The Web of Science, which is broadly used in bibliometrics, can provide extensive and authoritative global academic data bibliometric software needs (20, 21). We mainly use the Web of Science Core Collection (WoSCC). The data were obtained on 21 July 2022 from the WoSCC database. The search formula was [TS = (alternative splicing)] AND TS = (cancer* OR anticancer* OR tumor* OR oncology OR neoplasm* OR carcinoma* OR lymphoma* OR sarcoma* OR leukemia*). The wildcard character (*) was used to allow variable endings of keywords to capture as much data as possible. There was a limitation on the publication year (2012–2021). The inclusion of English-language literature was limited to original articles and reviews. We downloaded the search results as “Full Record and Cited References” and “Plain Text.” Following this, the files were renamed as “download_*.txt” to be analyzed by the CiteSpace software.

We used three bibliometric tools, CiteSpace 6.1.R2 Basic (22), R-bibliometrix 4.0.0 (23), and VOSviewer 1.6.18 (24), to conduct bibliometric evaluation and visualization and Microsoft Excel 2021 for statistics and the plotting part of the figure. The first step was to clean our data. For example, “splicing factors” and “splicing factor” were merged as “splicing factor” and “rna splicing,” “pre-mrna splicing,” and “alternative splicing” were unified as “alternative splicing” (25).

CiteSpace is capable of discovering collaboration, keywords, domain research structure, future direction, and evolution as a bibliometric and visual analysis tool in a scientific field (26). Using CiteSpace, we utilize its co-occurrence, timeline, bursts, and dual-map functions to draw a series of figures about countries/regions and institutions, journals, references, citations, and keywords. We use the analysis process and parameter settings recommended by the software developers, where the time span is 2012–2021, the time slice is 1 year, the selection criterion is the g index (k = 25), pruning is none, and the minimum duration of burstness is 2 years. The size of the node in the CiteSpace visualization portrays the number of co-occurrence. Furthermore, linkages show the relationships between the co-occurrences (22). As time passes from 2012 to 2021, the node and line’s colors change from purple to red to symbolize the different years. High betweenness centrality (≥0.10) nodes with purple circles serve as a hub between distinct networks (26–28).

Another bibliometric tool that excels in producing and visualizing knowledge maps is VOSviewer, which displays the kinds of clusters, overlays, or density colors (24). We mainly applied the co-occurrence analysis function of VOSviewer, including authors, journals, references, and their co-citations, as well as keywords. The specific parameters applied are mentioned in the corresponding chapters. In addition, we use the full counting method as a counting method. The meaning of the size of the node is the same as that of CiteSpace. Nodes with the same color belong to the same cluster. Additionally, links show the relationship between co-occurrences, and their degree of thickness is determined by the estimated strength value. The value is related to the number of papers published by the two authors or the frequency of the co-occurrence of the two keywords (24). The co-cited frequency is positively correlated with the word and round sizes as well as the yellow opacity in density maps. The color on the overlay map represents the typical publishing year.

R-bibliometrix is an R package for executing a comprehensive science mapping analysis of scientific literature (23). To be adaptable and make integration with other statistical and graphical R packages easier, R-bibliometrix was programmed in R. Maps of the geographical distribution of the countries/regions were produced using it. The network map displays the current state of research and communication between various countries/regions (23).

Excel 2021 software was used to anatomize the yearly publications and cited frequency. Additionally, we obtained impact factor (IF), journal citation report (JCR), average per item (ACI), journal classification, and author H-index from Web of Science on 1 August 2022.

From the WoSCC database, we acquired 3,544 papers, and we ultimately included 3,507 publications that qualified ( Figure 1 ; Supplementary Table 1 ). The number of publications related to AS of cancer and the frequency of citations have both steadily increased over the past 10 years, as illustrated in Figure 2 .

