Evolutionary landscape and collaborative networks of global theranostic research in nuclear medicine: a visual bibliometric analysis

Abstract

Background

Theranostics, integrating diagnostic imaging with targeted radionuclide therapy, has become central to nuclear medicine, but its global research landscape and collaboration patterns remain incompletely mapped.

Objective

To characterize publication dynamics, cooperative networks, and emerging topics in theranostic research from 2006 to 2024.

Methods

Articles and reviews indexed in the Web of Science Core Collection were retrieved for bibliometric analysis. Excel, CiteSpace, VOSviewer, Scimago Graphica, ArcGIS, and the R package bibliometrix were used for quantitative analysis and visualization.

Results

A total of 1801 articles and reviews were included, with a compound annual growth rate of 18.7%. The USA, Germany, and China were the leading producer countries, and the USA–Germany collaboration showed the strongest bilateral link strength. The Helmholtz Association, Memorial Sloan Kettering Cancer Center, and the University of California System were the most productive institutions, while Herrmann K., Wester H.J., and Werner R.A. led individual output. Keyword co-occurrence analysis identified five major themes: prostate‐specific membrane antigen (PSMA)–targeted imaging and therapy, somatostatin receptor–based peptide therapy, neuroendocrine tumors, alpha-particle therapy, and fibroblast activation protein-targeted probes. Burst detection showed that fibroblast activation protein, actinium-225, and ¹⁷⁷Lu-PSMA-617 have emerged as recent research hotspots.

Conclusion

Theranostic research in nuclear medicine has expanded rapidly over the past two decades, with a trans-Atlantic and Asia-Pacific collaborative pattern. Current frontiers are shifting from beta-emitters toward alpha-emitters and tumor-microenvironment-targeted probes. Greater intercontinental cooperation and continued focus on novel radionuclides and under-explored targets may support future innovation and clinical translation.

Introduction

Theranostics integrates diagnostic imaging and radionuclide therapy functionalities of a single molecular-targeted probe, providing a precision-medicine paradigm that enables a ‘molecular-imaging screening → targeted radiopharmaceutical therapy → posttherapy response assessment’ loop within the same patient [1,2]. Over the past decade, a series of exemplar radiopharmaceuticals – ⁶⁸Ga/¹⁷⁷Lu-DOTATATE, ⁶⁸Ga/¹⁷⁷Lu-prostate-specific membrane antigen (PSMA)-617, and ¹⁸F/²²⁵Ac-FAPI, among others – have either been approved or advanced into phase III clinical trials, steering nuclear medicine from a traditionally ‘diagnosis-driven’ model toward a fully integrated ‘diagnosis-and-therapy’ era and attracting intense interdisciplinary investment and industrial interest [3–5].

Benefiting from the continuous expansion of molecular targets and medically relevant radionuclides, the annual growth rate of theranostics-related publications worldwide has now surpassed 15%, and research priorities have shifted from β-emitters toward α-particle therapies and tumor-microenvironment-targeted probes [6–8]. For example, ²²⁵Ac-PSMA radioligand therapy has achieved deeper responses and superior survival benefits compared with ¹⁷⁷Lu-PSMA in castration-resistant prostate cancer. Concurrently, new targets typified by fibroblast activation protein (FAP) and its inhibitor derivatives (FAPI) are rapidly accruing evidence, emerging as the most promising ‘second curve’ in the radiopharmaceutical pipeline. Yet, the resulting radionuclide supply bottleneck (e.g. the scarcity of ²²⁵Ac) and the complexity of multinational, multicenter clinical trials have set a higher bar for scientific collaboration [9–12].To date, most bibliometric investigations have focused on a single target (PSMA, somatostatin receptor, FAPI, etc.) or a single disease domain, lacking a macrolevel nuclear-medicine perspective that systematically maps the evolutionary landscape of theranostic research, cross-country/cross-institutional collaboration networks, and topic migration. This gap constrains our understanding of global innovation resource allocation and technology-diffusion pathways and hampers the ability of policy-makers and industry stakeholders to make fully informed, panoramic decisions [13–16]. In addition to the widely studied pairs such as ⁶⁸Ga/¹⁷⁷Lu-DOTATATE and ⁶⁸Ga/¹⁷⁷Lu-PSMA, other clinically relevant or emerging radionuclides – including ⁹⁰Y, ⁶⁴Cu/⁶⁷Cu, ⁸⁹Zr, ²¹²Pb, and ¹⁶¹Tb – as well as promising targets such as human epidermal growth factor receptor 2 (HER2), integrins (αvβ6, αvβ3), and CXCR4 are also expanding the theranostic landscape in nuclear medicine.

Building on these considerations, we mined the Web of Science Core Collection (2006–2025) and applied bibliometrix, VOSviewer, and CiteSpace to construct a multilayer visual-analysis framework that systematically dissects: (a) publication and citation trajectories; (b) country-, institution-, and author-level collaboration networks; (c) emergent themes/keywords and research frontiers; and (d) milestone papers and prospective growth nodes in nuclear-medicine theranostic research. The goal is to provide investigators, translational teams, and industry stakeholders with a data-driven, panoramic perspective that will catalyze ongoing innovation and cross-domain synergy in the theranostics arena amid the precision-medicine era.

