Chemotherapy induced neurotoxicity in cancer survivors assessed through a dual database bibliometric analysis from 2005 to 2025

Abstract

Background

Cancer survivors frequently suffer from chemotherapy-induced peripheral neuropathy (CIPN) and chemotherapy-related cognitive impairment (CRCI), two of the most common and persistent adverse effects of anticancer therapies. These neurotoxicities not only compromise survivors’ quality of life and daily functioning but also contribute to long-term survivorship challenges. Understanding the mechanisms underlying CIPN and CRCI and identifying effective intervention strategies are essential to improving survivorship care and health outcomes.

Methods

We conducted a dual-database bibliometric analysis using both the Web of Science Core Collection (WoSCC) and Scopus, covering studies published between 2005 and 2025. This dual-database cross-validation approach minimizes database-specific bias and enhances analytical robustness by integrating broader publication coverage and citation patterns. After deduplication, 2,837 articles (1,474 from WoSCC and 1,363 from Scopus) were analyzed. VOSviewer, CiteSpace, and Bibliometrix were employed to visualize research trends, mechanistic investigations, intervention strategies, and global collaboration networks.

Results

Research activity on CIPN and CRCI has significantly increased, with a notable shift from symptomatic descriptions to deeper mechanistic insights, including neuroinflammation, oxidative stress, glial activation, and mitochondrial dysfunction. Recently, novel pathways such as ferroptosis, the gut-brain axis, and BDNF signaling have emerged. Intervention studies have expanded from conventional symptom control to integrated survivorship strategies combining neuroprotective agents, cognitive rehabilitation, physical activity, and psychosocial support.

Conclusion

This study presents the first dual-database bibliometric landscape of CIPN and CRCI, offering valuable insights into their pathophysiological underpinnings and intervention development. By framing these neurotoxicities as critical survivorship issues, the findings emphasize the urgent need for comprehensive, mechanism-driven, and survivor-centered management strategies to enhance long-term quality of life in cancer survivors.

Introduction

With the continuous advancement of systemic cancer therapies—including chemotherapy, targeted agents, and immunotherapies—overall survival rates among patients with malignancies have significantly improved [1, 2]. However, chemotherapy-induced peripheral neuropathy (CIPN) and cancer-related cognitive impairment (CRCI) remain the most common neurotoxic complications, substantially affecting treatment adherence, neurocognitive function, and overall quality of life [3–5]. As the population of cancer survivors expands and multimodal regimens become routine, the long-term neurological sequelae of anticancer agents have attracted growing attention from both oncology and neuropharmacology communities [6]. The urgent need to clarify their underlying mechanisms and to develop effective interventions has shifted these conditions from being considered secondary adverse effects to becoming central issues in precision oncology and pharmacological risk management [7]. Importantly, CIPN occurs in approximately 40–70% of patients receiving neurotoxic agents, while CRCI is reported in up to 40% of survivors, often persisting for years after therapy completion [8]. Despite their high prevalence and clinical significance, these complications remain underrecognized and insufficiently managed in routine practice.

Extensive research over the past two decades has revealed that neurotoxicity induced by anticancer therapies is mediated by a complex interplay of molecular, cellular, and systemic mechanisms. For CIPN, dysregulation of ion channels, mitochondrial dysfunction, oxidative stress, microtubule disruption, axonal degeneration, and aberrant neuroimmune signaling have all been implicated [9, 10]. Distinct chemotherapeutic classes such as platinum agents, taxanes, vinca alkaloids, and proteasome inhibitors display characteristic neurotoxic profiles, yet they frequently converge on overlapping downstream pathways of neuronal injury [11, 12]. Recent evidence also emphasizes the role of neuroinflammatory cascades, transcriptomic and proteomic alterations, and glial activation in shaping persistent nociceptive states [13–15]. For CRCI, preclinical and clinical studies indicate a multifactorial etiology involving hippocampal synaptic loss, impaired neurogenesis, white matter and oligodendrocyte pathology, cerebrovascular changes, and neuroimmune dysregulation [16, 17]. Disruption of the blood–brain barrier, oxidative injury, altered gut–brain axis signaling, and deficits in neurotrophic support such as reduced BDNF have further been reported [18, 19]. Genetic and host-related factors including estrogen deficiency, aging, and APOE polymorphisms may modulate susceptibility, particularly in vulnerable populations [20, 21]. Emerging technologies such as single-cell RNA sequencing, spatial transcriptomics, and patient-derived neuronal models are beginning to provide cell-type–specific insights into drug-induced neurotoxicity, yet cross-study comparability remains limited by heterogeneity in regimens, patient genetics, and methodological approaches [22, 23]. Altogether, anticancer drug-induced neurotoxicity represents a spectrum of overlapping syndromes with both peripheral and central components, underscoring the need for systems-level approaches integrating molecular pathology, clinical phenotypes, and biomarker discovery to advance precision diagnostics and targeted neuroprotection [24].