There were 3,507 papers in total, representing 80 different countries/regions and 405 institutions. The United States, China, and Germany are the top 3 countries in terms of the number of published articles, with 1,165, 1,013, and 249, respectively ( Table 1 ). China’s centrality, however, was less than 0.10, suggesting that it may not be a “hub” node in the AS of cancer studies (20). In contrast, the United States (centrality = 0.41), England (centrality = 0.25), and Germany (centrality = 0.20) had high centrality, which is depicted in Figure 3A by a purple circle. As shown in Figure 3A , the closeness of the co-occurrence atlas of countries/regions was 0.1864, denoting lively collaboration between them (27). Figure 3C shows the relative proportion of annual publications for the top 10 countries from 2012 to 2021. It was evident that China’s share had gradually increased. A country/region co-authorship network was created by R-bibliometrix ( Figure 3D ). The network map showed the current state of research and communication activities between these countries/regions. The Chinese Academy of Sciences is the scientific research institution with the largest number of published papers, as shown in Figure 3B ; however, its centrality is just 0.07 (n = 76). By contrast, the National Cancer Institute (n = 65, centrality = 0.14), the University of California San Diego (n = 48, centrality = 0.2), and Karolinska Institute (n = 25, centrality = 0.14) had a high centrality. The publication counts, H-index, and ACI of the top 10 most productive institutions are displayed in the bar graph of Figure 3E .

There were 20,406 authors active in AS of cancer research, and 25 of them published 10 or more articles, as shown in Figure 4A and Supplementary Table 2 . Scorilas Andreas, a National and Kapodistrian University of Athens scholar, was the most prolific author (n = 25), followed by Adamopoulos Panagiotis G. and Karni Rotem ( Table 2 ). Different colors represent different clusters in Figure 4 , a total of 15 (29). Active partnerships, such as Ladomery Michael and Oltean Sebastian, are frequently found in the same cluster. Additionally, there were alliances between two linked authors of different colors, which seemed to be Ghigna Claudia and Fu Xiangdong.

Authors who have been cited in one article are known as co-cited authors (30). From Figure 4B and Supplementary Table 2 , we can see that 83 authors out of the 85,667 co-cited authors had more than 100 co-citations. They are displayed as a density map in Figure 4B , which made it easy to identify the writers who were co-cited frequently. The hue gets warmer with more citations (31). Wang Eric T., Venables Julian P., and David Charles J. had the most co-citations, as indicated in Table 2 and Figure 4B . The images cannot display all the information due to CiteSpace and VOSviewer visualization’s intrinsic restrictions. As a result, we have supplemented this with detailed information on the graphs in the Supplementary Material .

Articles about AS of cancer research have been published in 766 scholarly journals overall. Nine hundred and five papers, or 25.81% of the total publications, were published in the top 15 journals ( Table 3 ). The number of papers published in Plos One is the largest (n = 145, 4.1%), followed by the International Journal of Molecular Sciences (n = 84, 2.4%) and Oncotarget (n = 77, 2.2%). A co-citation map of journals was produced by VOSviewer, as shown in Figure 5A . The required minimum was established at 100 citations, and 292 journals satisfied the requirement.

Of the 6,707 cited journals, 84 have more than 500 citations. Among them, the Journal of Biological Chemistry (n = 7,922), Nature (n = 7,333), and Proceedings of the National Academy of Sciences of the United States of America (PNAS) (n = 6,656) ranked in the top 3 in the number of citations ( Table 3 ). Additionally, 31.16% of all cited sources came from the top 15 co-cited journals.

The dual-map overlay of journals represents the field distribution of citing and cited journals, as seen in Figure 5B (32). Citation relationships are indicated by colored paths, with citing journals on the left and cited journals on the right (22). The principal citation channel indicates that papers published in Molecular/Biological/Immunology journals mainly cite papers published in Molecular/Biological/Genetics journals ( Figure 5B ).

Thirty-seven references out of the 132,804 cited ones were quoted at least 100 times ( Supplementary Table 3 ). The top 10 co-cited references are included in Table 4 , with a minimum of 157 co-citations. The article by Wang Eric T. et al. from Nature in 2008 (n = 465) is the one that has received the most co-citations out of all of them. In addition, four of the top 10 were reviews, and six of the top 10 were research articles.