Methods

This work was carried out as a bibliometric data-analysis study. All datasets were retrieved from publicly accessible sources on the internet, and no animal or human participants were involved; therefore, institutional review board or ethics committee approval was not required.

Search strategy

We queried the Web of Science Core Collection for records from 1 January 1900 through 31 May 2025 using the following search strategy: TOPIC (‘Theranostic*’ OR ‘Theragnostic*’ OR ‘Radiotheranostic*’ OR ‘Theranostic agent*’) AND TOPIC (‘Nuclear Medicine’ OR ‘Radionuclide Therapy’ OR ‘Therapeutic Radiopharmaceutical*’ OR ‘Radioligand* Therapy’ OR ‘Radioligand*’ OR ‘Peptide Receptor Radionuclide Therapy’ OR ‘PRRT’ OR ‘PSMA’ OR ‘DOTA-TATE’ OR ‘Lutetium*’ OR ‘177Lu’ OR ‘68Ga’ OR ‘Radioimmunotherapy’ OR ‘Radiotheragnostic*’).The search was limited to publications classified as original research articles or reviews and restricted to English-language records.

Analytical procedures

Using Microsoft Excel, we plotted annual publication trends; the R package bibliometrix furnished an overall bibliometric profile of the included literature – identifying the most-cited authors, journals, documents, inter-country collaboration patterns, and high-frequency keywords; CiteSpace performed reference cocitation clustering, keyword burst detection, timeline and heat-map visualizations, and a dual-map overlay of citing versus cited journals; VOSviewer generated keyword co-occurrence and overlay maps and preliminarily visualized collaboration among publishing countries; and finally, Scimago Graphica and ArcGIS provided geospatial analyses that further clarified transnational publication volumes and bilateral link strengths, together forming a multilayered analytic framework for examining global theranostic research.

Results

Overall characteristics of the literature included in this study

Figure 1 illustrates the study’s search workflow. An initial 3441 records were retrieved from the Web of Science Core Collection; after excluding conference abstracts, early access items, non-English papers, and other low-comparability documents, 2452 high-quality journal articles were included – 1748 original research papers and 704 reviews. Figure 2 presents the temporal evolution of global theranostic research in nuclear medicine from 2006 to 2024, showing annual publication counts (blue line, left axis) and cumulative totals (orange bars, right axis). The trajectory displays a marked increase in publication activity, with annual output rising rapidly after 2016 and peaking at approximately 450 papers in 2024, in clear contrast to the sporadic publications observed in the early years. The fitted curve also indicates a pronounced upward growth trend in research output across the study period.

Fig. 1.

Flowchart of literature identification and inclusion.

Fig. 2.

Annual publication output from 2006 to 2024.

National publication impact and collaborative contribution

Table 1 lists the 10 most-cited countries. When theranostic nuclear-medicine publications from 2006 to 2024 are ranked by total citations and citations per paper (CPP), Germany tops the chart with 15 003 citations and a CPP of 42.7 – well ahead of the USA (11 991 citations; CPP: 23.8). Although China ranks third in total citations (4189), its CPP is only 16.8, indicating that its publication volume is expanding rapidly while its scholarly impact is still catching up. Australia, Italy, and the Netherlands – all part of the Europe–Oceania research axis – enter the top 10 by virtue of high CPP values (28–36), highlighting the European academic community’s strong citation advantage in theranostics. By contrast, Canada, India, and France show medium-level influence and room for upward momentum.

Table 1.

Top 10 most-cited countries

Country	TC	Average article citations
Germany	15 003	42.70
USA	11 991	23.80
China	4189	16.80
Australia	3968	35.70
Italy	2243	14.60
The Netherlands	1912	28.10
Switzerland	1895	26.30
Canada	1699	19.50
India	1657	14.20
France	1634	18.40

Table 2 lists the top 10 countries ranked by the number of papers for which their researchers served as the corresponding author, highlighting publication volume and collaboration patterns. The USA leads with 503 papers (20.5% of the global total), of which 31.4% are multicountry publications (MCPs), indicating a still-strong capacity for in-house research and development. Germany produced 351 papers (14.3%), but more than half (51.3%) involve international collaboration, underscoring its ‘hub’ status within the European research network. By contrast, only 31.3% of China’s 249 papers were internationally co-authored, suggesting room to expand global partnerships. Although Switzerland and the Netherlands each account for less than 3% of total output, their MCP ratios are as high as 58.3 and 51.5%, respectively, showing how smaller nations can achieve substantial academic impact in theranostics by deeply engaging in international projects. Collectively, these data reveal a ‘US–EU bipolar core with an emerging Asia-Pacific cohort’ and emphasize that deeper cross-border collaboration is essential for boosting research quality and influence.

Table 2.