Although research on the mechanisms and clinical manifestations of chemotherapy-induced neurotoxicity has steadily expanded, the literature remains fragmented across cancer types, drug classes, and methodological designs. While systematic reviews are valuable for synthesizing evidence on specific clinical questions or therapeutic interventions, they are less suited for analyzing large-scale, interdisciplinary bodies of literature spanning decades [25]. In contrast, bibliometric analysis enables the quantitative assessment of publication patterns, collaboration networks, citation structures, and thematic evolution, making it particularly appropriate for mapping knowledge landscapes. Most prior bibliometric studies rely on a single database—typically WoS or Scopus—making them susceptible to bias due to differences in journal coverage, indexing delays, and metadata structures [26]. For example, WoSCC is recognized for its rigorous citation tracking and long-term coverage of high-impact biomedical journals, whereas Scopus offers broader interdisciplinary coverage, faster indexing, and better inclusion of non-English and regional literature [27]. As a result, analyses based solely on one database may overlook important publications or introduce systematic bias. Dual-database cross-validation provides a more comprehensive and reproducible strategy by improving data representativeness, reducing systemic bias, and enhancing generalizability [28]. This approach enables a more reliable identification of evolving research themes, collaboration networks, and knowledge gaps. It is particularly valuable in interdisciplinary fields such as oncology neuropharmacology, where mechanistic complexity and thematic heterogeneity require a panoramic, data-driven understanding. Integrating dual-source bibliometric analysis not only strengthens the methodological foundation of trend mapping but also facilitates strategic planning for future mechanistic and translational research.

To bridge these gaps, the present study performs a dual-database bibliometric and knowledge-mapping analysis of anticancer drug-induced neurotoxicity using both WoSCC and Scopus. By combining cross-database validation with multidimensional visualization, we systematically characterize global research output, collaboration patterns, mechanistic hotspots, and citation trajectories from 2005 to 2025. We place particular emphasis on the evolution of peripheral and central neurotoxic pathways and on translating mechanistic insights into intervention strategies. Furthermore, we highlight landmark publications and emerging frontiers that are shaping the field. Our goal is to provide a panoramic, data-driven framework that informs both basic research and translational neuropharmacology. To our knowledge, this is the first study to apply a dual-database strategy to map the global landscape of anticancer drug-induced neurotoxicity, offering novel insights into its thematic evolution and future priorities. Ultimately, this study seeks not only to clarify structural and thematic trajectories but also to establish a data-driven foundation for future mechanistic studies, biomarker discovery, and translational strategies aimed at neuroprotection. By mapping the intellectual landscape and identifying research gaps, our findings are expected to support more precise, mechanism-informed approaches to managing neurotoxic side effects in oncology pharmacology.

Materials and methods

Data source and search strategy

This study used two bibliographic sources—Web of Science Core Collection (WoSCC) and Scopus—as inputs for the bibliometric analysis. We retrieved records published from January 1, 2005 to July 5, 2025 (final search on July 8, 2025). We included English-language Articles and Reviews only. After deduplication by DOI and title, 1,474 WoSCC records (1,147 Articles; 327 Reviews) and 1,363 Scopus records (1,037 Articles; 326 Reviews) were retained. The screening workflow is shown in Fig. 1 and detailed in the “Data Cleaning” section.

Fig. 1.

Data Collection and Bibliometric Analysis Flow

Data cleaning and inclusion criteria

To ensure the accuracy and reliability of the analysis, we performed the following steps. Deduplication: Articles were deduplicated using DOI and article title to ensure each publication appeared only once in the dataset. Document Type Filtering: Non–peer-reviewed literature, including meeting abstracts, editorial materials, conference proceedings, and book chapters, was excluded. Only Articles and Reviews in English were retained. Metadata Harmonization: Records from WoSCC and Scopus were reconciled for author names, affiliations, source titles, and keywords, with manual checks performed when necessary. The final dataset included 2,837 valid records, comprising 1,474 from WoSCC and 1,363 from Scopus, which formed the basis for subsequent analyses.