The entire network map might be partitioned into several clusters using the clustering function, and studies inside a cluster might have distinct study themes to studies from other clusters ( Figure 6A ). Each cluster’s most prevalent terms were designated as cluster labels (38). The references’ timeline view could allow users to see how various research hotspots have changed over time. As shown in Figure 6B , cluster #0 (post-transcriptional regulation), #3 (prognostic alternative), #4 (alternative splicing), #6 (circular RNA), #7 (androgen receptor), and #8 (non-coding RNA malat1) started earlier, while cluster #1 (splicing signature), #2 (aberrant splicing), and #5 (oncogene srsf3) are still ongoing, which could be considered as the frontier.

with citation bursts indicate that their citations have grown by leaps and bounds in a particular period (27). Two hundred and fifty references were included in the citation bursts analysis, and the top 25 are listed in Figure 6C . One paper, entitled “Alternative pre-mRNA splicing regulation in cancer: pathways and programs unhinged,” had the strongest burstness (strength = 31.24) with citation burstness from 2012 to 2015 (10), which was published in Gene Dev by David Charles J. et al. in 2010. It is worth noting that five references (15, 37, 39–41) were still in burstness. Climente-González Héctor et al. (37) explored the functional influence of AS in cancer, Kahles André et al. (15) conducted a comprehensive analysis of AS across tumors, Ryan Michael et al. (39) provided a web-based resource for exploring the AS patterns of TCGA tumors, Li Yuan et al. (40) discovered a series of AS signatures in non-small cell lung cancer, and Seiler Michael et al. (41) analyzed the functional consequences of somatic mutation of splicing factor genes, respectively.

In total, 6,116 keywords were recovered, 121 of which appeared at least 10 times, and there were 31 keywords that appeared at least 30 times. As represented in Table 5 , alternative splicing (n = 1,238), cancer (n = 215), and prognosis (n = 125) were the three most popular keywords. We divided the keywords into three categories, namely, molecules, pathological processes, and diseases associated with AS of cancer, and listed the top 15 keywords in Table 6 , respectively. Obviously, P53 (n = 39), CD44 (n = 35), androgen receptor (n = 31), srsf3 (n = 24), and esrp1 (n = 16) were several molecular keywords with the highest frequency; apoptosis (n = 94), EMT (n = 61), metastasis (n = 55), angiogenesis (n = 40), proliferation (n = 34), and epigenetics (n = 23) were several pathological process keywords with the highest frequency; and breast cancer (n = 115), colorectal cancer (n = 82), prostate cancer (n = 76), hepatocellular carcinoma (n = 76), and gastric cancer (n = 41) were several disease keywords with the highest frequency in AS of cancer studies.

Keywords with high co-occurrence counts (n ≥ 10) are displayed as an overlay map in Figure 7A , with the hue denoting the typical year of publication. As we can see, the emerging fields that were given the color yellow include splicing factor, prognosis, immunotherapy, and TCGA (The Cancer Genome Atlas). Each cluster displayed the top 3 keywords over time in the timeline view ( Figure 7B ). Six of the seven clusters were still active except for cluster #6. Among them, #0 (splicing event) is the biggest cluster, followed by #1 (therapeutic target), #2 (epithelial–mesenchymal transition), and #3 (splicing factor). Supplementary Table 2 provides further details.

Bursts of keywords are those that were used a lot during a certain time frame (27). Figure 7C demonstrates that in vivo had the strongest bursts, with a strength of 15.42, followed by a splicing factor at 12.08 and visualization at 10.82. Notably, until 2021, the landscape, prognosis, splicing factor, and other keywords were in burstness. The results of VOSviewer and CiteSpace had a considerable similarity regarding keywords, indicating the reliability of our analysis.

Based on the information from the WoSCC database, a total of 3,507 AS of cancer publications by 20,406 writers in 405 institutions from 80 countries/regions were published between 2012 and 2021 in 766 academic journals. Increasing publications indicate that curiosity and attention are growing regarding the AS of cancer. The AS research officially started in 1977, when Roberts R.J. and Sharp P.A. found the phenomenon of “alternative splicing” (42, 43). Since that time, AS research has expanded quickly. Our focus is on the relationship between alternative splicing and cancer. AS research of cancer has steadily increased in steps during the last 10 years, and the relevant published article and cited frequency in 2021 are almost twice and over 80 times those of 2012, respectively.