Corresponding-author country distribution for the top 10 nations by publication volume

Country	Articles	Articles %	SCP	MCP	MCP %
USA	503	20.5	345	158	31.4
Germany	351	14.3	171	180	51.3
China	249	10.2	171	78	31.3
Italy	154	6.3	94	60	39
India	117	4.8	100	17	14.5
Australia	111	4.5	73	38	34.2
France	89	3.6	60	29	32.6
Canada	87	3.5	57	30	34.5
Switzerland	72	2.9	30	42	58.3
The Netherlands	68	2.8	33	35	51.5

MCP, multicountry publications; SCP, single-country papers.

Figure 3a (bar plot of corresponding-author country distribution) ranks global theranostic publications by the country of the corresponding author, with blue bars representing single-country papers and red bars indicating multicountry collaborations. The USA, Germany, and China occupy the top three positions. Notably, MCP proportions are highest in the USA and Germany, suggesting that leading nations not only generate substantial output but also favor transnational team science. In contrast, emerging countries, such as India and Iran, display larger blue segments, indicating a predominance of domestically conducted studies. These patterns vividly reveal a ‘core-periphery’ structure and differing depths of international collaboration, providing a quantitative foundation for subsequent network-analysis work.

Fig. 3.

National publication output and collaborative contributions: (a) bar chart of corresponding-author country distribution; (b) global publication-density heat map; (c) inter-country collaboration flow map; and (d) VOSviewer circular co-authorship network.

Figure 3b (global publication-density heat map) shows a world map shaded from green (low) to red (high) according to each country’s output of theranostics-related papers published between 1990 and 2024. North America, Western Europe, and East Asia form red-to-yellow high-density belts, indicating that research hotspots align with traditional medical powerhouses and major economies. In contrast, most of Africa, the Middle East, and South America remain green, reflecting the spatial imbalance in global research resources and nuclear-medicine uptake – and underscoring opportunities for international technology transfer and capacity-building initiatives.

Figure 3c (inter-country collaboration flow map, co-authorship based) overlays a geographic basemap with network features: node size denotes publication volume, node color marks five collaboration communities identified by a clustering algorithm, and line thickness represents co-authorship strength. A trans-Atlantic ‘blue cluster’ centered on the USA, Italy, and the UK forms the network’s core. Germany and the Nordic countries constitute a ‘green cluster’, whereas China, Japan, and India make up a tightly knit ‘purple cluster’ within the Asia-Pacific region. The map visualizes the main intercontinental ‘arteries’ of collaboration alongside intra-regional small-world structures, offering a spatial perspective on knowledge flow and technology diffusion.

Figure 3d (VOSviewer circular co-authorship network): in this high-density graph, node size corresponds to publication volume, node color denotes modular clusters, and the thickness/gradient of edges represents collaboration strength. A centrally positioned US node radiates connections worldwide, with secondary hubs – Germany, Italy, China, and others – forming concentric layers around it. The network exhibits a clear small-world structure and a high clustering coefficient, indicating that international collaboration in theranostics is both concentrated and diversified. Compared with the geographic perspective in the previous sub-panel, this visualization highlights the topological significance of structural holes and hub nodes, offering a useful basis for pinpointing high-impact partners and opportunities for resource integration.

Most influential authors and institutions

Table 3 shows that Herrmann et al. [17] and Wester et al. [18] share the highest h-index (30) among all authors. Herrmann K. entered the theranostics field in 2015 and has since published 79 papers, accumulating 2891 citations, with a g-index of 52 and an m-index of 2.727, exemplifying a ‘high-output, high-growth’ trajectory [19,20]. Wester H.J. entered the field in 2014; his 49 papers have attracted a field-leading 3250 citations, giving a g-index of 49 and an m-index of 2.500 – characteristic of a ‘high-impact, steady-production’ leadership model [21,22]. Next, Baum R.P. (h = 24, g = 45, total citations = 2766) and Hofman M.S. (h = 23, g = 44, total citations = 2846) both entered between 2011 and 2015 and maintained m-indices above 2, indicating sustained growth momentum [23,24]. Chen X.Y., Eiber M., Haberkorn U., Kulkarni H.R., Pomper M.G., and Buck A.K. each have h-indices of 18–22 and total citations ranging from 1353 to 3669, reflecting their rapid rise and robust impact over the past decade [25–40]. Collectively, these 10 scholars span entry cohorts from 2011 to 2015 – encompassing both early pioneers and high m-index emerging leaders – and together constitute the core knowledge-producer network within theranostics research.

Table 3.

Top 10 most influential authors

Author	h_index	g_index	m_index	TC	NP	PY_start
Herrmann	30	52	2.727	2891	79	2015
Wester	30	49	2.5	3250	49	2014
Baum	24	45	1.6	2766	45	2011
Hofman	23	44	2.091	2846	44	2015
Chen	22	36	1.467	1353	41	2011
Eiber	20	36	1.818	2392	36	2015
Haberkorn	20	26	1.429	3669	26	2012
Kulkarni	20	23	1.429	2274	23	2012
Pomper	19	32	1.357	1374	32	2012
Buck	18	36	1.636	1381	47	2015

NP, number of publications; TC, total citations.