Bibliometric and statistical analyses

To construct a comprehensive knowledge framework and reveal trends in cancer therapy-induced neurotoxicity research, several bibliometric tools and statistical methods were employed: CiteSpace (Version 6.4.R1): Keyword Burst Detection: CiteSpace was used to detect keyword bursts, identifying emerging hotspots in cancer neurotoxicity research, highlighting terms with rapid growth in specific time periods [29]. Reference Co-citation Clustering: Co-citation networks identified influential papers and research directions, revealing the interdisciplinary nature of research topics. Collaboration Networks: We used CiteSpace to construct networks identifying key authors, institutions, and countries, providing a global perspective on academic exchanges. VOSviewer (Version 1.6.20): Keyword Co-occurrence Networks: VOSviewer analyzed co-occurrence of keywords, identifying core terms frequently mentioned in cancer neurotoxicity research. A threshold of ≥ 10 occurrences created the high-frequency co-occurrence network, reflecting major research trends. Co-authorship and International Collaboration Networks: VOSviewer also constructed co-authorship networks and visualized international collaboration patterns, emphasizing cross-border cooperation. Bibliometrix R-package (Version 4.2.1): Publication Trend Analysis: We analyzed annual publication counts and citation data to visualize trends, highlighting research growth in various topics. Thematic Evolution and Author Timelines: Bibliometrix tracked thematic evolution and showcased influential scholars and emerging topics. Institutional Productivity Analysis: We identified the top producers of publications and citations in the field. Scimago Graphica (Version 1.0.51): Data visualization tools refined and finalized visual outputs, ensuring all figures met publication-quality standards.

Statistical methods

Descriptive Statistics: Descriptive statistics were performed on publication counts, citation distributions, and H-index values to evaluate the distribution and impact of research in cancer therapy-induced neurotoxicity. Basic statistical measures including the mean, median, standard deviation, and range were calculated, along with citation count distribution to assess academic impact. Regression and Correlation Analysis: Linear Regression: Linear regression was used to assess the relationship between publication volume and time (2005–2025), revealing trends in research output over time. The regression model’s slope coefficient and R² value were used to quantify the trend and goodness of fit. Polynomial Fitting: Polynomial fitting was applied to explore potential non-linear trends in research output and citation frequency. This method captures more complex, curvilinear patterns than linear regression, allowing for the detection of accelerating or decelerating trends in research focus, particularly for immunotherapy-related neurotoxicity. Spearman’s Rank Correlation Coefficient: Spearman’s rank correlation coefficient was used to analyze the association between publication volume and citation frequency. This non-parametric test identified the strength and direction of the relationship between these two variables. Statistical Significance: A two-tailed P-value of < 0.05 was considered statistically significant. For both linear regression and Spearman correlation, P-values and confidence intervals (CIs) were calculated to assess statistical significance and the precision of estimates. Robustness of Analysis: Sensitivity analyses were conducted to test the consistency of findings across subsets of data (e.g., by cancer type). These steps ensured the results were not influenced by outliers or biases.

Results

Dual database analysis of research trends and citation impact

In this study, we performed cross-validation using WOS and Scopus to assess research output and citation impact in cancer treatment drug-induced neurotoxicity. This dual-database approach ensures result consistency while highlighting subtle differences. Figure 2A shows the annual trends in publications and citations from 2005 to 2025. Both databases demonstrate consistent increases, with WoSCC showing a marked surge in publications after 2019, whereas Scopus maintained a steady growth rate. Figure 2B compares the subject category distribution, with oncology remaining dominant in both databases, followed by pharmacology, pharmacy, and neurosciences. Scopus reported more publications in pharmacology, while WoSCC had a slight advantage in clinical neurology. Figure 2C presents cubic polynomial fits for publication data, with R² values of 0.92 and 0.95, respectively, indicating strong model fit and steady growth in both databases.

Fig. 2.

Publications and Citation Analysis. A Line charts illustrating the annual research output (upper panel) and citation trends (lower panel) in the WoSCC and Scopus databases from 2005 to 2025. Each line represents the cumulative publication or citation volume for both databases, providing a comparison of research activity and academic impact over time. B Bar charts comparing the distribution of publications across different subject categories in both WoSCC and Scopus. Categories include oncology, pharmacology, neuropharmacology, and related fields. Data for each category are presented for both databases to highlight their distinct indexing practices. C The polynomial fitting curves were applied to the publication and citation data from WoSCC and Scopus. Higher-degree polynomial fitting allowed us to capture more complex patterns in the data, revealing accelerating growth in research output and citation frequency over time. The fitted curves for WoSCC and Scopus data have R² values of 0.92 and 0.95, respectively, indicating that the model explains the relationship between time trends and research output with a high degree of precision

Global research collaboration patterns and National research output

Using WOS and Scopus, we assessed global research output and collaboration patterns in cancer treatment drug-induced neurotoxicity. Table 1 summarizes the top corresponding author countries, showing the USA as the leading contributor (29.6% in WoSCC; 28.6% in Scopus), followed by China and Italy. Notably, the USA also exhibited the highest number of single-country publications (SCP), while Germany and Canada had the highest proportion of multi-country publications (MCP %), reflecting strong international collaboration. Figure 3A shows the geographical distribution of research output, with the USA, China, and several European countries leading in research efforts. The USA contributed 29.6% of total publications in WOS and 28.6% in Scopus, followed by China, Italy, and other European countries. Figure 3B presents the country collaboration network, revealing strong ties between the USA, China, and European nations. Scopus shows a broader network of international collaboration, especially between Asian and European countries. Figure 3C compares research output trends for key countries from 2005 to 2025, with the USA, China, and Italy showing consistent growth. Figure 3D highlights the collaboration network among top publishing countries, with stronger ties in Scopus, particularly with Japan, the Netherlands, and Canada. Table 2 shows the USA’s leading citation impact, with 25,617 citations in WOS and 20,270 in Scopus, consistent with its research output.