The volume of published articles is a very important indicator, but the centrality is a crucial indicator for the quality of published articles in country/region analysis, where high centrality nodes (≥0.10) indicate the “hub” influence of particular countries/regions in the worldwide collaboration map (26, 28, 38, 44). The United States and China made the largest contributions to papers on AS of cancer, as shown in Table 1 and Figure 3A . The top 10 institutions by the number of publications were mainly from China and the United States: six were from China and four were from the United States, respectively. China and Chinese institutions, on the other hand, had a centrality of less than 0.1, while the United States had a centrality of 0.41, suggesting that the United States may continue to dominate AS of the cancer field. Furthermore, the United States, England, Germany, Australia, and France had high betweenness centrality, which indicates that they were important hubs in AS of the cancer field’s international collaboration. Moreover, it could be seen that countries/regions and institutions had an active collaboration in terms of network density, respectively.

Scorilas Andreas not only published the most articles related to AS of cancer but is also the top 1 H-index scholar ( Table 2 ), demonstrating his excellent contribution to AS of cancer research. Scorilas is a researcher at the National and Kapodistrian University of Athens, committed to AS, gene transcription, and next-generation sequencing. For the past 10 years, his group has published a series of articles (45–50) that described the process of identification of novel alternative splicing variants using next-generation sequencing methodology and discussed their expression situation and pathophysiological implications. Furthermore, Oltean Sebastian is also a scholar working at the University of Exeter; his team centered on the regulation and form of AS in prostate cancer and therapeutic targets associated with AS in cancer. In 2014, Nature published an Oltean and Bates’s review entitled “Hallmarks of alternative splicing in cancer” (11). They summarized how the numerous phenotypic traits that tumors acquire are influenced by AS and identified a new class of anticancer treatments called alternative splicing inhibitors; This article was co-cited up to 234 times with citation bursts from 2015 to 2019. Significantly, the top 1 co-cited author, Wang Eric T., a professor at the University of Florida, focused on exploring the function of the short and long gamma subunit splice variants of human GABA(A) receptors. He published the top 1 co-cited reference that reviewed the regulation of AS (5), with the second strongest citation burst strength.

According to a journal analysis ( Table 3 ), Plos One was the ninth most referenced journal and published the most AS of cancer studies. The fact that Nucleic Acids Research was among the top 5 academic journals, as well as co-cited journals, shows how important it is to the dissemination of AS of cancer research. The top 10 most-cited papers were published in the journals, which are also generally the most-cited journals. For instance, Nature obtained the second-highest number of co-citations, in part due to three of the top 10 extremely co-cited references (1, 5, 35) ( Table 4 ). Similar to the dual-map study ( Figure 5 ), we found that the majority of the journals were on the subjects of molecular, genetics, and comprehensive fields.

The knowledge network may be represented in part by the co-cited references cited by the papers in the corresponding field (22, 27, 51). The study’s top 10 co-cited articles found that three articles primarily explored the relationship between AS and hallmarks of cancer (11, 36, 37), two were about the discovery of alternative splicing isoforms using high-throughput sequencing (5, 33), two were related to the regulation of AS (6, 10), and two papers focused on several specific molecules, splicing factor SF2/ASF, and heterogeneous nuclear ribonucleoproteins (hnRNPs) controlled by c-Myc, respectively (34, 35). Moreover, one review discussed the mechanisms of proteome expansion by AS (1). As shown in our citation bursts results ( Figure 6C ), five references in the AS of cancer research are still active: three are related to the function and mechanisms of AS (37), one is a dedicated effort to mine prognostic AS signatures (40), and three present the landscape of AS in TCGA tumors (15, 39, 41).

Overall, as far as we know, this article may be the first to apply bibliometric methods to comprehensively examine papers associated with AS of cancer research published in the last 10 years. The bibliometric method offers a fresh and unbiased perspective on the changing research hotspots and fashion in contrast to conventional reviews (21). Simultaneously, we conducted an investigation using various bibliometric tools, which could produce more prosperous outcomes across numerous dimensions (21, 125). This study will educate the public about the significance of AS of cancer, offer scholars a complete image of AS of cancer study, and additionally provide a detailed and impartial direction for the field’s upcoming growth. This study unavoidably has certain shortcomings. First, we only obtained the WoSCC database’s English-language articles, leaving out non-English or non-WoSCC items. However, WoSCC’s English articles are the most often utilized data source in bibliometrics; thus, to some extent, they may represent the majority of the field (21, 126). Secondly, some studies report that bibliometric methods are inevitably biased because they are based on natural language processing (17, 20). Our findings, however, are comparable to recent conventional reviews while offering more comprehensive and unbiased data (4, 11, 86).