Table 4 indicates that, in terms of institutional output, Germany’s Helmholtz Association firmly occupies first place with 304 papers – 54 more than the second-ranked Memorial Sloan Kettering Cancer Center (the USA, 250 papers). They are followed by the University of California System (240), the University of Duisburg-Essen (234), and the German Cancer Research Center (DKFZ; 231). With Germany and the USA each holding two of the top five positions, a clear ‘dual German–US core’ emerges. The Peter MacCallum Cancer Center (Australia, 226 papers) and the Technical University of Munich (187) come next, illustrating the complementary roles of comprehensive medical centers and multidisciplinary engineering-oriented institutions in advancing theranostics research. The publication gap among institutions ranked sixth through 10th is less than 70 papers (160–226), suggesting that no absolute ‘output chasm’ has yet formed at the top and that new centers still have potential to break into the leading tier. Overall, large research consortia, cancer-specialty centers, and cross-campus system platforms are the primary drivers of high output in nuclear-medicine theranostics, reflecting the field’s strong internationalization and interdisciplinary collaboration.

Table 4.

Top 10 publishing institutions

Affiliation	Articles
Helmholtz Association	304
Memorial Sloan Kettering Cancer Center	250
University of California System	240
University of Duisburg-Essen	234
German Cancer Research Center (DKFZ)	231
Peter Maccallum Cancer Center	226
Technical University of Munich	187
University of Wurzburg	176
National University of Singapore	174
University of Munich	160

Reference and journal cross-citation analysis

Table 5 lists the 10 journals that publish the largest number of articles on global nuclear-medicine theranostics. The Journal of Nuclear Medicine (229 papers) and Theranostics (212 papers) dominate the landscape, together accounting for roughly one-third of all publications. The European Journal of Nuclear Medicine and Molecular Imaging (132 papers) and the multidisciplinary oncology title Cancers (89 papers) rank third and fourth, respectively, indicating that the field bridges dedicated nuclear-medicine outlets and broader oncology venues. The remaining six journals – such as Nuclear Medicine and Biology, Seminars in Nuclear Medicine, and Molecular Pharmaceutics – serve as high-output platforms for basic radiotracer development, educational reviews, and pharmacy-oriented translational research. Overall, the table shows that core knowledge dissemination in theranostics is concentrated within a few high-impact flagship nuclear-medicine journals, while also progressively diffusing into oncology and pharmaceutical science publications.

Table 5.

Top 10 most frequently cited journals

Sources	Articles
Journal of Nuclear Medicine	229
Theranostics	212
European Journal of Nuclear Medicine and Molecular Imaging	132
Cancers	89
Nuclear Medicine and Biology	63
Seminars in Nuclear Medicine	52
Molecular Pharmaceutics	48
Pharmaceuticals	46
Clinical Nuclear Medicine	43
Applied Radiation and Isotopes	42

Figure 4a presents the CiteSpace journal dual-map overlay for theranostics literature published between 2006 and 2025, projecting ‘citing–cited’ relationships onto disciplinary basemaps on both sides: the left reflects the subject categories of the publishing journals, while the right represents the primary citation sources. Arcs of varying thickness and color trace the flow of knowledge. Early studies were chiefly published in basic-science outlets grouped under ‘Physics/Materials/Chemistry’ and ‘Molecular Biology/Immunology’, citing heavily within the same domains. Over time, the arcs converge toward the right-hand clusters ‘Health/Nursing/Medicine’ and ‘Chemistry/Materials/Physics’, indicating that diagnostic tracers and therapeutic radiopharmaceuticals have migrated from laboratory synthesis and molecular-mechanism research toward clinical application and health sciences. This ‘left-basic to right-clinical’ trajectory reveals a two-decade evolution from physicochemical foundations through molecular biology to clinical translation, underscoring the pivotal role of interdisciplinary integration in accelerating knowledge diffusion and technological iteration.

Fig. 4.

Cross-citation analysis of references and journals: (a) CiteSpace journal dual-map overlay and (b) CiteSpace reference mountain-ridge plot.

Figure 4b displays the reference ‘mountain-ridge’ plot, in which frequently cocited papers from 2006 to 2025 are grouped into clusters numbered #0–#19 in descending size; the height and thickness of each ridge represent the annual citation intensity of that cluster, while the ridge color shifts from red to purple as time progresses. In the early phase (2004–2010), clusters #0 ‘tumor lesion’ and #3 ‘fluorodeoxyglucose F-18’ dominate, anchoring the field’s molecular-imaging foundations around fluorine-18 fluorodeoxyglucose. Between 2011 and 2015, the ridges for #1 ‘prostate cancer’ and #2 ‘peptide receptor radionuclide therapy’ rise sharply, marking the rapid growth of PSMA-targeted imaging and peptide receptor radionuclide therapy. Subsequent spikes in clusters such as #7 ‘bioorthogonal click chemistry’ and #10 ‘human transferrin’ during 2016–2020 denote breakthroughs in drug-delivery systems and radiopharmaceutical conjugation strategies. Over the past five years (2020–2024), smaller new ridges – for example, #12 ‘lung radiolabeled somatostatin analogue’ and #15 ‘potential application’ – have emerged, suggesting that highly specific targets and multimodal diagnostic–therapeutic schemes are the latest hotspots. Overall, the ridge plot vividly traces the rise, peak, and decline of thematic clusters, revealing an evolution from early tumor-imaging foundations through targeted radiotherapy to intelligent drug design and novel-target exploration in nuclear-medicine