Table 1.

Top corresponding author countries

WOS
Country	Articles	Articles %	SCP	MCP	MCP %
USA	437	29.6	357	80	18.3
China	154	10.4	138	16	10.4
Italy	127	8.6	105	22	17.3
France	76	5.2	62	14	18.4
Japan	75	5.1	71	4	5.3
Germany	69	4.7	50	19	27.5
Australia	56	3.8	44	12	21.4
Greece	48	3.3	37	11	22.9
Netherlands	46	3.1	35	11	23.9
Spain	44	3	34	10	22.7
Scopus
USA	390	28.6	313	77	19.7
China	151	11.1	130	21	13.9
Italy	105	7.7	79	26	24.8
Australia	62	4.5	41	21	33.9
Germany	46	3.4	27	19	41.3
Spain	43	3.2	29	14	32.6
Japan	41	3	34	7	17.1
Canada	37	2.7	24	13	35.1
France	35	2.6	30	5	14.3
India	33	2.4	24	9	27.3

Fig. 3.

Global Research Collaboration and Output. A Geographical distribution and collaboration map. Each circle represents a country, with the size proportional to publication volume. Connecting lines indicate international collaborations, and different colors denote distinct collaboration clusters. Larger circles highlight major contributing countries and their broad partnerships. B Country collaboration network. A circular chord diagram visualizing co-authorship and collaborative ties among countries. Line thickness indicates the strength of collaboration, while outer labels identify specific countries. C Research output trends. Line charts show annual publication trends from 2005 to 2025 for the most productive countries, as indexed in WoS (upper panel) and Scopus (lower panel). D Collaboration network analysis. A network map illustrating inter-country collaborations. Node size is proportional to publication volume, and edge thickness reflects collaboration intensity

Table 2.

Research output and citations by country

WOS
Country	TC	Average Article Citations	Freq
USA	25,617	58.60	2101
Italy	6975	54.90	639
Netherlands	4237	92.10	252
China	3600	23.40	526
France	3258	42.90	519
Australia	2593	46.30	278
Japan	2398	32.00	394
Greece	2132	44.40	181
Germany	1554	22.50	461
Canada	1132	36.50	268
Scopus
USA	20,270	52.00	1811
Italy	4701	44.80	485
Australia	3030	48.90	380
United Kingdom	2892	115.70	261
China	2721	18.00	570
Netherlands	1890	82.20	143
Spain	1380	32.10	240
Canada	1370	37.00	224
France	1339	38.30	291
Germany	1231	26.80	309

Author and institutional collaboration networks and their scholarly contributions

Through cross-database validation, we analyzed author and institutional collaboration in cancer research. Figure 4A and B display author collaboration networks in WOS and Scopus, respectively, with key institutions like Mayo Clinic, Memorial Sloan Kettering Cancer Center, and University of Texas MD Anderson Cancer Center at the center. Scopus shows a broader scope of global collaboration, especially in interdisciplinary research. Figure 4C illustrates the institutional collaboration network, where node size reflects publication volume, and color indicates collaboration type. Mayo Clinic and Memorial Sloan Kettering Cancer Center lead in both research output and collaboration. Table 3 lists the top 10 institutions, with Mayo Clinic ranking first in both publications (55 articles) and citations (5,080). Figure 4D highlights strong collaboration ties between leading institutions, especially in North America, Europe, and Asia. Figure 4E shows key authors’ publication trends, emphasizing their sustained contributions to neurotoxicity research. Figure 4F displays the flow between authors and research topics, emphasizing the growing trend of cross-disciplinary collaboration.

Fig. 4.

Author and Institutional Collaboration. A Author collaboration network. A co-authorship network map where each node represents an author, node size indicates the number of publications, and connecting lines denote co-authorship relationships. Cluster colors reflect collaborative groups within the field. B Top collaborative authors. Visualization of the most frequently collaborating authors, with node size corresponding to publication output and edge thickness representing collaboration intensity. C Institutional collaboration network. A network showing inter-institutional collaborations. Node size indicates institutional publication volume, and edge thickness represents collaboration strength. Different colors mark clusters of institutions with frequent collaboration. D Major institutional collaboration ties. Highlighted connections between leading institutions, illustrating global hubs of scientific cooperation. E Author publication trends. Line charts tracking annual publication activity of the most prolific authors from 2005 to 2025, showing consistent increases and identifying researchers driving growth in the field. F Author-topic flow analysis. A Sankey diagram visualizing how leading authors’ research themes evolved over time, mapping shifts in focus from early descriptive studies to recent mechanistic and translational investigations

Table 3.

