPs are increasingly developed as multifunctional therapeutic cores. Based on the mode and complexity of combination, these strategies can be categorized as single PDT, single-combination, and multi-combination approaches. This progression not only reflects the growing complexity in material design but also a shifting focus toward balancing therapeutic efficacy with clinical translatability. Although single PDT studies still dominate the literature, the proportion of combination therapy studies has steadily increased over time, as illustrated in Fig. S7.

Single PDT, as the earliest application, faces inherent limitations in deep-seated tumors, including hypoxia, insufficient ROS generation, and limited light penetration, which hinder its effectiveness [120]. To address these issues, researchers quickly moved toward single-combination therapies, most commonly PDT coupled with chemotherapy. Such combinations not only compensate for the limitations of PDT but also reduce the toxicity associated with high-dose monotherapy [121]. For instance, the UCNP/Ce6@mMnO₂/DSP-TFA platform designed by Han et al. enables PDT under NIR-II imaging guidance while MnO₂ generates oxygen in the TME and releases cisplatin, alleviating hypoxia and enhancing chemotherapeutic efficacy in a non-small cell lung cancer model [122]. Similarly, UCNPs integrated with photothermal agents can simultaneously induce thermal therapy and ROS generation, creating complementary effects that improve treatment in deep tissues. Another study by Yoon et al. employed NaErF₄@NaLuF₄ core–shell UCNPs with a silica surface layer to load Ce6, the oxygen carrier PFC, and paclitaxel, while mannose modification and macrophage membrane coating enabled specific targeting of M₂ tumor-associated macrophages (TAMs). Upon 808 nm excitation, this system efficiently released oxygen and generated singlet oxygen (¹O₂) while promoting TAM polarization toward the antitumor M₁ phenotype. In both 2D and 3D coculture systems, significant cytotoxicity was observed. In a 4T1 breast cancer mouse model, the treated group had an average tumor volume of 332.4 mm³, compared with 709.5–1185.5 mm³ in controls, demonstrating a clear therapeutic advantage [123]. These results highlight that UCNP-PDT combined with immune modulation can overcome single-treatment limitations, achieving efficient and multi-level antitumor effects.

Building on these findings, some studies have explored multi-combination and theranostic platforms to integrate imaging, drug delivery, and multi-pathway therapy within a single system. Theranostics has become a major trend in tumor nanomedicine, aiming to combine diagnostic and therapeutic functions in one nanocarrier for precise, image-guided intervention. For multimodal imaging, UCNPs provide deep-tissue upconversion luminescence under near-infrared excitation [124], which can be combined with magnetic resonance imaging (MRI) [125], computed tomography (CT) [125], and photoacoustic imaging [126] to achieve a cross-scale, multi-dimensional diagnostic system. This approach enables real-time monitoring of drug release, tumor boundary delineation, and treatment response. For example, Shao et al. developed a UCNP@MOF platform in which efficient energy transfer between the UCNP core and porphyrin-based MOF shell triggers ROS generation under near-infrared irradiation, achieving deep PDT. The MOF shell also delivers hypoxia-activated prodrug tirapazamine, selectively releasing it in hypoxic tumors to enhance chemotherapy. Furthermore, the platform supports dual UCL/CT imaging, improving lesion localization and providing real-time therapeutic monitoring. When combined with PD-L1 inhibitors, the system not only suppressed primary tumors but also induced cytotoxic T-cell infiltration, achieving complete inhibition of distant, untreated tumors [127]. This study demonstrates the potential of UCNPs as theranostic platforms, where multimodal imaging enables precise tumor targeting and drug monitoring, and structural design allows integration of PDT, chemotherapy, and immunotherapy to enhance overall efficacy.

Overall, the shift from single to multi-combination therapies shows that UCNP platforms can expand the functionality of PDT. However, combination therapy is not automatically better than monotherapy, nor is it merely a simple stack of treatments. Effective combinations depend on mechanistic complementarity and optimized energy pathways [121]. For example, a recent NIR light-controlled PROTAC delivery system combines UCNP-PDT with precise BRD4 degradation via light-triggered PROTAC release, enhancing antitumor efficacy. This design does not just add another module—it uses the optical properties of UCNPs to solve delivery and timing challenges [128].