Keyword and research hotspot analysis

Figure 5a, based on a keyword co-occurrence network, maps the average publication year for 2006–2025 onto a blue-to-yellow color gradient. The central nodes ‘theranostics’ and ‘PET’ are the largest, indicating that they remain the core concepts in nuclear-medicine theranostic research. Moving upward along the vertical axis, yellow-green nodes such as ‘fibroblast activation protein’ and ‘artificial intelligence’ appear, showing that these emerging terms have surged in 2022–2023. In contrast, the lower blue region retains earlier hotspots like ‘neuroendocrine tumors’ and ‘positron-emission tomography’. The visualization clearly reveals a temporal shift in research focus – from the early foundations of tumor imaging toward recent emphases on FAP-targeted probes and artificial intelligence-enabled diagnostic–therapeutic strategies – thus providing a real-time ‘thermometer’ of keyword-level evolution in this field.

Fig. 5.

Keyword and research hotspot analysis: (a) VOSviewer keyword time-overlay map; (b) CiteSpace keyword cluster plane; (c) keyword timeline clustering map; and (d) keyword burst plot.

Figure 5b, generated via hierarchical clustering, partitions high-frequency keywords into ten island-like thematic clusters – #0 molecular imaging, #1 prostate cancer, #2 peptide receptor radionuclide therapy, #3 breast cancer, #4 FAP, and others. These multidisciplinary blocks are discretely distributed and linked only by sparse boundaries, indicating that theranostics research has evolved into parallel yet complementary knowledge communities spanning tumor-specific diagnosis and therapy, targeted radiopharmaceutical development, and the exploration of new molecular-imaging targets.

Figure 5c plots the same 10 clusters along a time axis: in the early phase (2005–2012), clusters #6 ‘PET’ and #0 ‘molecular imaging’ form the dominant peaks; between 2013 and 2018, clusters #2 ‘peptide receptor radionuclide therapy’ and #1 ‘prostate cancer’ rise steadily; after 2019, clusters #4 ‘fibroblast activation protein’, #7 ‘pretargeted radionuclide therapy’, and #8 ‘photodynamic therapy’ emerge as new high points. The timeline, therefore, reveals a three-stage evolutionary trajectory – from foundational imaging probes, through targeted radionuclide therapy, to tumor-microenvironment/immune-coupled strategies – offering visual evidence for dissecting the field’s technological cadence and milestones.

Figure 5d lists the 20 keywords with the highest burst strengths between 2006 and 2025, together with their burst intervals. Early bursts such as ‘in-vivo evaluation’ (2007–2019) and ‘PET’ (2011–2016) map to the peak of foundational research; mid-period bursts – including ‘Ga-68-labeled PSMA ligand’ (2016–2020) and ‘Lu-177 PSMA-617 radionuclide therapy’ (2018–2021) – correspond to the wave of PSMA-targeted radiotherapy. The most recent bursts, ‘fibroblast activation protein’ (2022–2025) and ‘CXCR4-directed endoradiotherapy’ (2022–2023), signal cutting-edge trends in new targets and immune-radiation coupling concepts. Burst analysis thus quantifies when research hotspots arise and wane, providing data-driven support for predicting future breakthroughs and guiding resource allocation.

Analysis of the three-way relationship among references, authors, and keywords over the past three decades

Figure 6 presents a three-field plot that simultaneously maps cocited core references (CR), prolific authors (AU), and high-frequency descriptors/keywords (DE), outlining the knowledge base, key author cluster, and thematic foci of nuclear-medicine theranostics research. Milestone papers such as Zaknun et al. [37], Strosberg et al. [3], and Hofman et al. [13] occupy the left field, while central nodes are dominated by leading authors – Herrmann K., Hofman M.S., and Wester H.J – whose prominence links them tightly to the field’s evolution. The right field is densely packed with terms like ‘theranostics’, ‘prostate cancer’, ‘PET’, ‘PSMA’, and ‘radionuclide therapy’. Connection lines indicate that (a) most top authors are strongly tied to ‘theranostics’ and ‘PET’, marking them as primary drivers of ongoing conceptual development; (b) Hofman M.S., Wester H.J., and colleagues show intense linkage to ‘prostate cancer’ and ‘PSMA’, underscoring PSMA-guided theranostic integration as the current hotspot; and (c) guideline and review articles such as Zaknun et al. [37] extend to numerous core authors, demonstrating their role as shared knowledge pillars. Collectively, the visualization uncovers a vertical knowledge-transfer chain – ‘classic cocited references → core authors → frontier keywords’ – and illustrates how global theranostics research, grounded in seminal studies, is steered by a small cadre of high-output scholars toward collaborative networks and thematic clusters centered on PET imaging and PSMA-targeted oncologic theranostics.