Institutional and Co-author collaboration analysis

Top 10 Organizations
Organization	Documents	Citations	Total Link Strength
Univ Milano Bicocca	46	3515	582
Mayo Clinic	55	5080	407
Univ Padua	19	1402	310
Univ Sydney	31	813	260
Univ Texas MD Anderson Cancer Center	53	3529	193
Univ Michigan	25	1004	185
Univ Rochester	24	2047	179
Mem Sloan Kettering Cancer Center	43	4067	166
Ohio State University	29	1796	164
Johns Hopkins University	22	2312	151
Top 10 Co-author Organizations
Mem Sloan Kettering Cancer Center	43	4067	53
Mayo Clinic	55	5080	47
Univ Texas MD Anderson Cancer Center	53	3529	44
Dana Farber Cancer Institute	25	2388	42
Duke University	20	3791	40
Univ Rochester	24	2047	35
Ohio State University	29	1796	31
Univ Michigan	25	1004	31
Univ Milano Bicocca	46	3515	25
Univ Penn	21	998	22

Journal citation and academic source analysis

The journal and citation network analysis reveals key trends in cancer research. Figure 5A presents the journal co-citation network, with influential journals like Journal of Clinical Oncology and Annals of Oncology at the center. These journals play a pivotal role in disseminating research findings on cancer treatment drug-induced neurotoxicity. Figure 5B shows the top collaborative journals, with larger nodes indicating more frequent co-citations. Journal of Clinical Oncology and Annals of Oncology again show extensive collaboration within the research community. Table 4 provides detailed data on the top 10 journals by citation count and productivity. Journal of Clinical Oncology leads in both citation count (131,409) and total link strength (6,196), followed by Annals of Oncology and New England Journal of Medicine. Figure 5C illustrates core journal sources based on Bradford’s Law, showing the decreasing concentration of core journals as the total number of journals increases. Figure 5D presents the most globally cited documents, with larger nodes indicating higher influence. Figure 5E highlights the top 25 references with the strongest citation bursts from 2005 to 2025, illustrating their significant impact on the field.

Fig. 5.

Citation and Source Analysis. A Journal co-citation network. Nodes represent journals, and edges between them indicate co-citation relationships, with edge thickness proportional to co-citation frequency. B Top collaborative journals. Bar length corresponds to the number of collaborative publications, while color differentiates categories of journals. C Core journal sources. Bars are proportional to publication volume, with colors distinguishing between different research domains. D Most globally cited documents. Each bar reflects the total citation frequency of a document, with different colors denoting the source journal. E Top 25 references with the strongest citation bursts. The red segments indicate periods of strong citation bursts, while blue lines represent the full citation timeline

Table 4.

Top 10 journals by citation count and productivity

Top 10 Journals by Citation Count			Top 10 Productive Journals by Output
Source	Citations	Total Link Strength	Source	Documents	Citations	Total Link Strength
J Clin Oncol	6196	131,409	Supportive Care in Cancer	63	2369	242
Ann Oncol	1847	49,821	European Journal of Cancer	27	2137	131
New Engl J Med	1424	37,728	Annals of Oncology	29	2007	117
Clin Cancer Res	1226	34,799	Breast Cancer Research and Treatment	23	1025	114
Cancer-Am Cancer Soc	1034	33,002	Oncologist	20	1021	104
Support Care Cancer	1279	32,385	Journal of Clinical Oncology	30	5882	101
Eur J Cancer	1054	28,291	Cancers	51	836	86
Blood	1109	27,111	Acta Oncologica	16	578	69
Pain	1052	23,241	Cancer	26	1724	67
Neurology	748	23,173	BMC Cancer	30	696	64

Keyword co-occurrence and research topic evolution trends

The keyword co-occurrence and topic network analysis reveal key research trends in cancer treatment drug-induced neurotoxicity. Figure 6A shows the keyword co-occurrence network in both WOS and Scopus, highlighting core terms like neurotoxicity, chemotherapy, and paclitaxel. Figure 6B illustrates the top collaborative keywords, with larger nodes indicating stronger co-occurrence, reflecting the interdisciplinary nature of the research. Figure 6C presents the trend analysis of key terms over time, showing the rise of topics such as chemotherapy-induced peripheral neuropathy and cognitive impairment in recent years. Figure 6D shows the most globally cited documents, with larger nodes indicating greater citation impact. Figure 6E highlights the top 25 keywords with the strongest citation bursts from 2005 to 2025, emphasizing the rapid dissemination of critical terms in cancer-related neurotoxicity research. This analysis underscores the growing focus on understanding the long-term side effects of cancer treatments and the increasing trend of cross-disciplinary collaboration in the field.