Currently, most UCNP-PDT combination therapies remain at the animal study stage. Some combinations, like PDT plus chemotherapy or PDT plus PTT, have been studied more extensively. In contrast, combinations with immunotherapy, gas therapy, or gene editing are still limited. The choice of combination should consider the underlying mechanisms, as effectiveness can vary with tumor type and stage. This underscores the importance of clinical feasibility and translatability in developing multimodal UCNP platforms.

aPDT

The global spread of multidrug-resistant (MDR) bacteria poses a serious public health challenge, while the effectiveness of conventional antibiotics is increasingly limited. In this context, PDT and its antibacterial branch, aPDT, have attracted significant attention as alternative strategies. aPDT does not rely on traditional antibiotics and can directly kill pathogens through ROS generation by activated PSs [26, 129]. Several studies have demonstrated that aPDT effectively inhibits a variety of MDR pathogens, including Staphylococcus aureus, Acinetobacter baumannii, Klebsiella pneumoniae, and Escherichia coli [130–133].

Through surface functionalization, UCNPs can be covalently linked with targeting peptides, antibiotics, or metal ions. Such modifications not only improve light penetration depth but also enable selective recognition and precise eradication of specific pathogens, thereby enhancing the accuracy of aPDT [134]. However, our bibliometric analysis indicates that only 5.70% (27/474) of the included studies investigated UCNPs for aPDT, reflecting the relatively limited attention this area has received so far. Consequently, UCNP-aPDT remains in its early stages, with most studies focusing on in vitro or small animal models of oral biofilms [27, 135], Mycobacterium tuberculosis [136], or single MDR strains such as MRSA [26, 137]. Its applicability in complex infection environments still requires further validation.

Recent advances in nanocomposite design have promoted multifunctional UCNP platforms. Integration with MOFs, functionalized liposomes, or multimodal imaging components allows UCNPs to combine PDT with imaging, diagnostics, and combination therapies. For instance, the Fe@UCNP-HMME platform combines MRI and upconversion luminescence imaging with ROS generation, achieving near-complete eradication of resistant bacteria [138]. Wang and co-workers reported a cationic aggregation-induced emission PS, MTTTPy, which generates oxygen-independent ROS via a type I pathway. At a low concentration of 0.5 μM, MTTTPy achieved ~ 95% killing efficiency against Gram-positive bacteria under white light irradiation, even under hypoxic conditions. In vivo experiments further confirmed significant antibacterial efficacy in S. aureus-infected mice. These results highlight the potential of incorporating type I PSs into UCNP-aPDT systems to overcome hypoxia-related limitations and enhance antibacterial performance [28]. The study by Xu et al. reports yolk-structured multifunctional UCNPs designed for synergistic PDT and sonodynamic therapy (SDT) against antibiotic-resistant bacteria. The yolk–shell structure improved PS loading efficiency and biocompatibility. Importantly, the combined PDT-SDT therapy achieved 100% inhibition of antibiotic-resistant bacteria, which was significantly higher than PDT alone (74.2%) or SDT alone (70%) [139]. This result demonstrates that the synergistic effect of dual modalities substantially enhances antibacterial efficacy compared with either single therapy, likely due to more efficient ROS generation and better utilization of the loaded PSs.

Despite these advances, translation of UCNP-aPDT from the laboratory to the clinic faces multiple challenges. Most studies are conducted in single-strain or acute infection small-animal models, lacking systematic validation in mixed infections or complex microenvironments such as hypoxia, biofilms, and necrotic tissue. In addition, the long-term biocompatibility, immunogenicity, and in vivo metabolism of UCNP surface modifications—including PEG, SiO₂, or liposomes—remain unclear, limiting clinical safety assessment [81]. Bacterial structural differences also influence efficacy: Gram-negative bacteria require strategies to overcome the outer membrane barrier, whereas Gram-positive bacteria are more susceptible to PS penetration, making ROS generation in the cytoplasmic membrane the key mechanism [140, 141]. Recognizing these structural differences offers a rational framework for designing more effective and targeted aPDT interventions.

Future development of UCNP-aPDT should focus on three key directions: systematic evaluation of antibacterial efficacy and biosafety in large animal models; development of targeted and responsive nanosystems for deep or complex infections; and advancement of multimodal combination therapy and theranostic platforms to improve efficacy and reduce clinical translation risk. Through continued interdisciplinary and multi-scale research, UCNP-aPDT holds promise as a viable new solution for combating MDR infections.