Fig. 6.

Three-field visualization association plot generated with the R package.

Analysis of the most frequently cited references

As shown in Table 6, the PSMA-targeted radioligand study published by Lindner et al. [5] in 2018 in the Journal of Nuclear Medicine leads the list with 639 total citations and an average of ≈80 citations per year, underscoring its milestone role in advancing the paradigm of precision therapy for prostate cancer. Although only released in 2022, the review by Rindi et al. [41] ranks second, thanks to the highest average annual citations (≈122) and the highest normalized citation score (22.53), indicating that molecular theranostics for neuroendocrine tumors is rapidly becoming a research hotspot. Several Journal of Nuclear Medicine papers published around 2015 (e.g. Benesová et al., Weineisen et al., and Baum et al.) have accumulated substantial citations and laid the technical groundwork for the ⁶⁸Ga/¹⁷⁷Lu dual-isotope ‘diagnosis-therapy integration’ strategy [42,43].

Table 6.

Top 10 references ranked by citation frequency

Paper	DOI	TC	TC per year	Normalized TC
Lindner et al., 2018, J Nucl Med	10.2967/jnumed.118.210443	639	79.88	14.45
Rindi et al., 2022, Endocr Pathol	10.1007/s12022-022-09708-2	487	121.75	22.53
Benesová et al., 2015, J Nucl Med	10.2967/jnumed.114.147413	460	41.82	6.15
Weineisen et al., 2015, J Nucl Med	10.2967/jnumed.115.158550	425	38.64	5.68
Hofman et al., 2015, Radiographics	10.1148/rg.352140164	408	37.09	5.45
Baum et al., 2016, J Nucl Med	10.2967/jnumed.115.168443	395	39.50	7.04
Larson et al., 2015, Nat Rev Cancer	10.1038/nrc3925	376	34.18	5.02
Calais et al., 2019, Lancet Oncol	10.1016/S1470-2045(19)30415-2	356	50.86	8.00
Loktev et al., 2019, J Nucl Med	10.2967/jnumed.118.224469	351	50.14	7.89
Violet et al., 2019, J Nucl Med	10.2967/jnumed.118.219352	338	48.29	7.60

TC, total citations.

Analysis of inter-author collaborative and cocitation relationships

Figure 7a presents the author collaboration network (weighted by the number of co-authored publications), depicting the global ‘author–author’ structure in nuclear-medicine theranostics. Node size denotes the volume of joint papers, and link thickness represents cooperation strength. The red cluster, centered on German scholars such as Herrmann K. and Eiber M., radiates to radioligand development teams led by Hicks R.J. and Hofman M.S., forming a tightly knit Euro-Australian ‘PSMA innovation chain’. The green cluster, revolving around Wester H.J., Buck A.K., and Lapa C., focuses on dosimetric studies of the ⁶⁸Ga/¹⁷⁷Lu ‘diagnosis-therapy integration’ strategy. The blue and purple clusters represent, respectively, American and German imaging experts collaborating on neuroendocrine tumors and FAP-targeted tracers. A loosely connected yellow cluster (Engle J.W., Basu S., and colleagues) links small radiochemistry groups in North America and Asia, suggesting that emerging regions are entering the mainstream network via key nodes. Overall, the network’s high-density and moderate modularity indicate a German-centered yet multinational collaborative innovation landscape, providing a robust cooperative foundation for the rapid iterative advance of theranostic technologies.

Fig. 7.

Analysis of inter-author collaboration and citation relationships: (a) author collaboration network weighted by number of co-authored publications and (b) author collaboration network weighted by cocitation counts.

Figure 7b presents the author collaboration network weighted by cocitation counts. Within the same network topology, node size now represents the total citations received by jointly authored papers, capturing the scholarly impact of collaborative output. Nodes in the red core cluster enlarge further, showing that the Herrmann K., Hofman M.S., and Eiber M. groups are not only highly productive but also the most frequently cited, serving as ‘high-impact engines’ driving paradigm shifts in the field. In the green cluster, the nodes for Wester H.J. and Buck A.K. expand markedly, highlighting their citation advantage in ¹⁷⁷Lu-PSMA and α/β dual-nuclide therapeutic strategies; by contrast, nodes in the blue cluster (Roecker F. et al.) shrink slightly, suggesting that although their early work involved close collaboration, subsequent citation growth has been relatively modest. Emerging yellow–orange nodes (Zhang J., Chen X., and others) remain small, indicating that Asian teams are still in the early phase of transitioning from ‘high output’ to ‘high impact’. Comparing the two maps reveals that productivity and influence do not fully align; some prolific clusters must enhance citation quality, and leading future hotspots will depend on deeper cross-regional and cross-disciplinary collaboration to achieve the efficient translation and global diffusion of integrated diagnostic–therapeutic technologies.