Fig. 6.

Keyword and Topic Network Analysis. A Keyword co-occurrence network. Nodes represent keywords, and edges between them indicate their co-occurrence across studies. The thickness of the edges reflects the frequency of co-occurrence. B Top collaborative keywords. Bar length reflects the number of co-occurrences, and colors differentiate categories such as mechanisms, clinical outcomes, or interventions. C Keyword trend analysis. Each line represents the usage trend of a specific keyword, with line slope and trajectory indicating whether a term is gaining or losing prominence. D Citation trends of the most influential documents. Lines represent annual citation counts of selected documents, with different colors distinguishing individual studies. E Top 25 keywords with the strongest citation bursts. The red segments indicate periods of strong citation bursts, while blue lines represent the full citation timeline

Citation burst analysis and milestone references

The citation burst and key reference analysis provide insights into the evolution of cancer treatment drug-induced neurotoxicity research. Figure 7A illustrates the keyword co-occurrence network, highlighting central terms such as chemotherapy-induced peripheral neuropathy and neurotoxicity, which reveal the core topics that have shaped the research landscape. Figure 7B shows the trend analysis of key topics, emphasizing the growing prominence of terms such as paclitaxel and cognitive impairment, indicating increasing interest in the neurotoxic effects of cancer treatments. Figure 7C presents the top 10 representative influential publications from 2005 to 2025, identified through bibliometric indicators including citation frequency, normalized citation counts, co-citation centrality, and burst detection. These studies span different but interconnected dimensions of the field. Early foundational work (e.g., Postma 2005 in Eur J Cancer) established the clinical significance and characterization of chemotherapy-induced peripheral neuropathy, providing a historical cornerstone for subsequent investigations [30]. Highly cited clinical trials and cohort studies published in Journal of Clinical Oncology (Falcone 2007; Lin 2024) and Lancet Oncology (Bahadoer 2021) have contextualized neurotoxicity within evolving treatment strategies, highlighting risk–benefit trade-offs and survivorship issues [31–33]. Mechanistic and translational research, including contributions from Nature Medicine (Norelli 2018) and Molecular Cancer Therapeutics (McWhinney 2009) [34, 35], has clarified drug- and immune-mediated pathways linking treatment exposure to neurotoxicity, while reviews in Nature Reviews Immunology (Morris 2022) have consolidated immune-related toxicity as a central theme [36]. Additional references from JAMA Oncology (Dueck 2015) and Journal of Immunotherapy (Morgan 2013) underscore the methodological advances in symptom measurement and the integration of immunotherapy safety data, whereas interdisciplinary contributions such as Gorelick (2010, Ann NY Acad Sci) bridge oncology with neurology and cognitive outcomes [37–39]. Together, these publications illustrate how the field has evolved from early descriptive work to clinically anchored and mechanistically enriched investigations, while simultaneously integrating immuno-oncology, pharmacology, and survivorship perspectives. The analysis highlights the ongoing shift in research priorities from short-term treatment outcomes to long-term survivorship, patient-reported neurocognitive function, and the development of mechanism-informed interventions.

Fig. 7.

Citation Burst and Key Reference Analysis. A Keyword co-occurrence network. Visualization of the most frequent keywords and their interconnections, with node size proportional to keyword frequency and edge thickness representing co-occurrence strength. Different colors denote thematic clusters. B Key topic trend analysis. Temporal distribution of emerging research topics identified by burst detection, illustrating the evolution of the field from 2005 to 2025. C Top 10 representative influential publications. The ten most influential and highly cited papers identified through bibliometric indicators (citation frequency, normalized citation counts, co-citation centrality, and burst detection)

Discussion

This study offers a comprehensive bibliometric and knowledge-mapping analysis of the global research landscape on cancer treatment drug-induced neurotoxicity between 2005 and 2025. By uniquely adopting a dual-database cross-validation strategy that integrates data from both WoSCC and Scopus, we enhanced the completeness, representativeness, and methodological rigor of our findings while reducing structural biases inherent in single-database analyses. WoSCC emphasizes high-impact biomedical journals, whereas Scopus provides broader interdisciplinary coverage and faster indexing, making their combination particularly suitable for emerging cross-disciplinary topics. Our results demonstrate a steady increase in publications and citations in this area, reflecting growing global interest in the long-term neurotoxic effects of cancer therapies. Keyword co-occurrence and citation burst analyses consistently identified CIPN and CRCI as major hotspots, with more recent attention shifting toward neurotoxicity related to immunotherapy and targeted therapy. Importantly, the thematic evolution maps and burst detection analysis further revealed a progressive transition in research focus—from early descriptive studies centered on symptom characterization (e.g., “neuropathy,” “pain,” “cognitive dysfunction”) toward mechanistic investigations emphasizing neuroinflammation, ion channel dysregulation, and glial activation. Journal co-citation analysis confirmed the central role of high-impact journals such as the Journal of Clinical Oncology and Annals of Oncology, and collaboration network mapping revealed increasingly mature global, institutional, and author-level partnerships. It is noteworthy that Scopus revealed relatively denser cross-regional collaborations, particularly between Asian and European countries. This difference may be attributable to its broader journal coverage, which includes more interdisciplinary and non-English sources, thereby capturing a wider spectrum of international collaboration. Nonetheless, both databases consistently converged on the same set of leading countries, underscoring the robustness of the observed collaboration patterns. Beyond neurotoxicity, radiotherapy-associated cardiotoxicity has also been reported, underscoring the need to evaluate long-term multi-organ adverse outcomes across treatment modalities [40]. Taken together, this study provides a structured knowledge framework and formally demonstrates, through bibliometric evidence, that the field has shifted from simple clinical description to mechanistic exploration, thereby establishing a methodological reference point for future multidisciplinary investigations of drug-induced neurotoxicity.