Conclusion and future perspectives

Using bibliometric analysis, this study systematically reviewed the research progress and trends of UCNPs in PDT-based drug delivery and biomedical targeting. Research has evolved from simple PS loading to multifunctional nanoplatforms, with applications expanding from material design to cancer therapy, infection control, and immune modulation. Despite these advances, most outcomes remain at the proof-of-concept stage, highlighting the gap between laboratory research and clinical translation.

Although UCNPs offer unique advantages in excitation modalities and spectral tuning, several key limitations remain. First, particle size poses a trade-off: smaller UCNPs favor in vivo metabolism but often exhibit lower luminescence efficiency, while larger particles provide higher brightness but risk accumulation in the liver and spleen, potentially causing long-term toxicity [81]. Preclinical toxicological outcomes vary with particle composition, surface chemistry, dosage, and administration route, so without rigorous biodistribution and toxicity data, claims about safety can be misleading [142]. Second, under clinically permissible low-power irradiation, upconversion efficiency remains limited. Energy transfer between UCNPs and PSs and stable ROS production are constrained in complex physiological environments, particularly under hypoxia or dense biofilm conditions. Future research should develop more controllable and responsive surface modification strategies. Third, excitation at 980 nm can induce tissue heating, causing local temperature elevation and potential damage. While 808 nm systems partially mitigate this effect, overall emission efficiency remains suboptimal. Fourth, clinical translation of nanomedicines faces dual constraints from Good Manufacturing Practice (GMP) compliance and regulatory approval. Limited scalability and batch-to-batch variability present significant barriers from laboratory synthesis to GMP-compliant production [13, 143]. Fifth, systematic long-term biosafety and toxicological evaluations are insufficient. Most studies focus on short-term cytotoxicity, with limited investigation into immunotoxicity, genotoxicity, and neuro- or reproductive toxicity [121].

In summary, advancing UCNP-PDT requires more than enhancing anticancer or antimicrobial effects. A systematic integration of material optimization, mechanistic understanding, and clinical translatability is essential. Balancing efficiency, safety, and scalable production will be crucial to move UCNP-PDT beyond laboratory proof-of-concept and toward practical solutions for major health challenges such as drug-resistant infections and malignant tumors [144].

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

PDT

Photodynamic therapy

PSs

Photosensitizers

ROS

Reactive oxygen species

UCNPs

Upconversion nanoparticles

NIR

Near-infrared

MOFs

Metal–organic frameworks

WOS

Web of science

TME

Tumor microenvironment

aPDT

Antimicrobial photodynamic therapy

TLCS

Total local citation score

TGCS

Total global citation score

PTT

Photothermal therapy

ETU

Energy transfer upconversion

UCL

Upconversion luminescence

TNBC

Triple-negative breast cancer

CDT

Chemodynamic therapy

GSH

Glutathione

ATP

Adenosine triphosphate

TAMs

Tumor-associated macrophages

MRI

Magnetic resonance imaging

CT

Computed tomography

MDR

Multidrug-resistant

SDT

Sonodynamic therapy

GMP

Good manufacturing practice

Author contributions

Wan-Tong Zheng: Conceptualization, Methodology, Formal Analysis, Writing—Original Draft. Hao-Lin Chen: Conceptualization, Visualization, Writing—Original Draft. Xin-Ting Hou: Methodology, Writing—Original Draft. Alina Abakirova: Visualization. Meng-Yi Han: Data Curation. Yong-Qi Feng: Validation. Wan-Ting Huang: Software. Xiao-Chun Mo and Hao-Chun Zhu: Validation. Si-Kai Huang, Huan-Dong Lv, Ting Yang and Sha Huang: Writing—Review & Editing. Zhong-Miao Shi: Conceptualization, Supervision, Project administration, Funding acquisition.

Funding

No funding was received.

Data availability

The original contributions presented in this study are included in the article/supplementary material. Further inquiries can be directed to the corresponding author(s).

Declarations

Ethics approval and consent to participate

Review and/or approval by an ethics committee as well as informed consent was not required for this study because this article did not involve any direct experimentation/studies on living beings.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.