Discussion

Research overview and stage-wise evolution

Between 2006 and 2025, a total of 2452 publications related to nuclear-medicine theranostics were retrieved. Annual output rose from fewer than 40 papers in the first year to nearly 290 in 2024, yielding an average compound annual growth rate of approximately 18.7%. Based on technological milestones, collaboration network dynamics, and the timing of thematic bursts, the field’s evolution can be divided into three stages: (A) exploratory stage (2006–2012) – work centered on synthesizing β-emitting radioligands, such as ¹³¹I and ¹⁷⁷Lu and testing them in early animal models; yearly growth was modest (<10%), and international cooperation showed a loose ‘point-to-line’ pattern; (b) acceleration stage (2013–2018) – dual-nuclide ‘diagnosis-therapy’ concepts, typified by ⁶⁸Ga/¹⁷⁷Lu-PSMA, somatostatin receptor, and FAPI, took shape; researchers such as Herrmann K. and Wester H.J. forged intercontinental collaborations spanning Germany, the USA, and Australia; annual growth surpassed 20%, boosted by major funding from the NIH, EU-EURATOM, and others [2,44]; and (c) translational deepening stage (2019–present) – α-particle therapies (²²⁵Ac-PSMA), tumor-microenvironment targets, and artificial intelligence-driven personalized dosimetry became focal points; global multicenter phase III trials commenced, and the Asia-Pacific region (notably China) exceeded a 25% share of publications, although highly cited work remains dominated by German–American teams. Overall, the three-stage progression – ‘radiochemical groundwork → clinical validation → intelligent precision’ – has laid a solid foundation for theranostics to advance from proof of concept to large-scale clinical adoption.

Comparison with previous literature

Compared with earlier bibliometric studies that focused on a single target or radionuclide – such as Rahmim et al.’s 2016 analysis of 318 papers on ¹³¹I radioiodine therapy and Schuchardt et al.’s 2020 network visualization of 542 ¹⁷⁷Lu-PSMA articles – our investigation spans multiple technological pathways, including β- and α-emitters as well as tumor-microenvironment targets. Both the sample size (2452 papers) and the time window (20 years) are substantially larger, enabling a comprehensive portrayal of the field’s full life-cycle evolution [33,34,45,46]. Beyond traditional cocitation and co-word clustering, we introduced three-field plots and dual-map overlays to uncover, within a three-dimensional framework of ‘knowledge sources–author communities–research themes’, the stepwise progression from physicochemical synthesis to animal studies, clinical validation, and artificial intelligence-enabled precision. This finding aligns closely with the ‘target–radionuclide–platform’ iteration model proposed by Lindner et al. (2018) and with Rindi et al.’s (2022) call for cross-disciplinary collaboration [47,48]. Unlike earlier work that emphasized publication counts or collaboration density alone, our study combines normalized citation counts, burst-term strength, and structural variation coefficients to quantify, for the first time, the speed of hotspot migration and the key bridge nodes in knowledge diffusion. We show that the German–American–Australian triad is not only highly productive but also functions as a ‘conduit’ for knowledge flow, whereas Asian teams – despite rapid output growth – still lag in high-impact citations and cross-domain collaboration, a point seldom addressed in previous literature. In addition, we project trends for the rapid rise of α-particle therapies, new radionuclide supply chains, and artificial intelligence-driven imaging-omics, findings that corroborate recent systematic reviews by Kendig (2024) and Lage (2025) on the core issues of the next 5 years. Collectively, our study significantly extends prior literature in data scale, analytical depth, and theoretical integration, providing forward-looking evidence for strategic planning and policy formulation in the theranostics field.