Our bibliometric analysis reveals a clear thematic evolution in cancer treatment drug-induced neurotoxicity research, highlighting the increasing integration of mechanistic investigation and therapeutic development. Initially centered on symptom characterization, the field has progressively shifted toward uncovering cellular and molecular underpinnings—notably neuroinflammation, ion channel dysregulation, glial activation, and impaired central nervous system plasticity [41]. This shift is not only evident in the experimental literature but is also quantitatively supported by our bibliometric findings: burst detection showed a decline in early descriptive keywords such as “neuropathy” and “pain,” while recent years saw the emergence of mechanistic terms like “microglia,” “ion channel,” and “neuroinflammation.” Co-citation clustering further revealed highly cited groups of papers focused on cellular and molecular pathways, confirming a field-wide transition toward mechanistic exploration. In CIPN, accumulating evidence implicates sensory ion channels such as TRPA1, TRPV4, Nav1.7, and P2 × 3 in aberrant nociceptive signaling [42]. Preclinical studies, including those by Pesce et al. (2022), suggest that oxaliplatin can modulate TRPA1 via oxidative stress and inflammatory mediators, while taxane-induced neuropathy has been linked to P2 × 3 upregulation [43, 44]. These ion channels are therefore being explored as potential drug targets, though most findings remain at the experimental stage. At the neuroimmune level, activation of microglia and astrocytes by agents such as paclitaxel, cisplatin, and doxorubicin is consistently associated with the release of pro-inflammatory cytokines, including TNF-α, IL-6, and IL-1β [45]. Inhibition of glial activation has shown beneficial effects in animal models [46], and microRNAs such as miR-124 have been implicated in regulating these inflammatory cascades [47]. The prominence of keywords such as “microglia,” “miRNA,” and “neuroinflammation” in our analysis underscores the growing focus on immunomodulation in this field [48]. For CRCI, converging evidence from neuroimaging and molecular studies points to hippocampal atrophy, white matter demyelination, reduced BDNF expression, and impaired synaptic plasticity as key biological correlates [49]. Functional MRI has revealed connectivity deficits, and inflammatory markers have been linked to memory and executive dysfunction [50]. Recent studies further implicate signaling pathways involving BDNF, synapsin-1, NF-κB, and the NLRP3 inflammasome in chemotherapy-related hippocampal injury [51, 52]. These molecular cascades correspond with bibliometric signals showing a growing cluster of references around neuroinflammatory and neurodegenerative pathways, providing convergent evidence for the mechanistic reframing of CRCI. Non-pharmacological interventions are also gaining attention, with aerobic exercise, resistance training, and acupuncture shown to modulate neuroinflammation and support synaptic function [53, 54]. Our clustering analysis likewise highlighted “exercise,” “BDNF,” and “non-pharmacologic intervention” as rising keywords, further validating the translational integration observed in experimental studies. The gut–brain axis has recently emerged as another research focus. Chemotherapy-induced microbiota alterations may affect systemic inflammation and blood–brain barrier integrity, thereby influencing neurotoxic outcomes [55]. Early studies suggest probiotics or prebiotics can mitigate CIPN symptoms and cognitive decline, potentially via short-chain fatty acid signaling and microglial modulation [56, 57]. Overall, current evidence suggests that inflammation, ion channel dysfunction, glial activation, and microbiota imbalance together form a multifaceted framework for understanding neurotoxicity [58]. These insights provide opportunities for interventions ranging from small-molecule inhibitors and glial modulators to RNA-based and microbiota-directed therapies, complemented by lifestyle strategies. Future efforts should prioritize validation in large, multicenter human cohorts, the development of mechanism-guided biomarkers, and the application of high-throughput platforms to identify novel neuroprotective agents [41, 59]. As survivorship continues to grow, establishing robust pharmacological frameworks for preventing and managing neurotoxicity is both scientifically essential and clinically urgent. Our findings offer an evidence-based overview of this evolving landscape and point toward mechanism-informed, translationally actionable directions.

This study adopted a rigorous dual-database strategy, combining the WoSCC and Scopus to enhance data comprehensiveness, minimize selection bias, and improve analytical robustness. This cross-validation approach allowed us to identify converging trends while capturing subtle database-specific differences in subject classification, journal indexing, and citation structures. By leveraging the complementary strengths of WoSCC in citation tracking of high-impact biomedical journals and Scopus in broader interdisciplinary and regional coverage, our design ensured that the knowledge map is both systematic and globally representative. However, several limitations remain. First, the structural differences between databases may still result in the exclusion of relevant but unindexed studies, particularly those published in regional, non-English, or newly established journals. Although our search strategy was carefully constructed using Boolean logic and thematic refinement, keyword inconsistencies and indexing delays—particularly in WoSCC—may introduce coverage bias. Thus, while the dual-database approach improves robustness and reduces the risk of systematic bias compared with single-source studies, it cannot fully eliminate disparities caused by indexing practices and metadata quality. Second, the bibliometric tools used, including CiteSpace and VOSviewer, rely on co-occurrence frequency and algorithmic clustering, which are sensitive to parameter settings. For instance, applying a minimum keyword frequency threshold may overlook emerging concepts. Third, metadata quality is dependent on automated indexing and may include errors in author affiliation, keyword tagging, or institutional classification. Finally, as a bibliometric study, our work reflects published trends but cannot assess causal relationships or clinical effectiveness. Future research should consider supplementing bibliometric insights with systematic reviews, expert validation, and real-world data to further substantiate thematic interpretations and support translational research.

By cross-validating data from both WoSCC and Scopus, this study constructed a comprehensive global knowledge map of cancer treatment drug-induced neurotoxicity, systematically delineating thematic trajectories, mechanistic priorities, and intervention trends. Mechanistically, neuroinflammation, ion channel dysfunction, and glial activation have emerged as core pathways, indicating that neurotoxicity is not merely an isolated adverse effect but a multifactorial pathological process involving neuroimmune dysregulation, metabolic perturbations, and impaired brain plasticity. Future research should prioritize several directions: the establishment of mechanism-based translational models to assess the reversibility and therapeutic potential of critical pathways in both animal models and multicenter clinical cohorts; the promotion of standardized diagnostic criteria and grading systems for CIPN and CRCI to improve cross-study comparability—an approach exemplified by evidence-based guidelines for organ-specific toxicities such as radiation-induced bladder injury in the radiotherapy field [60], which highlight the progressive move toward standardized cross-modality toxicity management; the integration of patient-reported outcomes, wearable devices, and neuroimaging to enable personalized and dynamic monitoring of neurotoxicity; and the further exploration of underinvestigated domains such as the gut–brain axis, epigenetic regulation, and sex-specific susceptibility, which may provide novel targets for neuroprotective interventions. In summary, despite certain limitations, this study provides a rigorous and data-driven framework that supports drug development, risk stratification, and mechanism-guided interventions, thereby advancing both research and clinical practice.

Conclusion

This bibliometric study provides a systematic and methodologically robust mapping of the research landscape on cancer treatment drug-induced neurotoxicity, integrating WoSCC and Scopus data. Our findings document a clear thematic transition from descriptive observations to mechanistic inquiry, particularly focused on neuroinflammation, glial activation, and ion channel dysregulation. Future work should prioritize: (1) sustained dual-database surveillance complemented by advanced bibliometric algorithms to capture emerging trends; (2) integration of bibliometric evidence with multi-omics and clinical registries to strengthen translational relevance; (3) establishment of consensus-based diagnostic and grading frameworks for CIPN and CRCI through multicenter collaboration; and (4) incorporation of digital health platforms and biomarker-driven monitoring for individualized neurotoxicity management.

Author contributions

CRediT: Boxiang Zhang: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Writing – original draft; Lucy Yue Lau: Conceptualization, Data curation, Methodology, Software; Zhimin Wu: Conceptualization, Data curation, Formal analysis, Methodology, Software, Supervision, Visualization, Writing – review & editing; Yi Chen: Conceptualization, Data curation, Formal analysis, Methodology, Software, Supervision, Visualization, Writing – review & editing.

Funding

This study was not funded.

Data availability

Datasets are available upon request from the corresponding author.

Declarations

Ethical approval and consent to participate

As this study is a bibliometric analysis based on previously published data, no human participants, animals, or clinical samples were directly involved. Therefore, ethical approval was not required.

Consent for publication

All authors have read and approved the final version of the manuscript and consent to its publication in this journal.

Competing interests

The authors declare no competing interests.