Research hotspots and future trends

An integrated reading of keyword burst curves, structural variation coefficients, and co-occurrence analyses of frontier citations indicates that theranostic research is rapidly evolving from a field dominated by a limited number of β-emitter platforms toward a more diversified landscape encompassing α-emitters, novel β-/positron-emitting radionuclides, and broader target classes. In addition to extensively studied systems such as ¹⁷⁷Lu-PSMA and ²²⁵Ac-based radioligand therapy, radionuclides, including ⁹⁰Y, ⁶⁴Cu/⁶⁷Cu, ⁸⁹Zr, ²¹²Pb, and ¹⁶¹Tb, are attracting increasing attention because of their complementary physical properties and translational potential in imaging-guided therapy. Likewise, beyond PSMA, somatostatin receptor, and FAP/FAPI, emerging targets such as HER2, integrins (αvβ6, αvβ3), and CXCR4 are expanding theranostic applications across a wider spectrum of malignant diseases. Alpha-particle therapies, represented by ²²⁵Ac-PSMA and ²¹³Bi-labeled agents, remain highly promising because of their high linear-energy transfer and short tissue range, although radionuclide-production capacity, radiochemical stability, and distribution logistics continue to limit broad implementation. At the same time, tumor-microenvironment-related targets and immune-associated strategies are moving closer to clinical translation, while artificial intelligence is increasingly being integrated into dosimetry prediction, image registration, response monitoring, and treatment personalization. Taken together, these findings suggest that the next phase of theranostic development will likely be characterized not by a single dominant target or radionuclide, but by parallel advances in diversified isotope supply, target expansion, multitarget combination strategies, and artificial intelligence-assisted individualized treatment planning. The rationale for highlighting large-scale production of α-nuclides, multitarget probes for the tumor microenvironment, artificial intelligence-driven personalized dosimetry, and cross-regional multicenter trials as likely dominant directions over the next 3–5 years is that these themes sit at the intersection of the strongest recent keyword bursts, current translational bottlenecks, and the practical requirements for broader clinical implementation. In particular, α-nuclide supply remains a recognized rate-limiting step for targeted alpha therapy; tumor-microenvironment-oriented probes such as FAP/FAPI and CXCR4-linked approaches are among the most rapidly expanding target classes; artificial intelligence-based dosimetry addresses an urgent clinical need for scalable treatment personalization; and multicenter trials are indispensable for converting promising single-center experiences into robust evidence suitable for guideline adoption and wider regulatory acceptance [2,15,28–30,33–36,39,44,47–51]. To further reflect the breadth of current development in this area, other widely used radionuclides such as ⁹⁰Y, ⁸⁹Zr, and copper-based systems (e.g. ⁶⁴Cu/⁶⁷Cu), as well as emerging alpha-emitters such as ²¹²Pb and beta-plus/Auger-emitting candidates such as ¹⁶¹Tb, should also be considered as important components of the evolving theranostic landscape [33–36,39,47–51].

Academic and clinical implications

The high-impact collaboration networks and hotspot-migration patterns identified in this study offer three key takeaways for research planning and clinical translation. First, the sustained citation leadership of German-, American-, and Australian-led teams stems from a closed-loop ecosystem that links universities, research institutes, industrial radionuclide-production facilities, and multicenter clinical sites, suggesting that emerging economies should use policy incentives and cross-border good manufacturing practice-center partnerships to build an end-to-end pipeline from radiochemistry to registration trials. Second, data sharing and open science have become critical drivers of citation quality; health authorities worldwide are therefore encouraged to promote interoperable databases of theranostic imaging-omics and clinical outcomes – within ethical frameworks – to accelerate the reproducible validation of artificial intelligence models. Third, clinical practice must strengthen interdisciplinary training and target-validation capacity in parallel; joint ‘nuclear medicine + radiochemistry + data science’ master’s and doctoral programs sponsored by medical and engineering faculties, coupled with incentives for physicians to participate in early phase trial design, could shorten the lag between laboratory discovery and guideline adoption. A triad strategy of policy support, open collaboration, and talent integration is expected to propel theranostics from a research hotspot to a routine clinical standard worldwide, ultimately benefiting a broader population of cancer patients.

Strengths and limitations

Strengths included: (a) a large sample size and long time span enable the capture of macro-level evolutionary trends; (b) cross-validation with multiple tools – bibliometrix, CiteSpace, and VOSviewer – ensures robust conclusions; and (c) the introduction of three-field plots and mountain (peak) charts presents knowledge diffusion and topic migration from multiple dimensions. Limitations were: (a) restricting retrieval to the Web of Science Core Collection may underestimate non-English scholarly contributions; (b) the long-term impact of the most recent papers still requires follow-up; and (c) software threshold settings could affect clustering granularity, although sensitivity analyses indicate that overall trends remain stable.

Conclusion

Over the past two decades, global research in nuclear-medicine theranostics has leapt from β-emitter-based therapies to an era of α-emitters, tumor-microenvironment targets, and artificial intelligence-enabled precision. A US–German dual hub drives most high-quality knowledge output, whereas the Asia-Pacific region, despite a surge in publications, still needs stronger high-impact collaborations. Future breakthroughs are likely to center on four areas: large-scale production of α-nuclides, multitarget probes for the tumor microenvironment, artificial intelligence-driven personalized dosimetry, and cross-regional multicenter trials. This projection is supported by the convergence of recent hotspot analyses with current translational bottlenecks in radionuclide supply, target expansion, personalized dosimetry, and evidence generation through multicenter validation. In parallel, broader clinical translation will also depend on continued development of additional radionuclides such as ⁹⁰Y, ⁶⁴Cu/⁶⁷Cu, ⁸⁹Zr, ²¹²Pb, and ¹⁶¹Tb, together with expanding target systems, including HER2, integrins (αvβ6, αvβ3), and CXCR4 [28–30,33–36,39,47–51]. By reinforcing international cooperation and open data sharing, theranostic technologies can move more rapidly from a research hotspot to routine clinical practice, delivering sustainable momentum for global precision oncology.

Acknowledgements

Conflicts of interest

There are no conflicts of interest.